Trailing-Edge
-
PDP-10 Archives
-
decus_20tap5_198111
-
decus/20-0137/stp/stp4.doc
There are 3 other files named stp4.doc in the archive. Click here to see a list.
western michigan university
computer center
library program #1.1.4
calling name: stp
prepared by: richard houchard*
programmed by: richard houchard
approved by: jack r. meagher
date: september 1, 1974
statpack
statistical package
* with much assistance from nancy attwell,
berenice houchard, george r. kohrman, charles
lane jr., james e. nadonly, and delores vlahon
PAGE 2
statpack
purpose
the statpack was written to enable a large section of the western
michigan university community to do much of their own statistical
analysis on a terminal with only a simple working knowledge of the
decsystem-10.
limitations
maximum core allowable
nv - number of variables
no - number of observations no*nv+nv*nv+2*max+3*nv<8001
max - larger of nv or no
see also table of variable-observation combinations.
description
statpack is an integrated, interactive package written for
terminal use. it allows the user to issue simple commands for data
analysis and will prompt him for necessary information. when
questions of a procedural nature arise, the user may ask for an
additional explanation by simply typing "help". the standard output
device is the terminal; however, a command is available to channel
output to the line printer, providing the user with the ability to
obtain multiple copies.
data input may be from terminal, disk, magnetic tape, or a
structured data bank. input consists of observations each containing
a value for every variable. variables are defined by a number or an
alphabetic name of not more than five characters. data must be
entered before issuing any of the statistical commands. once data has
been entered, the statistical commands will continue analyzing it
until the data is modified or replaced. options exist for evaluating
data with missing values. it is also possible to restrict the data to
only those observations where a certain set of circumstances occurs.
statpack v4 PAGE 3
table of variable-observation combinations
------------------------------------------
the following is a table illustrating the various
variable-observation data combinations that can be processed by
statpack. letting the rows represent the number of observations and
the columns represent the number of variables, one can easily
determine if a specific variable-observation data combination is
possible by simply determining the point where the variable line
crosses the observation line. a "yes" indicates that the combination
is possible; a blank indicates that statpack cannot analyze the amount
of data necessary for that variable-observation combination.
number of variables
1 2 3 4 5 10 15 20 25 30 40 50 75 100
10 yes yes yes yes yes yes yes yes yes yes yes yes yes
20 yes yes yes yes yes yes yes yes yes yes yes yes yes
30 yes yes yes yes yes yes yes yes yes yes yes yes
40 yes yes yes yes yes yes yes yes yes yes yes yes
50 yes yes yes yes yes yes yes yes yes yes yes yes
60 yes yes yes yes yes yes yes yes yes yes yes yes
n 70 yes yes yes yes yes yes yes yes yes yes yes yes
u 80 yes yes yes yes yes yes yes yes yes yes yes yes
m 90 yes yes yes yes yes yes yes yes yes yes yes yes
b 100 yes yes yes yes yes yes yes yes yes yes yes yes
e 125 yes yes yes yes yes yes yes yes yes yes yes
r 150 yes yes yes yes yes yes yes yes yes yes
200 yes yes yes yes yes yes yes yes yes yes
o 250 yes yes yes yes yes yes yes yes yes
f 300 yes yes yes yes yes yes yes yes
350 yes yes yes yes yes yes yes
o 400 yes yes yes yes yes yes yes
b 450 yes yes yes yes yes yes yes
s 500 yes yes yes yes yes yes
e 600 yes yes yes yes yes yes
r 700 yes yes yes yes yes
v 800 yes yes yes yes yes
a 900 yes yes yes yes yes
t 1000 yes yes yes yes yes
i 1100 yes yes yes yes yes
o 1200 yes yes yes yes
n 1300 yes yes yes yes
s 1400 yes yes yes
1500 yes yes yes
1600 yes yes
1700 yes yes
1800 yes yes
1900 yes yes
2000 yes
statpack v4 PAGE 4
list of commands
----------------
"data" - data input by terminal
"fetch" - read data from disk
"form" - enter special input format
"manip" - manipulate data in core (includes appending)
"trans" - data transformations
"store" - store data on disk
"print" - print selected variables on line printer
"type" - type selected variables on terminal
"acbnk" - access a stored data bank
"mabnk" - create a bank from data in stp
"sort" - sort data into ascending order
"mta/i" - read data from magtape
"desc" - description of data - means, st. dev., var.
"basic" - medians, modes, and ranges
"erana" - std error of mean, coeff of skewness, coeff of var
"estat" - "desc","basic", and "erana"
"zscor" - z scores
"kolm" - 1 or 2 sample kolmogorov-smirnov tests
"corr" - correlation matrix
"pcorr" - partial correlations
"kendl" - kendall tau correlations
"srank" - spearman rank correlation
"ptbis" - point biserial correlation
"ttest" - t test (significance between means)
"corrt" - correlated t tests
"mann" - mann-whitney u test
"wilcx" - wilcoxon rank
"anov1" - single factor analysis of variance
"anov2" - 2-way analysis of variance
"1wayr" - 1-way analysis of variance w/ repeated measures
"anoc1" - 1-way analysis of covariance
"regr" - regression
"stepr" - stepwise regression
"facto" - factor analysis
"prob" - probability assoc. with t, f, or chi square
"chisq" - chi square
"cvsmt" - exponential curve smoothing model
"plot" - scatter plot
"hist" - histogram
"bargr" - bar graph
"freq" - frequency
"xtab" - cross tab
statpack v4 PAGE 5
"xtab*" - cross tab (table form - only if "assig" is used)
"pcent" - percentiles
"stop" - restart
"help" - for commands
"fini" - end run
"info" - general information
"assig" - assign output to line printer
"deass" - reinitialize output to terminal
"copys" - indicate more than 1 printer copy ("assig" and "print")
"title" - label output with identification
"name" - give names to variables
"make" - make a text to be inserted into lineprinter output
statpack v4 PAGE 6
program transfer
----------------
purpose: initiate the run of another program while in statpack
description: stat pack may be used to transfer control to another
program (initiate a run to a different program). as the
following programs become available, they may be called directly
from stat pack.
bank
freq
tab
corl
regr
when a call to another program is executed, the output file (if
one has been created) is queued to the lineprinter, and the
program specified is executed. to run a program type a "/" and
the program name in response to "which command?".
example:
which command? /bank
bank?
statpack v4 PAGE 7
command: acbnk
---------------
purpose: read data from a binary structured data bank. options are
available to subset data, bypass observations which do not meet
user specified criteria, and reject observations containing
missing values. provisions have been made for variable names,
and elimination of input formats.
limitations: data must be in the form of a well-structured data bank.
a maximum of 20 variables may be accessed at one time.
description: the "acbnk" command is used to read data from a stored
structured data bank located on the disk. when prompted, the
user types the name of the bank (no extension is necessary;
".bnk" will automatically be added), and the project-programmer
number enclosed in brackets (if other than his own), of the area
where the file containing the bank is located.
this is followed by the switches enclosed in parentheses,
which are representative of the options available. if no options
are desired, no options are necessary. options are:
"i" - independent samples
"m" - allow observations with missing data to be recovered
"q" - select observations to be used by specifying criteria
they must satisfy
"s" - specify starting position
next the user is asked to indicate which variables are to be
read. bank codes (the number of the variable as situated in the
bank) or variable names are typed in separated by commas. ranges
of variables may be entered by typing the extremes of the range
separated by a "-". if all variables are to be read use a "*".
it should be noted that the first bank code specified becomes
statpack's first variable, the second bank code the second
variable, etc. thus, bank code 1 may end up as variable number
4. although the variable name obs is illegal, the user may
indicate this variable both as a variable to be read and as a
qualifier. the observation number will be referenced. in stp if
obs is read as a variable it will be changed to obser.
if no options are specified, data will be considered an
observation at a time, starting with the first observation. if
any of the variables to be recovered contain missing data for
that observation, the entire observation is discarded. stp will
continue in this manner, always checking the next observation,
until either the entire bank has been considered or the data set
whose size was specified at the beginning has been filled.
statpack v4 PAGE 8
if an "s" switch is specified, the user will be asked to
supply the starting observation number for the bank. any value
between 1 and the number of the last observation in the bank may
be used. this does not, however, inhibit the "acbnk" command
from checking the entire bank for data.
an "m" switch allows the collecting of samples containing
missing data. the number "-9999e-20" will be used to denote
values which are missing.
the "q" switch specifies the qualifying option. by using
the "q" switch, a user may select a subset of the bank where all
the observations conform to a certain user-specified criteria.
for example, data may be chosen where each observation in the
data set is a male, less than 15 years old, with an i.q. of over
110.
if the "q" switch has been specified, the user is instructed
to type one qualifier after each "?". each qualifier consists of
three parts: a variable, a relationship, and a value to be
compared with the variable. hence, before each observation is
included in the data set it must satisfy all the qualifiers
specified. (for each observation to be accepted, the variable
selected in each qualifier must have the indicated relationship
to the value specified.) the relationships possible are:
sign code alpha code relationship
--------- ---------- ------------
"=" ",eq," or ".eq." equal
"<>" or "><" ",ne," or ".ne." not equal
">" ",gt," or ".gt." greater than
"<" ",lt," or ".lt." less than
">=" or "=>" ",ge," or ".ge." greater than or equal to
"<=" or "=<" ",le," or ".le." less than or equal to
the qualifier is constructed by typing the variable (either
number or name), the sign code or alpha code, and finally the
value. when the last qualifier has been entered, return and type
a ^z(control z), a <cr>, or "stop"(the stop typed here simply
terminates entry of the qualifiers, it is not the same as the
"stop" command). the maximum number of qualifiers possible is 20
minus the number of variables selected for entry to the data set.
it is not necessary to access a variable as data in order to use
it as a qualifier.
the "i" switch is used to assemble samples where variables
in the same observation may not be related. for example, it is
possible for variable 1 to be the i.q. of males and variable 2
the i.q. of females. each variable or set of variables is
collected by looking through the entire data set once. when the
last sample has been recovered, type a <cr> or a ^z(control z).
all variables are forced to have the same number of observations.
statpack v4 PAGE 9
examples (command explanations follow these examples):
(1) which command? acbnk
what bank name and switches? nick[220,220]
list bank codes separated by commas
case,2,3,weight,sex
(2) which command? acbnk
what bank name and switches? nick[220,220](s)
what is the starting position? 123
list bank codes separated by commas
case,2,3,weight,sex
(3) which command? acbnk
what bank name and switches? nick(q)
list bank codes separated by commas
case,sex,weight,7-11
a ? indicates a qualifier should be inserted. after
last one type a ^z(control z), a <cr>, or stop
? sex=1
? 2<>499.3
? 3,gt,1.2
? weight,ne,189
? ^z
(4) which command? acbnk
what bank name and switches? rslt(qsm)
what is the starting position? 121
list bank codes separated by commas
iq,sex,weight
a ? indicates a qualifier should be inserted. after
last one type a ^z(control z), a <cr>, or stop
? iq>115
? sex=1
? weight<=210
? ^z
statpack v4 PAGE 10
(5) which command? acbnk
what bank name and switches? nick(qi)
independent samples will be taken on successive
"list bank codes". when all independent samples have been
given, type ^z(control z), a <cr>, or stop to the
question
independent sample 1
list bank codes separated by commas
weight
a ? indicates a qualifier should be inserted. after
last one type a ^z(control z), a <cr>, or stop
? sex=1
? age<24
? ^z
independent sample 2
list bank codes separated by commas
weight
a ? indicates a qualifier should be inserted. after
last one type a ^z(control z), a <cr>, or stop
? sex=1
? age>=24
? ^z
independent sample 3
list bank codes separated by commas
weight
a ? indicates a qualifier should be inserted. after
last one type a ^z(control z), a <cr>, or stop
? sex=2
? age<24
? ^z
independent sample 4
list bank codes separated by commas
weight
a ? indicates a qualifier should be inserted. after
last one type a ^z(control z), a <cr>, or stop
? sex=2
? age=>24
? ^z
independent sample 5
list bank codes separated by commas
^z
statpack v4 PAGE 11
explanations:
(1) read from data bank nick under area [220,220] the variables:
case, 2, 3, weight, and sex.
(2) read from data bank nick under area [220,220] the variables:
case, 2, 3, weight, and sex beginning with observation 123 in the
bank.
(3) read from data bank nick the variables: case, sex, weight,
and variable numbers 7 through 11; but only use those
observations where the variable: sex is equal to 1, variable: 2
is not equal to 499.3, variable: 3 is greater than 1.2, and
weight is not equal to 189.
(4) read from data bank rslt the variables: iq, sex, and weight.
starting at the bank's observation number 121 and using only
those observations where the variable: iq is greater than 115,
the variable: sex is equal to 1 and the variable: weight is less
than or equal to 210.
(5) read independent samples from bank nick, breaking the
variable: weight into four variables each recovered as a separate
sample where:
each observation for sample 1 has the variable: sex equal to
1 and variable: age less than 24;
each observation for sample 2 has the variable: sex equal to
1 and variable: age greater than or equal to 24;
each observation for sample 3 has the variable: sex equal to
2 and variable: age less than 24; and
each observation for sample 4 has the variable: sex equal to
2 and variable: age greater than or equal to 24.
when finished there will be four separate variables each made up
of a particular portion of the population.
statpack v4 PAGE 12
command: anoc1
---------------
purpose: perform one way analysis of covariance
limitations: the sum of the treatments and covariances must not
exceed 20.
reference: "statistical methods", snedecor and cochran, chapter 14.
description: the "anoc1" command calculates analysis of covariance
for one or more sets of data. the user will first be instructed
to enter the options desired, separated by commas. available
options are:
"break"--rather than supplying a variable for each cell of the
analysis of covariance, separate a single variable and one
or more covariate variables into treatments based on values
of a breakdown variable.
"discr"--use values of the breakdown variable to form ranges,
with each distinct value forming a separate range. (if this
option is not used the user will be asked to enter ranges
for the breakdown variable).
"auto"--same results as "discr", however "auto" is not entered
when the other options are entered, it is entered when asked
for ranges.
"range"--list ranges calculated. (only available if "discr" is
used).
if no options are specified, the user will be asked to enter
the variables to be used as treatments. each variable entered
becomes a separate treatment. variables may be entered by
variable number or if names have been defined, as variable names.
ranges of variables may be entered by typing the extremes of this
range separated by a "-". next the user will be instructed to
enter the number of covariates. for each treatment, the user
will be asked to enter a variable to be used for each covariate.
either variable numbers or variable names (if names have been
defined) may be used. ranges of variables may be entered by
typing the extremes of the range separated by a "-". no variable
may be used as both a treatment and a covariate, and no variable
may be used as two covariates.
if "break" was specified as an option, the user will be
instructed to enter the variables to be analysed. up to 20
statpack v4 PAGE 13
variables may be entered separated by commas. variables are
indicated by variable number or variable name (if names have been
defined for the variables). ranges of variables may be entered
by typing the extremes of the range separated by a "-". for each
variable specified, a separate analysis of covariance will be
calculated. the user will then be asked to enter the breakdown
variable. only one variable may be entered, by either variable
number or variable name (if names have been defined).
if option "discr" was not supplied then the user will be
requested to enter the ranges for the breakdown variable. each
range is entered on a separate line, minimum first, followed by a
comma, and then the maximum. if names are to be given to the
ranges, the user should enter a comma and the name (5 characters
or less) following the maximum of the range. after the last
range has been entered, type a control z (^z) or line feed. if
the user wishes the ranges to be automatically created "auto" may
be used.
the user will now be instructed to enter covariates
separated by commas. either variable numbers or variable names
(if names have been defined) may be used. ranges of variables
may be entered by typing the extremes of the range separated by a
"-". the covariates will be broken into the same groups as the
treatments.
examples:
which command? anoc1
list options separated by commas
break,discr
which variables are to be analysed? a
which is the breakdown variable? 1
list covariates separated by commas
b,c,d
statpack v4 PAGE 14
***** 1-way anocov *****
analysis on variable: a with treatments determined
by a breakdown on variable: 1 ; covariates used:
b , c , d
unadjusted adjusted covariate means
treat size mean mean b c d
1 4 62.4 62.9 14.7 1.38 -50.3
2 1 19.5 13.2 13.3 -35.9 123.
3 1 73.2 72.7 17.0 12.6 7.53
4 5 36.0 37.8 11.4 21.0 -11.5
5 9 34.3 35.1 10.8 -0.610 -36.2
6 12 39.1 39.5 8.02 2.88 17.9
7 21 43.5 43.2 11.2 -0.605 4.24
8 8 49.5 50.0 10.6 -4.49 -38.0
9 8 38.5 36.4 13.1 -7.31 43.1
10 4 31.8 32.4 6.50 0.949 16.4
*totals 0.304e+04 790. 13.2 -42.6
*average 41.7 10.8 0.180 -0.583
*beta weights 0.217 -0.798e-01 0.234e-01
1 way anocov
source sum of squares df mean squares f prob
between
adjusted 5033.185 9 559.2 0.741 .670
treatments
error 45262.61 60 754.4
total 50295.79 69
statpack v4 PAGE 15
which command? anoc1
list options separated by commas
list variables to be used as treatments
d,f
how many covariates? 2
list covariates for var: d
a,b
list covariates for var: f
c,e
***** 1-way anocov *****
treatments and covariates are individual variables
treat cov 1 cov 2
d a b
f c e
unadjusted adjusted covariate means
treat size mean mean cov 1 cov 2
d 73 -0.583 -4.43 41.7 10.8
f 73 6.86 10.7 0.180 24.8
*totals 458. 0.306e+04 0.260e+04
*average 3.14 20.9 17.8
*beta weights 0.799e-01 -0.314
1 way anocov
source sum of squares df mean squares f prob
between
adjusted 3998.871 1 3999. 0.394 .531
treatments
error 1440020. 142 0.1014e+05
total 1444019. 143
statpack v4 PAGE 16
command: anov1
---------------
purpose: calculate one way analysis of variance
reference: "statistical principles in experimental design", winer,
pages 96-102.
description: the "anov1" command allows the user to calculate one-way
analysis of variance. the user will first be instructed to list
the options desired separated by commas (if no options are
desired type a <cr>). possible options are:
"break"--select samples from one variable based on the value of a
second variable. for each observation, the value of the
second variable (breakdown variable) will be used to
determine in which sample the variable being analyzed
belongs. this is accomplished by determining which of a
series of ranges the value of the breakdown variable fits
into, and then moving the value of the analysis variable to
the corresponding sample. (if this option is not used, the
analysis will be done using variables as the samples.)
note: the following options are to be used only if "break" has
been used.
"discr"--automatic breakdown. instead of the user entering
ranges, a separate range will be created automatically for
each value in the breakdown variable.
"auto"--automatic breakdown. this option is the same as the
"discr" option. do not enter "auto" with the other options;
it should be entered only when asked to type in the ranges.
the "discr" and "auto" options are equivalent. the only
difference is at which point in the program they are
entered.
note: the following option is available only if automatic
breakdowns are to be used.
"range"--list the ranges to be used for the automatic breakdown.
if the "break" option has not been specified, the analysis
of variance will be calculated with each variable specified by
the user occupying a single cell. when instructed the user lists
the variables to be used in the analysis of variance separated by
commas. up to 30 variables may be indicated by variable names
(if names have been defined) or variable numbers. ranges of
variables may be entered by listing the extremes of the range
statpack v4 PAGE 17
separated by a "-". one or more "*" may be used when listing the
variables to be analyzed. one at a time each variable not yet
specified in the analysis will be substituted for every "*".
those cases where the same variable would be listed twice in the
same analysis will be eliminated, as will be those cases that
except for a switch in the order of variables, duplicate an
analysis already performed.
if the "break" option has been used, it will be necessary
for the user to supply the following information:
(1) the variables for which analysis of variance are to be
calculated (up to 20). the samples for each set of analysis
of variance will be selected from a single variable.
variables may be listed using either variable names (if
names have been defined) or variable numbers. ranges of
variables may be specified by listing the extremes of the
ranges separated by a "-". where analysis of variances are
to be calculated for all variables, a "*" may be used
instead of variable names or numbers.
(2) the variable to be used for the breakdowns. only one
variable may be entered, specified by either its variable
name (if the name has been defined) or variable number. all
variables listed for analysis will be processed using the
same breakdown variable.
(3) ranges for the breakdown variable. if the "discr" option
has been used, this information will be automatically
calculated, and need not be supplied by the user. if the
"discr" option has not been used, the user may still request
the ranges to be automatically calculated by responding with
"auto". to specify ranges, the user types the extremes of
the range, smaller first, separated by a comma. only one
range may be entered per line. up to 50 ranges may be
specified. after the last range has been entered, the user
types a ^z(control z).
statpack v4 PAGE 18
examples:
which command? anov1
list options separated by commas
which variables? test1,test2,test3,test4
***** 1-way anova *****
tret. size mean std. dev.
test1 559 41.75 21.73741
test2 559 49.68 18.73237
test3 559 47.88 12.27697
test4 559 64.81 19.69569
source sum of sq. d.f. mean sq. f prob
between 160902.1 3 .5363e+05 157.5 0.0000
within 760030.1 2232 340.5
total 920932.1 2235
which command? anov1
list options separated by commas
break
which variables are to be analyzed? gpa
what is the breakdown variable? weight
enter ranges for breakdown variable: weigh
? 100,120,low
? 121,150,med
? 151,200,high
? ^z
***** 1-way anova *****
analysis on variable: gpa with treatments determined
by a breakdown on variable: weigh
tret. size mean std. dev.
low 25 2.918 0.5476086
med 29 3.002 0.5271557
high 34 2.958 0.5200980
source sum of sq. d.f. mean sq. f prob
between 0.9464264e-01 2 .4732e-01 0.1683 0.8454
within 23.90458 85 .2812
total 23.99922 87
statpack v4 PAGE 19
which command? anov1
list options separated by commas
break,discr,range
which variables are to be analyzed? gpa
what is the breakdown variable? age
breakdown ranges for variable: age
22.00 , 22.00
23.00 , 23.00
24.00 , 24.00
***** 1-way anova *****
analysis on variable: gpa with treatments determined
by a breakdown on variable: age
tret. size mean std. dev.
1 39 2.914 0.4997773
2 26 2.798 0.4729230
3 35 3.015 0.4958847
source sum of sq. d.f. mean sq. f prob
between 0.7007980 2 .3504 1.450 0.2397
within 23.44360 97 .2417
total 24.14440 99
statpack v4 PAGE 20
command: anov2
---------------
purpose: calculate two-way analysis of variance.
reference: "topics in intermediate statistical methods", snedecor and
cochran, pages 104-106.
description: the "anov2" command allows the user to calculate two-
way analysis of variance. the user will first be instructed to
list the options desired separated by commas (if no options are
desired type a <cr>). possible options are:
"headr"--eliminate means and standard deviation report.
"break"--select samples from one variable based on the value of
two other variables. for each observation, the value of the
two variables (breakdown variables) will be used to
determine in which cell the variable being analyzed belongs.
this is accomplished by determining which of a series of
ranges each breakdown variable fits into, and then moving
the value of the analysis variable to the corresponding
cell. (if this option is not used, variables will be used
as cells).
note: the following options are to be used only if the "break"
option is specified.
"discr"--establish ranges for the breakdown variables
automatically. rather than the user typing in ranges for
the breakdown variables, ranges will be automatically
calculated for each breakdown variable in the following
manner:
1) if a breakdown variable has fewer discrete values than
defined groups, a range will be calculated for each
value of the breakdown variable.
2) if a breakdown variable has more discrete value than
defined groups, a range will be calculated by finding
the difference between the maximum and minimum values
of the breakdown variable and separating the difference
into ranges; so that there is a range for each defined
group and all the ranges have equal intervals.
"auto"--if the "discr" option has not been used and the user
wishes to have ranges created automatically for a breakdown
variable, he may type "auto" when instructed to enter the
ranges for a particular breakdown variable. ranges will be
calculated in the same manner as for "discr". do not enter
"auto" with the other options; it should be entered only
when instructed to type in the ranges.
statpack v4 PAGE 21
"group"--specify number of groups (used when groups are to be
automatically broken down). preset to a maximum of 20.
note: the following is available only if automatic breakdowns
are used.
"range"--list the ranges to be used for automatic breakdown.
if the "break" option has not been specified, the analysis
of variance will be created with each specified variable
occupying a cell. the user will be asked for the following
information:
1) number of levels for factor 1. any number between 1 and 20
is acceptable.
2) number of levels for factor 2. any number between 1 and 20
is acceptable.
3) the variable to be put in each cell. one at a time the user
will be asked to specify which variable to be put into each
cell. either the variable name (if names have been defined)
or the variable number may be used to indicate the variable.
if a cell is empty the user may indicate this by typing
"empty". a "*" may also be used in one or more cells. one
at a time each variable not yet specified in the analysis
will be substituted for every "*".
if the break option has been specified, it will be necessary
for the user to supply the following information:
1) if the "group" option has been selected, the user will be
asked to indicate how many groups comprise a breakdown
variable. each breakdown variable can have 1 to 20 groups.
2) the variables for which the analyses are to be performed for
each two-way analysis of variance, the values comprising all
the cells will be selected from the same variable. up to 40
variables may be entered using either variable names (if
names have been defined) or variable numbers. ranges of
variables may be specified by listing the extremes of the
ranges separated by a "-". where analysis of variance are
to be calculated for all variables, a "*" may be used
instead of variable numbers or names.
3) the user will be asked to supply both breakdown variables.
either variable names (if names have been defined) or
variable numbers may be used.
4) if the "discr" option was not specified, the user will be
expected to enter the ranges for the breakdowns. up to 20
ranges may be submitted. each range is indicated by typing
the extremes (smaller first) separated by a comma. when
finished entering ranges for either of the breakdown
variables, type a ^z(control z). if ranges are to be
calculated automatically for a breakdown variable, type
"auto".
statpack v4 PAGE 22
examples:
which command? anov2
list options separated by commas
how many cells in factor 1? 2
how many cells in factor 2? 2
type in each variable after the
corresponding level, "empty"-indicates empty cell
cell( 1, 1)? test1
cell( 1, 2)? test2
cell( 2, 1)? test3
cell( 2, 2)? test4
*****2-way anova*****
factor
one factor two
level 1 2
1 n 100.00 100.00
mean 39.00 47.37
stdv 21.09 20.28
2 n 100.00 100.00
mean 45.96 62.21
stdv 12.00 19.20
preliminary anova
source df ss ms f prob
cells 3 28581.45 9527.15 27.84 .0000
1 ignoring 2 1 11878.98
2 ignoring 1 1 15150.97
within 396 135500.27 342.17
total 399 164081.73
final anova
source df ss ms f prob
cells 3 28581.45
1 eliminating 2 1 11878.98 11878.98 34.72 .0000
2 eliminating 1 1 15150.96 15150.96 44.28 .0000
1 by 2 1 1551.51 1551.51 4.53 .0338
within 396 135500.27 342.17
total 399
statpack v4 PAGE 23
which command? anov2
list options separated by commas
break
which variables are to be analyzed? gpa
which variable is breakdown variable 1? height
which variable is breakdown variable 2? weight
list the ranges for breakdown variable: heigh
50,62
63,64
65,70
^z
list the ranges for breakdown variable: weigh
100,118
119,137
138,200
^z
*****2-way anova*****
analysis run on variable: gpa with cells determined
by breakdowns on variable: heigh and variable: weigh
factor
one factor two
level 1 2 3
1 n 11.00 11.00 5.00
mean 2.95 2.86 3.11
stdv 0.47 0.58 0.29
2 n 8.00 9.00 13.00
mean 2.89 2.80 3.15
stdv 0.66 0.59 0.56
3 n 1.00 3.00 25.00
mean 2.90 3.18 2.90
stdv 0.00 0.45 0.52
preliminary anova
source df ss ms f prob
cells 8 1.18 0.15 0.51 .8490
1 ignoring 2 2 0.03
2 ignoring 1 2 0.25
within 77 22.47 0.29
total 85 23.64
statpack v4 PAGE 24
final anova
source df ss ms f prob
cells 8 1.18
1 eliminating 2 2 0.18 0.09 0.32 .7296
2 eliminating 1 2 0.40 0.20 0.68 .5085
1 by 2 4 0.75 0.19 0.64 .6359
within 77 22.47 0.29
total 85
which command? anov2
list options separated by commas
break,discr,range
which variables are to be analyzed? gpa
which variable is breakdown variable 1? sex
which variable is breakdown variable 2? age
breakdown ranges for variable: sex
.0000 , .0000
1.000 , 1.000
breakdown ranges for variable: age
21.00 , 21.00
22.00 , 22.00
23.00 , 23.00
*****2:way anova*****
analysis run on variable: gpa with cells determined
by breakdowns on variable: sex and variable: age
factor
one factor two
level 1 2 3
1 n 18.00 16.00 11.00
mean 2.90 2.80 2.74
stdv 0.47 0.45 0.46
2 n 17.00 23.00 15.00
mean 2.86 2.99 2.84
stdv 0.51 0.52 0.49
statpack v4 PAGE 25
preliminary anova
source df ss ms f prob
cells 5 0.67 0.13 0.55 .7370
1 ignoring 2 1 0.18
2 ignoring 1 2 0.21
within 94 22.70 0.24
total 99 23.37
final anova
source df ss ms f prob
cells 5 0.67
1 eliminating 2 1 0.18 0.18 0.76 .3861
2 eliminating 1 2 0.22 0.11 0.45 .6413
1 by 2 2 0.27 0.13 0.56 .5746
within 94 22.70 0.24
total 99
statpack v4 PAGE 26
command: assign
----------------
purpose: assign output to the line printer.
description: the "assign" command allows the user to print results on
the line printer. this command must be given prior to any
commands for which the output is to be assigned to the line
printer. once the command has been given, no other responses are
necessary. output will remain assigned to the line printer until
a "deass" command is given. it is permissible to channel output
back and forth between the line printer and terminal using the
"assign" and "deass" commands. questions to the user will still
be asked via the terminal. while output is assigned, no output
will appear on the terminal.
note: a "fini" command must be used to initiate printing of the output
assigned to the line printer. it is necessary to use the "fini"
command to ensure printing of the results which were assigned.
for multiple copies of the results use the "copys" command.
example:
which command? assign
output assigned to printer
which command?
statpack v4 PAGE 27
command: bargr
---------------
purpose: create bargraphs
reference: "basic statistical methods", downe and heath, page 27.
description: the "bargr" command allows user to construct one or more
bargraphs. when instructed, the user enters the variables
separated by commas for which the bargraphs are to be created.
up to 20 variables may be entered using either the variable names
(if names have been defined) or variable numbers. ranges of
variables may be indicated by typing the extremes of the range
separated by a "-". where bargraphs are to be constructed for
all variables, a "*" may be used in place of variable names or
numbers.
the user will next be instructed to enter the ranges for the
bargraphs. ranges are entered 1 per line, minimum first,
followed by a comma and the maximum. when the last range has
been entered, a control z (^z) or a blank line should be typed.
ranges may be calculated automatically, by placing each unique
value in an individual range, or if more than 40 values exist,
then 40 ranges are created each with an equal interval. to have
the ranges automatically calculated the user should respond with
"auto" when instructed to enter the ranges. if auto is used, and
the user wishes to limit the number of ranges to be created, the
"auto" should be followed by a "/" and the maximum number of
ranges the user desires.
examples:
which command? bargr
which variables? grade
enter ranges 1 per line
? 2,2
? 3,3
? 4,4
? 5,5
? 6,6
? 7,7
?
statpack v4 PAGE 28
***** bar graph for variable: grade *****
range of values freq pcent +----+----+----+----+
2.000 - 2.000 0 0.0 i
3.000 - 3.000 2 5.7 ixxx
4.000 - 4.000 7 20.0 ixxxxxxxxxx
5.000 - 5.000 10 28.6 ixxxxxxxxxxxxxx
6.000 - 6.000 11 31.4 ixxxxxxxxxxxxxxxx
7.000 - 7.000 5 14.3 ixxxxxxx
---- +----+----+----+----+
35 ^ ^ ^ ^
10.0 20.0 30.0 40.0
percentage
which command? bargr
which variables? 2,3
enter ranges 1 per line
? auto/4
***** bar graph for variable: grade *****
range of values freq pcent +----+----+----+----+
3.000 - 4.000 2 5.7 ixx
4.000 - 5.000 7 20.0 ixxxxxxx
5.000 - 6.000 10 28.6 ixxxxxxxxxx
6.000 - 7.000 16 45.7 ixxxxxxxxxxxxxxx
---- +----+----+----+----+
35 ^ ^ ^ ^
15.0 30.0 45.0 60.0
percentage
***** bar graph for variable: iq *****
range of values freq pcent +----+----+----+----+
42.00 - 75.00 8 22.9 ixxxxxxxxxxx
75.00 - 107.0 14 40.0 ixxxxxxxxxxxxxxxxxxxx
107.0 - 139.0 10 28.6 ixxxxxxxxxxxxxx
139.0 - 172.0 3 8.6 ixxxx
---- +----+----+----+----+
35 ^ ^ ^ ^
10.0 20.0 30.0 40.0
percentage
statpack v4 PAGE 29
command: basic
---------------
purpose: displays medians, modes, and ranges.
reference: "basic statistical methods", downe and heath, pages 32-34.
description: the "basic" command enables the user to display medians,
modes, and ranges for all variables. once the command has been
given no additional responses are necessary. output will be
appropriately labeled.
example:
which command? basic
var. median mode maximum minimum
sex 1.000000 1.000000 1.000000 0.0000000
age 23.00000 22.00000 27.00000 18.00000
heigh 64.00000 66.00000 73.00000 58.00000
weigh 139.0000 112.0000 251.0000 86.00000
iq 101.0000 113.0000 129.0000 70.00000
gpa 3.050000 2.910000 3.790000 1.850000
statpack v4 PAGE 30
command: chisq
---------------
purpose: calculate chi square
limitation: chisq will only calculate the chi square for raw data
already entered into statpack. there are no provisions for
calculating a chi square from a pre-calculated table.
reference: "non-parametric statistics", siegel, pages 42-47, 104-111,
196-202.
description: the "chisq" command allows the user to calculate either
a one-variable or a two-variables chi square. options exist to
group certain values together prior to forming the contingency
table, and collapsing the table once it is formed. when
instructed to list the options, the user enters the options
desired separated by commas (if no options are desired, type a
<cr>). possible options are:
group--group certain ranges of values together prior to forming
the contingency table
contg--eliminate contingency table from final output
colps--collapse contingency table (includes omitting)
fishr--calculate fishers exact probability (2x2 table only)
when instructed to enter variables, the user enters either
one or two variables separated by a comma. variables may be
listed by variable names (if names have been defined) or variable
numbers. if a one-variable chi square is desired for all
variables, the user may enter a "*". similarly, a two-variable
chi square between a single variable and all the remaining
variables may be entered by typing the variable which is to be
analyzed with all others, a comma, and a "*". if two variable
chi squares are to be calculated between all combinations of
variables, a "*,*" may be used.
if the "group" option has been given, the user will be
allowed to enter ranges for grouping certain values together. up
to 15 ranges may be entered for each variable, one range per
line. each range is entered by typing the extremes of the range,
smaller first separated by a comma. to finish entering ranges,
type a "stop" (this stop terminates the entry of ranges - it is
not the same as the "stop" command), <cr>, or ^z(control z).
if the "colps" option has been given, the user will be
expected to enter instructions for collapsing the contingency
table. each instruction is composed of a single character
indicating what is to be done, a variable name (if names have
statpack v4 PAGE 31
been defined) or variable number enclosed in parentheses, and a
string of numbers referencing levels to be acted upon.
single letter codes possible are:
c--combine levels
d---delete levels
the variable enclosed in parentheses must be one of the
variables being analyzed. the string of numbers represent the
level numbers in the contingency table which are to be acted
upon. ranges of levels may also be indicated by entering the
extremes of the range separated by a "-". collapsing
instructions are entered one per line. when finished, type
"stop" or ^z(control z).
the probability and contingency coefficient will be supplied
with the chi square. in the case of a 2 x 2 table the corrected
chi square will also be calculated. if in a two variable chi
square only one level exists for either of the variables, the chi
square will be calculated as a one-variable test.
examples:
which command? chisq
list options separated by commas
which variable or variables? sex,age
***** chi square *****
levels for horizontal variable: sex
level values comprising level
1 .0000
2 1.000
levels for vertical variable: age
level values comprising level
1 18.00
2 19.00
3 20.00
4 21.00
5 22.00
6 23.00
7 24.00
8 25.00
9 26.00
10 27.00
statpack v4 PAGE 32
sex
level 1 2 total
age ....................
1 . 27 18 45
2 . 28 21 49
3 . 23 35 58
4 . 36 28 64
5 . 30 36 66
6 . 22 30 52
7 . 26 38 64
8 . 30 34 64
9 . 25 29 54
10 . 34 25 59
total . 281 294 575
chi square = 11.94035 with 9 degrees of freedom
having a probability of 0.22
contingency coefficient = 0.47979e-01
which command? chisq
list options separated by commas
help
the chi square command operates from raw data not
pre calculated tables, options available are:
"group" - used to establish groupings prior to colps
"contg" - used to eliminate contingency table from final output
"colps" - collapse contingency table (includes omitting)
"fishr" - fisher's exact test probability (2x2 only)
if no options are desired type a return
list options separated by commas
group
which variable or variables? height,weight
enter ranges, 1 per line for variable: heigh
?58,65
?66,73
? ^z
enter ranges, 1 per line for variable: weigh
?100,150
?151,235
? ^z
statpack v4 PAGE 33
***** chi square *****
levels for horizontal variable: heigh
level values comprising level
1 58.00 , 59.00 , 60.00 , 61.00 , 62.00 ,
63.00 , 64.00 , 65.00
2 66.00 , 67.00 , 68.00 , 69.00 , 70.00 ,
71.00 , 72.00 , 73.00
levels for vertical variable: weigh
level values comprising level
1 100.0 , 102.0 , 104.0 , 106.0 , 108.0 ,
110.0 , 112.0 , 114.0 , 116.0 , 118.0 ,
120.0 , 122.0 , 124.0 , 126.0 , 128.0 ,
131.0 , 133.0 , 135.0 , 137.0 , 139.0 ,
141.0 , 143.0 , 145.0 , 147.0 , 149.0 ,
2 151.0 , 153.0 , 155.0 , 157.0 , 159.0 ,
161.0 , 163.0 , 165.0 , 167.0 , 169.0 ,
171.0 , 173.0 , 175.0 , 177.0 , 180.0 ,
182.0 , 184.0 , 186.0 , 188.0 , 190.0 ,
192.0 , 194.0 , 196.0 , 198.0 , 200.0 ,
202.0 , 204.0 , 206.0 , 208.0 , 210.0 ,
212.0 , 214.0 , 216.0 , 218.0 , 220.0 ,
222.0 , 224.0 , 227.0 , 229.0 , 235.0
heigh
level 1 2 total
weigh ...................
1 . 278 31 309
2 . 75 146 221
total . 353 177 530
chi square = 181.8592 with 1 degrees of freedom
corrected chi square = 179.3489
having a probability of 0.00
contingency coefficient = 0.5028279
statpack v4 PAGE 34
which command? chisq
list options separated by commas
colps
which variable or variables? age
levels for variable: age
level values comprising level
1 18.00
frequency = 45
2 19.00
frequency = 49
3 20.00
frequency = 58
4 21.00
frequency = 64
5 22.00
frequency = 66
6 23.00
frequency = 52
7 24.00
frequency = 64
8 25.00
frequency = 64
9 26.00
frequency = 54
10 27.00
frequency = 59
collapsing portion insert 1 instruction per line after the ?
? c(age)1,3,5
? d(age)2,9
? c(age)6-10
? ^z
statpack v4 PAGE 35
***** chi square *****
levels for variable: age
level values comprising level
1 18.00 , 20.00 , 21.00 , 22.00
frequency = 233
2 23.00 , 24.00 , 25.00 , 27.00
frequency = 239
chi square = 0.7627119e-01 with 1 degrees of freedom
having a probability of 0.78
statpack v4 PAGE 36
command: copys
---------------
purpose: obtain multiple copies of output.
limitation: maximum of 63 copies
description: the "copys" command allows the user to obtain multiple
copies of output. when requested, the user types in the number
of copies desired. up to 63 copies may be requested. if the
number entered is less than or equal to zero, the number of
copies will default to one.
the "copys" command is used in conjunction with the "print"
and "assign" commands. it may be issued at anytime prior to
either the "print" command (for a printing of the data) or the
"fini" command (for the output that was assigned to the line
printer). once the number of copies has been entered, it can
only be changed by another "copys" command.
example:
which command? copys
how many output copies? 3
statpack v4 PAGE 37
command: corr
--------------
purpose: compute the correlation matrix for all variables.
reference: "basic statistical methods", downe and heath, pages 78-90.
description: the "corr" command displays the computed correlation
matrix for all variables in the data set. only the lower
triangular portion of the correlation matrix will be displayed,
adjusting the number of correlations per line to fully utilize
the area available for the output. no responses are necessary
once the command has been given. output will be labeled with
variable names, if they have been defined; otherwise numbers will
be used.
example:
which command? corr
***** correlation matrix *****
var.
scor1 1.0000
2 0.5349 1.0000
index 0.2720 0.4078 1.0000
lengt 0.0895 0.1050 0.0612 1.0000
5 0.0718 -0.4686 -0.4367 -0.0507 1.0000
scor1 2 index lengt 5
statpack v4 PAGE 38
command: corrt
---------------
purpose: calculate correlated t tests
reference: "statistical methods", snedecor and cochran, pages 92-97.
description: the "corrt" command allows the user to calculate
correlated t tests for all combinations of variables. after the
command is given, no other user responses are necessary.
output will be labeled with variable names (if names have
been defined) or variable numbers. results will be adjusted to
fully utilize space available for output.
example:
which command? corrt
***** correlated t *****
iq 0.0000
test1 -25.22 0.0000
test2 -25.27 3.030 0.0000
test3 -31.13 6.117 -0.8085 0.0000
test4 -54.33 9.795 5.702 8.523 0.0000
iq test1 test2 test3 test4
statpack v4 PAGE 39
command: cvsmt
---------------
purpose: forecasting
reference: "decision rules for inventory management", r. g. brown.
description: the "cvsmt" command utilizes an exponential smoothing
model in forecasting cases, where time is the independent
variable and assumed the only condition changing. the user will
first be instructed to enter the variables separated by commas
for which forecasts are to be made. up to 20 variables may be
listed, by variables names (if names have been defined) or
variable numbers. ranges of variables may be specified by typing
the extremes of the range separated by a "-". if forecasts are
to be made for all variables the user may respond with a "*".
the user will next be instructed to enter options separated
by commas. possible options are:
"nlogy"--natural log transformation of dependent variable
(for each observation the natural log of the variable
will be used in the calculations).
"nlogt"--natural log transformation of time
"trend"--user enters trend line
"short"--short output (trend line, variance, projected
values)
"multi"--multiplicative rather than additive seasonal terms
if no options are desired type a <cr>.
the user will now be told to enter the number of
observations per cycle. at least two cycles of data must be
present to establish a forecasting model.
next the user will be asked to enter the weighting factors.
one at a time he will be prompted for the trend, steady state,
and seasonal factors.
finally the user will be instructed to enter the percentage
criteria for seasonal terms. those periods that consistently
deviate from the expected values by this percentage in the first
two cycles are assumed to have seasonal effects.
if the user has specified the "trend" option he will also be
instructed to enter the trend line. this is accomplished by
typing first the y intercept and then the slope when prompted.
statpack v4 PAGE 40
example:
which command? cvsmt
which variables? 1
list options separated by commas
how many observations per cycle? 8
weighting factors for:
steady state? .1
trend? .1
seasonal? .05
how many periods to be projected? 16
what is the percentage criterion for seasonal terms? 25%
***** curve smoothing model *****
for variable: 1 , with add. seasonal terms
percentage criterion for seasonal terms is 25.0%
weighting factors were: 0.100 for steady state
0.100 for trend, and 0.050 for seasonal
the trend line is y= 4.560484 + 0.4747067e-01x
actual predicted deviation % deviation mean avg dev.
1 6.000 6.000 0.0000 0.0000 0.0000
2 7.000 7.000 -0.5960e-07 0.8515e-06 0.2980e-07
3 2.000 2.000 -0.5960e-07 0.2980e-05 0.3974e-07
4 1.000 1.000 -0.4470e-07 0.4470e-05 0.4098e-07
5 5.000 4.798 0.2022 4.214 0.4043e-01
6 3.000 4.868 -1.868 38.37 0.3450
7 7.000 6.819 0.1812 2.657 0.3216
8 6.000 4.742 1.258 26.52 0.4386
9 8.000 6.304 1.696 26.91 0.5783
10 7.000 7.300 -0.2995 4.104 0.5504
11 3.000 2.295 0.7045 30.69 0.5644
12 3.000 1.291 1.709 132.3 0.6598
13 1.000 5.085 -4.085 80.33 0.9233
14 6.000 4.679 1.321 28.23 0.9517
15 8.000 6.943 1.057 15.22 0.9587
16 9.000 4.843 4.157 85.84 1.159
17 6.000 6.793 -0.7927 11.67 1.137
18 7.000 7.703 -0.4027 9.123 1.113
19 4.000 2.763 1.237 44.78 1.119
20 2.000 1.823 0.1772 9.720 1.072
21 6.000 5.545 0.4549 8.204 1.043
22 1.000 5.652 -4.652 82.31 1.207
23 7.000 7.372 -0.3717 5.042 1.171
24 8.000 5.218 2.782 53.32 1.238
25 8.000 6.977 1.023 14.67 1.229
statpack v4 PAGE 41
26 7.000 7.877 -0.8769 11.13 1.216
27 4.000 3.020 0.9802 32.46 1.207
28 5.000 2.013 2.987 148.4 1.271
29 6.000 5.712 0.2881 5.043 1.237
30 5.000 5.787 -0.7868 13.60 .1222
31 6.000 7.897 -1.9897 24.02 1.243
32 7.000 5.784 1.216 21.01 1.243
33 7.445
34 8.257
35 3.500
36 2.601
37 6.158
38 6.208
39 8.314
40 6.309
41 7.847
42 8.660
43 3.903
44 3.003
45 6.560
46 6.611
47 8.717
48 6.711
variance is 3.070
statpack v4 PAGE 42
command: data
--------------
purpose: enter data from terminal.
description: the "data" command allows users to enter data by
terminal according to a standard or user specified format. with
standard format, the user enters values separated by commas with
a maximum of 20 values per line. if 20 values will not fit on
one line, use the "form" command prior to "data". data is
entered by observation, the first value being variable 1, the
second value, variable 2, etc. when finished typing the last
observation, type a ^z(control z).
example:
which command? data
how many input variables? 3
enter input data
1,2,3
6,5,4
7,6.5,3
2,4,8
2.0,3.0,4.5
2,3,4
6,1,8
^z
statpack v4 PAGE 43
command: deass
---------------
purpose: reassign output to terminal from line printer.
description: the "deass" command allows the user to reassign output
to the terminal. once the command has been given, no other user
responses are necessary. this command is used in conjunction
with the "assign" command. it is permissible to channel output
back and forth between the line printer and terminal using the
"assign" and "deass" commands.
example:
which command? deass
output assigned to terminal
statpack v4 PAGE 44
command: desc
--------------
purpose: calculate the mean, standard deviation, and variance for all
variables.
reference: "statistical methods", snedecor and cochran, pages 44-46.
description: the "desc" command displays the mean, standard
deviation, and variance of all variables in the data set. after
the command has been given no other user responses are necessary.
output will be appropriately labeled, with variable names, if
they have been defined; otherwise, numbers will be used to label
the variables.
example:
which command? desc
there are 5 variables and 23 observations
var. means std.dev. variance
scor1 3.000000 3.089572 9.545455
2 4.217391 1.953005 3.814229
index 3.782609 3.029499 9.177866
lengt 5.478261 2.793980 7.806324
5 5.695652 2.457545 6.039526
statpack v4 PAGE 45
command: erana
---------------
purpose: determine the standard error of the mean, the coefficient of
skewness, and the coefficient of variation.
reference: "statistical methods", snedecor and cochran, pages 50,
62-63, 86.
description: the "erana" command enables the user to display the
standard error of the mean, the coefficient of skewness, and the
coefficient of variation for all variables. once the command has
been given, no additional responses are necessary. output will
be appropriately labeled.
example:
which command? erana
var. std err of mean skewness coef. of var.
sex 0.2088243e-01 -0.4522899e-01 97.84924
age 0.1160092 -0.3795715e-01 12.26888
heigh 0.1330046 0.4152664 4.981309
weigh 1.382710 0.5287957 23.29064
iq 0.7284885 -0.9540812e-01 17.29185
gpa 0.2106721e-01 -0.2985082 16.89175
statpack v4 PAGE 46
command: estat
---------------
purpose: incorporates the commands "desc", "basic", and "erana" into
one command.
description: the "estat" command combines the "desc", "basic", and
"erana" commands into one command, allowing the user to obtain
the following statistics: means, standard deviations, variances,
medians, modes, maximums, minimums, standard error of the means,
coefficients of skewness, and coefficient of variation for all
variables. no additional responses are necessary after the
command has been given.
example:
which command? estat
there are 6 variables and 575 observations
var. means std.dev. variance
sex 0.5113043 0.5003074 0.2503075
age 22.65391 2.779382 7.724963
heigh 63.97043 3.186565 10.15419
weigh 142.2348 33.12739 1097.424
iq 100.9339 17.45335 304.6193
gpa 2.988052 0.5047344 0.2547569
var. median mode maximum minimum
sex 1.000000 1.000000 1.000000 0.0000000
age 23.00000 22.00000 27.00000 18.00000
heigh 64.00000 66.00000 73.00000 58.00000
weigh 139.0000 112.0000 251.0000 86.00000
iq 101.0000 113.0000 129.0000 70.00000
gpa 3.050000 2.910000 3.790000 1.850000
var. std err of mean skewness coef. of var.
sex 0.2088243e-01 -0.4522899e-01 97.84924
age 0.1160092 -0.3795715e-01 12.26888
heigh 0.1330046 0.4152664 4.981309
weigh 1.382710 0.5287957 23.29064
iq 0.7284885 -0.9540812e-01 17.29185
gpa 0.2106721e-01 -0.2985082 16.89175
statpack v4 PAGE 47
command: facto
---------------
purpose: calculate factor analysis
reference: "scientific subroutine package".
description: the "facto" command allows the user to calculate a
factor analysis for selected variables. the user will first be
instructed to enter the options separated by commas (if no
options are necessary type a <cr>). possible options are:
"const"--specify constant to decide how many eigenvalues to
retain
"evect"--include eigenvectors in output
"factm"--include factor matrix in output
"varia"--include variances in output
"commu"--include communalities in output
"fscor"--include factor scores in output
"sfscr"--save factor scores
"all"----include all above options
if the "const" option has been specified, the user will
first be asked to supply the constant value for retaining
eigenvalues. unless the "const" option is specified, a constant
value of 1.0 is assumed.
when instructed the user should enter the variables,
separated by commas for which the factor analysis is to be
calculated. up to 40 variables may be entered using either
variable names (if names have been defined) or variable numbers.
ranges of variables may be indicated by typing the extremes of
the range separated by a "-". one or more "*" may be used when
listing the variables to be analyzed. one at a time each
variable not yet specified in the analysis will be substituted
for every "*". those cases where the same variable would be
listed twice in the same analysis will be eliminated, as will be
those cases that except for a switch in the order of the
variables, duplicate an analysis already performed.
statpack v4 PAGE 48
example:
which command? facto
list the options separated by commas
help
possible options:
"evect"-print eigen vectors
"factm"-print factor matrix
"const"-specify constant to decide how many eigenvalues to retain
"varia"-print variances
"commu"-print communalities
"fscor"-print factor scores
"sfscr"-save factor scores
"all" -all above options
if no options are desired type a carriage return
list the options separated by commas
all
what constant value would you like? 1.0
type in the variables, separated by commas
1-9
***** factor analysis *****
variables: 1 2 3 4 5 6 7 8 9
eigen values
2.949910 1.643716 1.555180 1.065827
cumulative percentage of eigenvalues
0.3277678 0.5104030 0.6832007 0.8016259
eigen vectors
vector 1 0.1643755 0.3483587 0.2879729 0.4966078
-0.1680629 -0.3292195 0.3993545 0.1287477e-01
0.4751829
vector 2 0.3483667 0.6551087e-01 -0.4464720 -0.1189327
0.6121037 -0.2642835 0.3886011 -0.2484407
-0.6013648e-01
vector 3 -0.2990021 -0.4682548 -0.2353323 0.1737775
0.1446729 -0.4354536 0.1880096e-01 0.6158744
0.1247019
vector 4 0.5444128 0.1690932 0.3828838 0.4162494e-01
0.3053685 -0.1616322 -0.4341069 0.4028344
statpack v4 PAGE 49
-0.2378870
factor matrix ( 4 factors)
variable: 1 0.2823199 0.4466322 -0.3728760
0.5620458
variable: 2 0.5983167 0.8398985e-01 -0.5839458
0.1745700
variable: 3 0.4946021 -0.5724106 -0.2934754
0.3952851
variable: 4 0.8529389 -0.1524807 0.2167124
0.4297313e-01
variable: 5 -0.2886532 0.7847629 0.1804170
0.3152591
variable: 6 -0.5654445 -0.3388313 -0.5430404
-0.1668673
variable: 7 0.5859035 0.4982158 0.2344608e-01
-0.4481672
variable: 8 0.2211280e-01 -0.3185196 0.7680375
0.4158818
variable: 9 0.8161411 -0.7709947e-01 0.1555118
-0.2455919
iteration variances
cycle
0 0.211289
1 0.336137
2 0.397023
3 0.403007
4 0.405181
5 0.405534
6 0.405587
7 0.405595
8 0.405596
9 0.405596
10 0.405596
11 0.405596
12 0.405596
statpack v4 PAGE 50
rotated factor matrix ( 4 factors)
variable: 1 0.5498048e-01 0.7184287e-01 -0.5578176e-01
0.8501882
variable: 2 0.2932897 -0.3965333 -0.3558099
0.6055122
variable: 3 0.5113546e-01 -0.8249465 0.1506893
0.3298538
variable: 4 0.7404132 -0.4140190 0.2458024
0.1397243
variable: 5 -0.9091001e-01 0.8066388 0.1352510
0.3922847
variable: 6 -0.6828682 -0.2158005 -0.4498387
-0.2050298
variable: 7 0.8699805 0.1829980 -0.3491893
0.8830109e-01
variable: 8 0.3602174e-01 -0.5499635e-01 0.9137716
-0.1596276
variable: 9 0.8053224 -0.3275969 0.9939831e-02
check on communalities
variable original final difference
1 0.73412 0.73412 0.00000
2 0.73650 0.73650 0.00000
3 0.81466 0.81466 0.00000
4 0.79957 0.79957 0.00000
5 0.83111 0.83111 0.00000
6 0.75727 0.75727 0.00000
7 0.92009 0.92008 0.00000
8 0.86478 0.86478 0.00000
9 0.75653 0.75653 0.00000
statpack v4 PAGE 51
factor scores
1 -1.490958 3.832233 -1.067173 -2.106546
2 1.198959 -0.6593465 -0.9128224 2.019913
3 1.424578 -1.453218 1.731582 0.6497848
4 4.805257 -1.572919 2.514179 -0.1405323
5 -5.638395 1.624943 1.808101 -2.416741
6 -3.724880 -1.390020 -1.040081 0.2053468e-01
7 -1.297509 3.529188 -0.6237091e-01 -2.239755
8 1.968393 -0.5822640 -0.5225442 3.182307
9 2.351696 -1.844730 -1.694724 1.499151
10 4.951994 -1.075510 2.141836 0.7312377
11 -4.471908 1.953006 2.304640 -1.182660
12 -3.467720 -1.791392 -0.8820030 -0.8098293
13 1.675869 -1.758742 -1.218959 0.6906289
14 0.5019671 -0.3294900 0.1332481 0.9855452
15 0.3846761 -0.5373469 -0.1268275e-01 0.6160892
16 1.381930 0.6014928 1.045720 1.713483
17 -0.5642744 3.375003 0.5186787 1.977829
18 -0.7676928 2.645897 -1.666019 -0.1859952
19 -0.9984410 1.030133 -2.250437 0.2990398
20 0.1735145 -0.7968859 0.1898848 -1.160393
21 0.6479561 -2.362979 1.079114 -2.574836
22 1.183074 -1.566329 1.541667 -0.5589452
23 -0.2280861 -0.8707238 -0.2156699 -1.009310
statpack v4 PAGE 52
command: fetch
---------------
purpose: enter data which is stored on disk.
limitation: will not input binary files.
description: "fetch" allows the user to read data from the disk
according to a standard format (separated by commas) or a user
specified format. to specify his own format, the user must give
a "form" command prior to the "fetch". in specifying the disk
file to be used for input, the name and extension must be typed
in separated by a period. by adding another user's project-
programmer number enclosed in brackets, the data will be read
from that area assuming the protection is correct.
example:
which command? fetch
what is the filename and extension? file.dat[420,420]
how many input variables? 2
statpack v4 PAGE 53
command: fini
--------------
purpose: terminate execution of stat pack.
description: the "fini" command allows the user to terminate
execution of stat pack. once the command has been given, no
other responses are necessary. output that was assigned to the
line printer will be entered into the print queue for printing.
the user will also be informed of the connect time and cpu time.
note: the "fini" command must be used to ensure printing of results in
instances where the output was assigned to the line printer.
example:
which command? fini
cpu time: 4:22.68 elapsed time: 32:49.12
no execution errors detected
exit
statpack v4 PAGE 54
command: form
--------------
purpose: specify input format.
limitation: no fixed point or alphanumeric formats.
description: when requested, the user enters a floating point format
enclosed in parentheses. the total length of the format
including the parentheses may be up to 480 characters. more than
1 line may be used to enter the format. blanks within lines will
be removed. a format of (20f) is assumed at the beginning of
each stp run, but a "form" command replaces it with the user
specified format.
example:
which command? form
enter your f-type data input format enclosed in parentheses
(2x,3f1.0,1x,2f2.0,1x,f3.0,3x,f2.0)
statpack v4 PAGE 55
command: freq
--------------
purpose: produce frequency tables
reference: "basic statistical methods", downe and heath, pages 16-19.
description: the "freq" command is used to produce frequency of
occurrence tables for one or more variables. the user will first
be asked if he desires percentages, to which the response must be
a "yes" or "no". next the user will be instructed to list the
variables separated by commas for which frequencies are to be
tabulated. up to 40 variables may be entered using either
variable names (if names have been defined) or variable numbers.
ranges of variables may be specified by typing the extremes of
the range separated by a "-". where frequencies are to be
tabulated for all variables, a "*" may be substituted for
variable names and numbers. the tabled results will be labeled,
and size of table adjusted to utilize space available for output.
note: positive and negative numbers as well as multiple digit numbers
may be processed with this command.
example:
which command? freq
do you also want percentages (yes or no)? yes
which variables? age,sex
var. frequency and percentages
----- --------------------------------------------------------
age value 18.0 19.0 20.0 21.0 22.0
freq 45 49 58 64 66
% 7.8% 8.5% 10.1% 11.1% 11.5%
value 23.0 24.0 25.0 26.0 27.0
freq 52 64 64 54 59
% 9.0% 11.1% 11.1% 9.4% 10.3%
sex value .000 1.00
freq 281 294
% 48.9% 51.1%
statpack v4 PAGE 56
command: help
--------------
purpose: answer questions, supply alternatives.
description: the "help" command is used when the response to a
question is not obvious. it will supply the user with an
explanation of what is necessary, the options available, or a
list of commands. once the explanation or information has been
supplied, the question will be restated, allowing the user to
continue.
examples:
which command? help
commands are broken into 6 groups:
"dc" - data control
"es" - elementary statistics
"gr" - graphs
"ia" - item analysis
"as" - advanced statistics
"pc" - program control
"al" - complete command code list
which set (type in the 2 character code)? dc
commands available
"data" - data input by tty
"fetch" - read data from disk
"form" - enter special input format
"manip" - manipulate data in core (includes appending)
"trans" - data transformations
"store" - store data on disk
"print" - print selected variables on line printer
"type" - type out selected variables on tty
"acbnk" - access a stored data bank
"mta/i" - read data from mag tape
which command? fetch
what is the file name and extension? help
the fetch command is used to read data from
the disk. both the file name and extension must be
specified. in order to read from another area the
project, programmer number must be inserted in brackets
directly adjoining the name and extension.
statpack v4 PAGE 57
command: hist
--------------
purpose: create histograms
reference: "basic statistical methods", downe and heath, pages 25-27.
description: the "hist" command allows the user to construct one or
more histograms. when instructed, the user enters the variables
separated by commas for which the histograms are to be created.
up to 20 variables may be entered using either the variable names
(if names have been defined) or variable numbers. ranges of
variables may be indicated by typing the extremes of the range
separated by a "-". where histograms are to be calculated for
all variables, a "*" may be used in place of variable names and
numbers.
scaling will be automatically done with the number of
divisions determined by the space available for output. the
range of values included in each division will be listed in the
output, as well as the frequency and percentage of observations
falling into each division.
example:
which command? hist
which variables? weight,height
statpack v4 PAGE 58
*** histogram for variable: weigh *****
40.00 +
i
i
i
i
30.00 +
i ixxxxxi
i ixxxxxixxxxxi
i ixxxxxixxxxxi
i ixxxxxixxxxxi
20.00 + ixxxxxixxxxxi
i ixxxxxixxxxxixxxxxi
i ixxxxxixxxxxixxxxxi
i ixxxxxixxxxxixxxxxi
i ixxxxxixxxxxixxxxxixxxxxi
10.00 + ixxxxxixxxxxixxxxxixxxxxi
i ixxxxxixxxxxixxxxxixxxxxixxxxxi
i ixxxxxixxxxxixxxxxixxxxxixxxxxi
i ixxxxxixxxxxixxxxxixxxxxixxxxxixxxxxi
i ixxxxxixxxxxixxxxxixxxxxixxxxxixxxxxi
--+-----+-----+-----+-----+-----+-----+-----+
^ 103 ^ 163 ^ 154 ^ 85 ^ 50 ^ 19 ^ 1 ^
^ ^ ^ ^ ^ ^ ^ ^
^ 111. ^ 161. ^ 211. ^ 261.
86.0 136. 186. 236.
***** histogram for variable: heigh *****
40.00 +
i
i
i
i
30.00 +
i
i
i
i ixxxxxi
20.00 + ixxxxxixxxxxi
i ixxxxxixxxxxixxxxxixxxxxi
i ixxxxxixxxxxixxxxxixxxxxi
i ixxxxxixxxxxixxxxxixxxxxi
i ixxxxxixxxxxixxxxxixxxxxi
10.00 + ixxxxxixxxxxixxxxxixxxxxi
i ixxxxxixxxxxixxxxxixxxxxixxxxxi
i ixxxxxixxxxxixxxxxixxxxxixxxxxixxxxxi
i ixxxxxixxxxxixxxxxixxxxxixxxxxixxxxxixxxxxi
i ixxxxxixxxxxixxxxxixxxxxixxxxxixxxxxixxxxxixxxxxi
--+-----+-----+-----+-----+-----+-----+-----+-----+
^ 47 ^ 98 ^ 120 ^ 132 ^ 103 ^ 40 ^ 23 ^ 12 ^
^ ^ ^ ^ ^ ^ ^ ^ ^
^ 60.0 ^ 64.0 ^ 68.0 ^ 72.0 ^
58.0 62.0 66.0 70.0 74.0
statpack v4 PAGE 59
command: info
--------------
purpose: give the user a general program description.
description: the "info" command allows the new user to obtain a brief
description of the program while sitting at the terminal. once
the command has been given, no other responses are necessary.
the text will simply be typed out.
example:
which command? info
stat pack is an integrated statistical package, written for
terminal use. it allows the user to issue simple commands
for data analysis. the program is in conversational mode and
will prompt the user for desired information. in most instances
when questions of procedure arise the user may request further
information, by simply typing "help". standard form of output
is terminal, but output may easily be channeled to the
printer. input is readily accepted from terminal or disk.
input consists of observations, each containing a value for
all of the variables. each observation must begin a new line.
the data may be input in either of two ways: 1) one observation
per line with values separated by commas; or 2) according
to your own input format which is entered using the command
"form". after the last observation enter a ^z (cntrl z).
to see the command list type "help" after "which command?"
and "al" for the 2 character code. a restriction on
data input is: 20 numbers maximum per line under
standard format. if a line requires more than 72
columns use your own input format.
statpack v4 PAGE 60
command: kendl
---------------
purpose: produce a kendall tau correlation matrix.
reference: "non-parametric statistics", siegel, pages 213-219.
description: the "kendl" command allows the user to calculate and
display kendall tau correlations between all variables. once the
"kendl" command is given no other responses are necessary from
the user. output is in the form of a well labeled matrix, with
its size adjusting to the space available.
example:
which command? kendl
***** kendall tau correlation matrix *****
iq 1.0000
test1 0.0091 1.0000
test2 0.0230 0.0981 1.0000
test3 0.0163 0.7648 0.3377 1.0000
test4 0.7208 0.2999 0.0597 0.2834 1.0000
iq test1 test2 test3 test4
statpack v4 PAGE 61
command: kolm
----------------
purpose: calculate one or two sample kolmogorov-smirnov tests.
reference: "statistical methods", b. h. lindgren, pages 329-335.
description: the "kolm" command allows user to calculate one or two
sample kolmogorov-smirnov tests. the user will first be
instructed to enter the options desired separated by commas.
possible options are:
*** one sample options ***
norml--test the variables to be specified against a normal
distribution. (if no options are specified this will be the
default).
expon--test the variables to be specified against an exponential
distribution.
cauch--test the variables to be specified against a cauchy
distribution.
unifm--test the variables to be specified against an uniform
distribution.
total--test the variables to be specified against a normal,
cauchy, uniform, and exponential distributions.
suppl--normally the parameters calculated from the sample will be
used to specify the distribution being tested, however, the
suppl option allows the user to enter his own parameters for
the distributions.
*** two sample options ***
2samp--indicates test will be a two sample kolmogorov-smirnov
test.
break--rather than having the two samples coming from two
variables, select the two samples from a single variable
based on the value of another variable (breakdown variable).
this option is available only if "2samp" has been used.
discr--ranges calculated automatically with a separate range
created for each unique value. if this option is not used
it will be necessary for the user to enter the ranges. if
statpack v4 PAGE 62
more than 20 unique values exist for the breakdown variable,
it will also be necessary for the user to enter ranges.
this option only available if "break" is used.
range--list ranges calculated. this option only available if
"discr" is used.
for the one sample test the user will be instructed to enter
the variables to be tested against the distributions specified.
up to 20 variables may be entered separated by commas. variable
numbers or variable names (if names have been defined) may be
used to indicate the variables. ranges of variables may be
specified by entering the extremes of the range separated by a
"-". if all variables are to be tested the user may use a "*".
no other responses are necessary unless the suppl option was
specified. if "suppl" was used, the user must supply the
parameters for the various distributions when requested.
for the two sample test the user will be instructed to enter
the variables separated by commas. up to 20 variable numbers or
variable names (if names have been defined) may be entered.
ranges of variables may be specified by typing the extremes of
the range separated by a "-". if all variables are to be used an
"*" may be entered. if the break option has not been used, no
other information need be entered. if the break option has been
used, it will be necessary for the user to enter the breakdown
variable. either the variable number or variable name (if names
have been defined) may be used. if the option "discr" has not
been specified the user must enter the ranges for the breakdown
variable. up to 20 ranges for the breakdown variable are
entered, one range per line. each range is comprised of a
minimum for the range, a comma, and a maximum. when the last
range has been entered, type a control z (^z) or carriage return.
statpack v4 PAGE 63
examples:
which command? kolm
enter options separated by commas
total
enter variables to be tested separated by commas
2,3
***** kolmogorov-smirnov test *****
var. distribution tested against d1 prob
grade normal mean= 5.28571 stdev= 1.12646 .1941 0.143*
exponential mean= 5.286 stdev= 1.126 .2688 0.013*
cauchy 1st quart.= 4.250 median= 5.000 .2523 0.023*
uniform from 3.000000 to 7.000000 .2429 0.032*
iq normal mean= 100.000 stdev= 30.0000 .6899e-01 0.996*
exponential mean= 100.0 stdev= 30.00 .1587 0.341*
cauchy 1st quart.= 79.75 median= 96.71 .1060 0.826*
uniform from 42.84538 to 172.0905 .1981 0.128*
the hypothesis that the sample is from the specified
distribution can be rejected with the indicated probability
(prob) of being incorrect.
*the probability may not be correct for the sample
sizes being used. for more accurate probabilities
check the tables in non-parametric statistics by siegel.
which command? kolm
enter options separated by commas
2samp
enter variables separated by commas
*
statpack v4 PAGE 64
***** kolmogorov-smirnov 2 sample test *****
var. size var. size d2 prob
1 60 2 60 0.2333333 0.076*
1 60 3 60 0.4833333 0.000*
2 60 3 60 0.4666667 0.000*
the hypothesis that the 2 samples are drawn from the same
population can be rejected with the indicated probability
(prob) of being incorrect.
*the probability may not be correct for the sample sizes
being used. for more accurate probabilities check the
tables in non-parametric statistics by siegel.
which command? kolm
enter options separated by commas
break,discr,range
break, discr, and range are only available on 2 sample tests
enter options separated by commas
2samp,break,discr,range
enter variables to be analyzed, separated by commas
2,3
which is the breakdown variable? 1
1.000 , 1.000
2.000 , 2.000
***** kolmogorov-smirnov 2 sample test *****
var. smp a size smp b size d2 prob
2 1 29 2 31 0.1434928 0.917*
3 1 29 2 31 0.1768632 0.737*
the hypothesis that the 2 samples are drawn from the same
population can be rejected with the indicated probability
(prob) of being incorrect.
*the probability may not be correct for the sample sizes
being used. for more accurate probabilities check the
tables in non-parametric statistics by siegel.
statpack v4 PAGE 65
which command? kolm
enter options separated by commas
break,2samp
enter variables to be analyzed, separated by commas
2,3
which is the breakdown variable? 1
enter ranges for breakdown variable 1
? 1,1
? 2,2
?
***** kolmogorov-smirnov 2 sample test *****
var. smp a size smp b size d2 prob
2 1 29 2 31 0.1434928 0.917*
3 1 29 2 31 0.1768632 0.737*
the hypothesis that the 2 samples are drawn from the same
population can be rejected with the indicated probability
(prob) of being incorrect.
*the probability may not be correct for the sample sizes
being used. for more accurate probabilities check the
tables in non-parametric statistics by siegel.
statpack v4 PAGE 66
command: mabnk
---------------
purpose: create a bank from data in stp.
description: the "mabnk" command allows users to make a bank file
from data located in stp. when the name for the bank is
requested, the user should type in a name of at most 6
characters; no extension is necessary as a ".bnk" will be
supplied. all variables in stp will be put into the bank in the
same order as they occur in stp. if variable names are not
assigned in stp, generated names of "v1", "v2", "v3", ..., will
be given to the variables in the bank. a standard protection
code of <155> will be assigned to the bank unless otherwise
specified by the user.
banks are stored as binary random access files to increase
recovery speed, and eliminate formatting problems. do not
attempt to print bank files; output will not be meaningful, and
large amounts of line printer paper will be wasted.
example:
which command? mabnk
bank name? john
what protection? 177
statpack v4 PAGE 67
command: make
--------------
purpose: document output with one or more pages of user submitted
information.
description: the "make" command allows the user to insert one or more
pages of documentation into program generated output. after the
make command is issued, the user is instructed to enter his page
or pages of documentation. when finished, the user types a <cr>,
an altmode, and another <cr>. the information submitted by the
user is entered into the output only once, at that point in the
output where the "make" command was requested. the text is
transferred to the output line by line, thus corrections to
previous lines are not possible. if corrections are to be made
to the line currently being typed in, rubouts may be used. each
rubout used erases the previous character. lines may be up to
120 characters long.
a disk file may also be used as the text to be inserted into
the output. to specify a disk file the user types an "@" and the
name of the file with the extension. the file must reside in the
area from which stat pack is being run.
examples:
which command? make
*** documentation ***
this output is from the stat pack program run on 4-12-74.
input is from the file oberl.dat in area 220,220.
its variables are defined as follows:
var var
num name description
--- ---- -----------
1 sex sex of respondant 1-female 2-male
2 age age in years (nearest birthday)
3 gpa grade point average (4.0=a)
$
which command? make
@sampl.doc
statpack v4 PAGE 68
command: manip
---------------
purpose: correct mistakes in data, make allowances for missing data,
add variables or observations, and modify or discard observations
which do not meet certain criteria.
description: "manip" is a command allowing the user to issue
instructions one at a time, to delete, type, modify, or add to
data which has been entered in statpack. each instruction has
the following three parts:
a. what operation is to be performed
each instruction must have as the first character a code
indicating what the instruction does.
"a"-add to the data
"d"-delete data
"e"-return to "which command?"
"r"-replace data with new values
"t"-type data on terminal
b. what portion of the data is to be acted upon by the operation
this portion of the "manip" instruction indicates which
section of the data is to be acted upon. the observation number
and variable number or name are used for such reference. to
indicate a particular observation an "o" is followed by the
observation number. to define a variable a "v" is followed by
the variable number or if names are defined for the variables,
the name enclosed in parentheses. ranges of variables or
observations may be indicated by following the "v" or "o" with
the smallest number in the range, a "-" (dash), and finally with
the largest number in the range. when only those values which
meet certain criteria are to be acted upon, a search is used.
the coding for a search consists of an "s" followed by a
conditional code and a numerical value used as a reference for
the conditional code. these codes are:
">" or "g"-greater than
"<" or "l"-less than
"=" or "e"-equal to
the condition code is modified by the value which follows
it, indicating the value to be searched for and the relationship
that must be satisfied in order to act upon the data.
c. what extra information is desired or previously defined
during execution of instruction strings, a report is written
on the terminal containing variable names or numbers, observation
numbers, and values at these points. if the report is
unnecessary or unwanted, a "w" will cause it to be suppressed.
statpack v4 PAGE 69
normally, when an instruction has been issued to replace or
add data, the values to be inserted are supplied by the user, one
at a time in response to a single question mark. a constant
value to be used for that instruction may be supplied
(eliminating the need to type in data upon cue), by inserting a
"c" followed by the numeric value to be used as the constant.
two special constants are also available: the mean of the
variable, and the mean of the variable calculated by ignoring all
occurrences of a particular value. if the mean of the variable
is to be used as the constant, replace the "c" and its associated
value by an "m". if the constant to be used is the mean
calculated by ignoring a particular value, as with missing data,
replace the "m" with an "l" followed by the value to be ignored
during calculation of the mean.
when typing an instruction to "manip" there may be no
spaces, and the first character must be the code indicating what
is to be done. no particular order must be maintained beyond the
first character. a double question mark indicates an instruction
should be inserted. a single question mark is used to indicate a
numeric value is to be inserted for data.
examples:(assume original data to be)
variable variable variable variable
(1) (2) (3) (4)
observation 1 1 3 5 7
observation 2 2 9 4 3
observation 3 6 8 4 3
observation 4 2 1 7 9
observation 5 5 8 6 8
observation 6 1 1 3 2
observation 7 4 7 3 1
observation 8 6 2 3 5
which command? manip
?? to1 type the values for all
variables in observation 1
var. obs value
1 1 1.00
2 1 3.00
3 1 5.00
4 1 7.00
statpack v4 PAGE 70
?? to1w type the values for all
variables in observation 1
leaving off the variable and
observation identification
1.00
3.00
5.00
7.00
?? tv2w type the values for all
observations in variable: 2
leaving off the variable and
observation identification
3.00
9.00
8.00
1.00
8.00
1.00
7.00
2.00
?? to2-3v2-3 type the values of observa-
tions 2 and 3, for variables
2 and 3
var. obs value
2 2 9.00
2 3 8.00
3 2 4.00
3 3 4.00
?? tse3 type all the values equal to
3 identified by variable and
observation numbers
var. obs value
2 1 3.00
3 6 3.00
3 7 3.00
3 8 3.00
4 2 3.00
4 3 3.00
?? tv2se3 type all those cases where
variable: 2 is equal to 3
var. obs value
2 1 3.00
statpack v4 PAGE 71
?? ro4v3 replace the value (7) in
observation 4 of variable: 3
with the new value (8)
entered from the terminal
in response to the
question mark
var. obs value new value
3 4 7.00 ?8
?? ro4v3c7 replace the value (8) in
observation 4 of variable:
3 with the constant value 7
var. obs value new value
3 4 8.00 7.00
?? rse3v3c8 replace those observations
in variable 3 having the
value (3) with the constant
value (8)
var. obs value new value
3 6 3.00 8.00
3 7 3.00 8.00
3 8 3.00 8.00
?? rv3o6 replace observation 6 of
variable: 3 with the new
value (3) entered from the
terminal in response to the
question mark
var. obs value new value
3 6 8.00 ?3
?? rv3o7-8c3 replace the value (8) in
observations 7 and 8 of
variable 3 with the constant
value (3.0)
var. obs value new value
3 7 8.00 3.00
statpack v4 PAGE 72
?? ao9 create a new observation (9),
with the values 3,5,7, and 4
(entered from terminal) being
used for variables 1-4
var. obs new value
1 9 ?3
2 9 ?5
3 9 ?7
4 9 ?4
?? av5c3.0 create a new variable (5)
where all observations are
automatically set equal to the
constant value 3
var. obs new value
5 1 ? 3.00
5 2 ? 3.00
5 3 ? 3.00
5 4 ? 3.00
5 5 ? 3.00
5 6 ? 3.00
5 7 ? 3.00
5 8 ? 3.00
5 9 ? 3.00
?? dv5 delete variable 5
?? ro9m replace observation 9 with the
mean of each variable. thus
variable 1, observation 9 will be
equal to the mean of variable 1;
variable 2, observation 9 will
be equal to the mean of variable
2, etc.
var. obs value new value
1 9 3.00 3.33
2 9 5.00 4.89
3 9 7.00 4.67
4 9 4.00 4.67
?? ro9 replace the values for all
variables in observation 9 with
values entered on the terminal in
in response to question marks
var. obs value new value
1 9 3.33 ?3
2 9 4.89 ?5
3 9 4.67 ?7
4 9 4.67 ?4
statpack v4 PAGE 73
?? ro9l3 replace observation 9 of all
variables with the mean of each
variable calculated by discarding
all the observations having a
value 3 for that variable
var. obs value new value
1 9 3.00 3.37
2 9 5.00 5.13
3 9 7.00 5.50
4 9 4.00 5.14
?? do9 delete observation 9
?? t type the entire set of data
var. obs value
1 1 1.00
1 2 2.00
1 3 6.00
1 4 2.00
1 5 5.00
1 6 1.00
1 7 4.00
1 8 6.00
2 1 3.00
2 2 9.00
2 3 8.00
2 4 1.00
2 5 8.00
2 6 1.00
2 7 7.00
2 8 2.00
3 1 5.00
3 2 4.00
3 3 4.00
3 4 7.00
3 5 6.00
3 6 3.00
3 7 3.00
3 8 3.00
4 1 7.00
4 2 3.00
4 3 3.00
4 4 9.00
4 5 8.00
4 6 2.00
4 7 1.00
4 8 5.00
?? e exit
statpack v4 PAGE 74
command: mann
--------------
purpose: calculate mann-whitney u
reference: "basic statistical methods", downe and heath, pages 213,
214. "non-parametric statistics", siegel, pages 116-127.
description: the "mann" command allows the user to calculate
mann-whitney u. the user will first be instructed to list the
options desired, separated by commas. if no options are desired
type a <cr>. possible options are:
"break"--select samples from one variable based on the value of a
second variable. for each observation, the value of the
second variable (breakdown variable) will be used to
determine in which sample the variable being analyzed
belongs. this is accomplished by determining which of a
series of ranges the value of the breakdown variable fits
into, and then moving the value of the analysis variable to
the corresponding sample. (if this option is not used,
mann-whitney u's will be calculated between variables.)
note: the following options are to be used only if "break" has
been used.
"discr"--automatic breakdown. instead of the user entering
ranges, a separate range will be created automatically for
each value in the breakdown variable.
"auto"--automatic breakdown. this option is the same as the
"discr" option. do not enter "auto" with the other options,
it should be used only when asked to enter the ranges. the
"discr" and "auto" options are equivalent; the only
difference is at which point in the program they are
entered.
note: the following option is available only if automatic
breakdown is to be used.
"range"--list the ranges to be used for the automatic breakdown.
if the "break" option has not been specified, mann-whitney u
will be calculated between all possible pairs of variables. no
other user responses will be necessary.
if the "break" option has been used, it will be necessary
for the user to supply the following information:
statpack v4 PAGE 75
(1) the variables for which mann-whitney u are to be calculated
(up to 20). the samples for each set of mann-whitney u will
be selected from a single variable. variables may be listed
using either variable names (if names have been defined) or
variable numbers. ranges of variables may be specified by
listing the extremes of the range separated by a "-". where
mann-whitney u's are to calculated for all variables, a "*"
may be used instead of variable names or numbers.
(2) the variable to be used for the breakdowns. only one
variable may be entered, specified by either its variable
name (if the name has been defined) or variable number. all
variables listed for analysis will be processed using the
same breakdown variable.
(3) ranges for the breakdown variable. if the "discr" option
has been used, this information will be automatically
calculated, and need not be supplied by the user. if the
"discr" option has not been used, the user may still request
the ranges be automatically calculated by responding with
"auto". to specify ranges, the user types the extremes of
range, smaller first, separated by a comma. only one range
may be entered per line. up to 50 ranges may be specified.
after the last range has been entered, the user types a
^z(control z).
example:
which command? mann
list options separated by commas
***** mann-whitney u-test *****
var. vs var. mean standard deviation z
n1 n2 u1 u2
----------------------------------------------------------------
iq test1 156240.5 5398.042 28.90354
559 559 312263.0 218.0000
iq test2 156240.5 5398.042 28.88862
559 559 312182.5 298.5000
test1 test2 156240.5 5398.042 -6.136114
559 559 123117.5 189363.5
statpack v4 PAGE 76
command: mta/i
---------------
purpose: read data stored on magnetic tape, with the ability to
select variables, and subset data (leave out observations which
do not meet user specified criteria).
limitation: data must be in ascii mode on magnetic tape, line-blocked
and recorded at 556 bpi. (these are normal w.m.u. magnetic tape
standards.) tape must be located at computer center or given to
operator prior to run.
description: the "mta/i" command is used to read data located on
magnetic tapes. if the format of the tape is other than the
standard format (20f) the "form" command should be used prior to
the "mta/i" command. the user will first be instructed to type
in some identification for the tape. a single line of 26
characters should be typed in, containing tape i.d. no.(if
available), name of user, or other identification. this message
is sent to the operator, along with the project-programmer number
of the user. the operator has a choice of five responses; they
are: no drives are available, tape cannot be located, the user
does not have one of the project-programmer numbers for which the
tape is reserved, more information needed, or the tape has been
mounted.
if a magnetic tape drive is free and the user may access the
requested magnetic tape, the magnetic tape will be mounted on the
free drive and the user informed of the drive number. if the
tape has not been mounted, check the following list to determine
proper course of action:
operator's reply action to be taken
---------------- ------------------
no drives available wait and request tape later
tape cannot be located 1. check and make sure right
i.d. number was used
2. make sure tape is at
computer center
3. contact operator to
determine problem
users project-programmer the owner of the tape must
number does not appear on the send a signed note to the
list of numbers which may to the computer center
access the tape containing any additions or
changes of proj-prog numbers
which may access the tape.
statpack v4 PAGE 77
operator needs more type in another line of up to
information 30 characters of i.d.
the user will then be asked which position on the tape the
file he wishes to access occupies. if the file is the only one
on the tape or was the first one written on the tape, then it
occupies position number 1. if it was the second one written on
the tape, then it occupies position number 2, etc. when the user
has answered this question, the magtape will advance to the
proper file. the user will then be asked how many variables it
will be necessary to look at. this includes variables which may
only be used as qualifiers. next, when instructed, the user
types in the qualifiers (50 maximum), one per line with no
spaces. each qualifier checks one variable to make certain it
contains a specified value or range of values. if the qualifier
is not satisfied the observation being scrutinized is discarded,
and the next observation considered. each qualifier is composed
of three parts: the number of the variable to be checked, the
relationship which must be satisfied if the observation is to be
accepted, and the value which the variable is compared against.
the relationships possible are:
,eq, - equal
,ne, - not equal
,lt, - less than
,gt, - greater than
,le, - less than or equal to
,ge, - greater than or equal to
after the last qualifier has been typed, a <cr>, ^z, or
"stop" may be typed. the user is then instructed to list the
numbers of the variables which are to be used as data, separated
by commas. the first variable listed becomes variable number 1
in the data set, the second variable listed becomes variable
number 2, etc. when either the entire file has been considered
or the data set has been filled, the user will be informed of the
number of observations in the data set, and the number of records
from which the data was selected. when finished, before logging
off the computer, indicate to the operator to remove the magnetic
tape from the drive.
example:
which command? mta/i
please give some identification for the tape? mag tape #1234
tape has been mounted on mta0 write protected. be
sure to ask to have the tape dismounted when done
what position does the file occupy on the tape? 1
statpack v4 PAGE 78
how many variables? 4
list qualifiers 1 per line
? 2,le,9
? 3,ge,0
? ^z
list the variables to be kept, separated by commas
1,2,4
data set consists of 36 observations
as selected from a sample of 36
statpack v4 PAGE 79
command: name
--------------
purpose: assign names to variables
limitation: maximum of five characters per name
description: the "name" command allows the user to assign names to
variables. one at a time, each variable number will be typed out
followed by a question mark. in response, the user may type in
the name he has selected for that variable, subject to the
following limitations:
(1) maximum of five characters
(2) first character must be alphanumeric
(3) ";", ",", "-", or blanks must not be present in the name
(4) "all", "stop", "help", "empty", and "obs" are not legal
names
(5) two variables must not be given the same name
if it is not necessary to name a variable, or if the
variable has been previously given a name and it does not need to
be changed, the user may type a <cr>.
names will be kept only until the end of the run or until a
new set of data is read. the "store" command does not store
names with the data, if the names are necessary, they will have
to be input each time the data is fetched.
the names are primarily used to label output, but they may
be used whenever it is necessary to specify a variable.
example:
which command? name
var 1? iq
var 2? test1
var 3? test2
var 4? test3
var 5? test4
statpack v4 PAGE 80
command: 1wayr
---------------
purpose: calculate one way analysis of variance with repeated
measures.
reference: "statistical principles in experimental design", winer,
pages 105-113.
description: the "1wayr" command allows the user to calculate a one
way analysis of variance with repeated measures. when
instructed, the user lists the variables separated by commas for
which he wishes to have an analysis of variance calculated. up
to 50 variables may be entered, using either variable names (if
names have been defined) or variable numbers. ranges of
variables may be entered by typing the extremes of the range
separated by a "-".
one or more "*" may be used when listing the variables to be
analyzed. one at a time each variable not yet specified in the
analysis will be substituted for every "*". those cases where
the same variable would be listed twice in the same analysis will
be eliminated, as will be those cases which, except for a switch
in the order of the variables, duplicate an analysis already
performed.
output will be labeled with variable names, if available;
otherwise, variable numbers will be used. output size will be
adjusted to fully utilize space available.
example:
which command? 1wayr
which variables? test1,test2,test3,test4
***** 1-way anova with repeated measures *****
tret. size mean std. dev.
test1 559 41.75 21.73741
test2 559 49.68 18.73237
test3 559 47.88 12.27697
test4 559 64.81 19.69569
source sum of sq. d.f. mean sq. f prob
between 403319.0 558
within 517612.4 1677
treat. 160901.6 3 .5363e+05 251.7 0.0000
resid. 356710.8 1674 213.1
total 920931.4 2235
statpack v4 PAGE 81
command: pcent
---------------
purpose: calculate user specified percentiles
reference: "basic statistics", downe and heath, pages 36,37.
description: the "pcent" command allows the user to calculate
selected percentiles for one or more variables. when prompted,
the user lists the variables separated by commas for which the
percentiles are to be calculated. up to 20 variables may be
entered using either the variable names (if names have been
defined) or variable numbers. ranges of variables may be
indicated by typing the extremes of the range separated by a "-".
where percentiles are to be calculated for all variables, a "*"
must be substituted for variable names and numbers.
the user will then be instructed to list the percentiles he
wishes to have calculated separated by commas. up to 20
percentiles may be entered. if either the deciles (10th, 20th,
30th,...80th, 90th percentiles) or quartiles (25th, 50th, and
75th percentiles) are desired, they may be obtained by responding
with "dec" for the deciles or "qtr" for the quartiles.
examples:
which command? pcent
which variables? height,weight,iq,gpa
type in percentiles you wish to have, separated by commas
dec
***** percentiles *****
variables
percentile heigh weigh iq gpa
---------- --------------------------------------------
10.00 60.00 102.0 76.00 2.250
20.00 61.00 112.0 82.00 2.535
30.00 62.00 120.0 89.00 2.690
40.00 63.00 131.0 96.00 2.890
50.00 64.00 139.0 101.0 3.050
60.00 65.00 148.0 108.0 3.190
70.00 66.00 157.0 113.0 3.340
80.00 66.00 170.0 119.0 3.500
90.00 68.00 188.0 125.0 3.640
statpack v4 PAGE 82
which command? pcent
which variables? height,iq,weight,gpa
type in percentiles you wish to have, seperated by commas
5,10,15,20,25,50,75,80,90,95,99
***** percentiles *****
variables
percentile heigh iq weigh gpa
---------- -----------------------------------------------
5.00 59.00 73.00 94.50 2.085
10.00 60.00 76.00 102.0 2.250
15.00 60.00 80.00 106.0 2.387
20.00 61.00 82.00 112.0 2.535
25.00 61.00 85.25 116.0 2.620
50.00 64.00 101.0 139.0 3.050
75.00 66.00 116.0 163.0 3.400
80.00 66.00 119.0 170.0 3.500
90.00 68.00 125.0 188.0 3.640
95.00 70.00 127.0 204.0 3.720
99.00 72.00 129.0 226.2 3.780
statpack v4 PAGE 83
command: pcorr
---------------
purpose: calculate partial correlations
reference: "an introduction to multivariate statistical analysis",
t.w. anderson, pages 31-32, 86-87.
description: the "pcorr" command allows the user to calculate a
partial correlation matrix for specified variables. when
instructed, the user enters the variables separated by commas,
for which partial correlations are to be calculated. up to 20
variables may be entered, using either variable names (if names
have been defined) or variable numbers. ranges of variables may
be entered by listing the extremes of the range separated by a
"-". one or more "*" may be used when listing the variables to
be analyzed. one at a time each variable not yet specified in
the analysis will be substituted for every "*". those cases
where the same variable would be listed twice in the same
analysis will be eliminated, as will be those cases that except
for a switch in the order of the variables, duplicate an analysis
already performed.
output will be labeled with variable names, if available;
otherwise, variable numbers will be used. output size will be
adjusted to fully utilize space available.
example:
which command? pcorr
which variables? gpa,iq,sex,age,weight,height
***** partial correlation matrix *****
iq 0.0139
sex 0.0001 -0.0090
age 0.0400 -0.0463 0.0480
weigh 0.0368 0.0674 0.3946 0.0380
heigh -0.0023 -0.0395 -0.0345 -0.0869 0.7753
gpa iq sex age weigh
statpack v4 PAGE 84
command: plot
--------------
purpose: produce bivariate scatter plot
reference: "basic statistical methods", downe and heath, pages 79-82.
description: the "plot" command allows the user to produce a
bivariate (two variable) scatter plot. scatter plots may be
produced for the following:
(1) a single variable vs a single variable
(2) a single variable vs all other variables
(3) all variables vs all variables
it will be necessary for the user to supply first the
horizontal and then the vertical axes. either variables names
(if names have been defined) or variable numbers may be used.
where all variables are to be used a "*" may be substituted for
variable names or numbers. the plot is automatically scaled in
both the horizontal and vertical directions, and its size is
adjusted to utilize space available.
the digits 1 through 9 indicate the number of observations
occupying a single point in the graph. letters (a-z) are used
when more than 9 observations occupy the same location, starting
with a to indicate 10 numbers, b to indicate 11, etc., and z to
indicate 35. if more than 35 observations occupy a single point
a "*" will be used.
example:
which command? plot
which is the horizontal variable? height
which is the vertical variable? weight
statpack v4 PAGE 85
plot of variable heigh (horiz.) vs variable weigh (vert.)
i---------+---------+---------+---------+
251.0 + 1
i
i 1 1
i 1 3 2
219.0 + 1 2 1 1 1 1 1
i 1 2 1 2 1
i 2 1 4 2 2 3
i 2 1 1 1 4 2 1 2
187.0 + 1 5 5 5 2 1 1
i 4 8 3 3 1
i 3 d a 3 4 1 1
i 1 6 3 9 7 3
155.0 + 2 4 6 c 6 e 4 1 3
i 1 2 1 7 9 f 5 4 1
i 1 7 7 f 7 8 9 2 1
i 4 9 9 9 d 4 6 1
123.0 +1 6 7 9 7 9 2 4
i1 9 c d 4 c 7 4 1
i a 4 3 9 7 2 1
i d 7 7 4 6 1
91.00 + a 8 5 3
i 1 2
i
i
59.00 +
i---------+---------+---------+---------+
57.60 65.60 73.60
61.60 69.60
statpack v4 PAGE 86
command: print
---------------
purpose: print selected variables on the line printer.
description: the "print" command is used to obtain a copy of the data
listed on the line printer. when asked for the variables to be
printed, the user types the variables on one line separated by
commas. variables may be entered using either variable names (if
names have been defined) or variable numbers. ranges of
variables may be specified by listing the extremes of the range
separated by a "-". where all variables are to be printed, use a
"*" instead of variable names or numbers. output is always to
the line printer with multiple copies available by using the
"copys" command prior to the "print" command. printouts may be
picked up at the output window in the computer center, by asking
for the printout and giving the users project-programmer number.
the computer center is located on the third floor of rood hall.
the "type" command is also available for copies of data in
smaller quantities, via the terminal.
example:
which command? print
which variables? sex,age,3,5-6
example output:
var
obs sex age 3 5 6
1 1.000000 23.00000 4.000000 9.000000 7.000000
2 1.000000 21.00000 6.000000 8.000000 9.000000
3 1.000000 25.00000 7.000000 9.000000 6.000000
4 2.000000 18.00000 6.000000 5.000000 3.000000
5 2.000000 18.00000 5.000000 6.000000 7.000000
6 2.000000 19.00000 7.000000 9.000000 6.000000
7 1.000000 16.00000 6.000000 7.000000 5.000000
8 2.000000 24.00000 5.000000 7.000000 8.000000
9 2.000000 21.00000 4.000000 7.000000 8.000000
10 2.000000 20.00000 3.000000 4.000000 2.000000
statpack v4 PAGE 87
command: prob
--------------
purpose: calculate probabilities associated with t tests, f tests,
and chi squares.
description: the "prob" command allows the user to calculate
probabilities associated with t tests, f tests, and chi squares.
to indicate to the user that an instruction should be inserted, a
"?" will be typed out. this will be the first reply to the
"prob" command, and will appear after each probability
calculated. instructions may be entered in either of two ways:
(1) a code indicating the type of probability desired (the user
will be prompted for additional information). the codes
are:
chi--chi square
t----t test
f----f test
(2) the code and information are entered in one line (no
prompting except for "?" is necessary here). the codes and
information are entered in the following manner, with no
spaces between any characters:
chi square
----------
chi##,df
the three character code "chi" followed by the
value of the chi square (the chi square is
substituted for the "##" in the above string).
this is followed by a comma, and finally, the
degrees of freedom (degrees of freedom substituted
for "df".)
t test
------
t##,df
the single character code "t" followed by the
value of the t score (the t score is substituted
for the "##" in the above string). this is
followed by a comma, and finally the degrees of
freedom (degrees of freedom substituted for "df").
f tests
-------
f##,ndf,ddf
the single character code "f" followed by the
value of the f score (f score substituted for the
"##" in the above string). next a comma, and then
the degrees of freedom in the numerator (degrees
statpack v4 PAGE 88
of freedom in the numerator will replace the "ndf"
in above string). finally, a comma followed by
the degrees of freedom in the denominator (degrees
of freedom in the denominator will replace the
"ddf" in above line).
included in the output will be a statement of the parameters
supplied as well as the probabilities calculated. both the
one-tailed and two-tailed probabilities will be calculated for
the t tests. to return to "which command?" enter an "exit",
<cr>, or ^z(control z).
examples:
which command? prob
"?" indicates program is waiting for instruct.
? chi
what is the value of the chi. sq.? 5.892
how many degrees of freedom? 13
the prob for a chi sq of 5.892 with 13 degrees of freedom is .95
? chi5.892,13
the prob for a chi sq of 5.892 with 13 degrees of freedom is .95
? chi5.892
how many degrees of freedom? 13
the prob for a chi sq of 5.892 with 13 degrees of freedom is .95
? t
what is the value of the t? 2.571
how many degrees of freedom? 5
the prob. for a t of 2.571 with 5 degrees of freedom is:
one tailed 0.0250; two tailed 0.0500
? t2.571
how many degrees of freedom? 5
the prob. for a t of 2.571 with 5 degrees of freedom is:
one tailed 0.0250; two tailed 0.0500
? t2.571,5
the prob. for a t of 2.571 with 5 degrees of freedom is:
one tailed 0.0250; two tailed 0.0500
statpack v4 PAGE 89
? f
what is the value of the f? 27.34
how many degrees of freedom in the numerator? 9
how many degrees of freedom in the denominator? 3
prob for an f of 27.34 with 9 degrees of freedom in the numerator
and 3 degrees of freedom in the denominator is 0.0100
? f27.340
how many degrees of freedom in the numerator? 9
how many degrees of freedom in the denominator? 3
prob for an f of 27.34 with 9 degrees of freedom in the numerator
and 3 degrees of freedom in the denominator is 0.0100
? f27.34,9,3
prob for an f of 27.34 with 9 degrees of freedom in the numerator
and 3 degrees of freedom in the denominator is 0.0100
? ^z
statpack v4 PAGE 90
command: ptbis
---------------
purpose: calculate point biserial correlations
reference: "basic statistical methods", downe and heath, pages 169-172
description: the "ptbis" command allows the user to calculate point
biserial correlations. when requested, the user enters up to 20
variables separated by commas. variable numbers or variable
names (if names have been defined) may be used. ranges of
variables may be specified by typing the extremes of the range
separated by a "-".
the user will next be instructed to enter the dichotomous
variable. either the variable number or variable name (if names
have been defined) may be used. if more than 2 unique values
exist, it will also be necessary to specify a breakpoint for the
dichotomous variable. all values less than or equal to the
breakpoint will be treated as one portion of dichotomy while all
values greater than the breakpoint will be the other portion.
when instructed the user enters a value for the breakpoint, or if
the median is to be used as the breakpoint "median" may be
specified.
examples:
which command? ptbis
which variables? 2,3
which is the dichotomous variable? 1
**** point biserial correlation ****
variable: rorw is the dichotomous variable
the lower group size is 14 and the upper group size is 21
point-biserial
mean of mean of standard correlation with
variable low group high group deviation variable:rorw
grade 5.0714 5.4286 1.1265 0.1553218
iq 93.671 104.22 30.000 0.1722583
statpack v4 PAGE 91
which command? ptbis
which variables? 2
what is the breakpoint for variable: iq ? median
**** point biserial correlation ****
variable: iq is the dichotomous variable
the breakpoint being used to split the variable is 96.70642
the lower group size is 18 and the upper group size is 17
point-biserial
mean of mean of standard correlation with
variable low group high group deviation variable:iq
grade 5.5000 5.0588 1.1265 -0.1957447
which command? fini
statpack v4 PAGE 92
command: regr
--------------
purpose: produce multiple regressions
reference: "statistical methods", snedecor and cochran, pages 381-418
description: the "regr" command allows the user to produce multiple
regression on selected variables. when instructed, the user
enters the option desired. only one option exists for the "regr"
command. it is:
"resid"--store residuals
if this option is not to be used, type a <cr>.
if the "resid" option has been specified the user will be
asked to indicate under which variable the residuals are to be
stored. the variable name (if the name already exists) or the
variable number may be used in storing the residuals. once the
residuals have been stored in the variable, the variable name
will be changed to "resid".
the user will be instructed to list the independent
variables separated by commas. up to 19 variables may be listed
by variable names (if names have been defined) or variable
numbers. ranges of variables may be indicated by typing the
extremes of the range separated by a "-". one or more "*" may be
used when listing the variables. one at a time each variable not
yet specified in the analysis will be substituted for every "*".
those cases where the same variable would be listed twice in the
same analysis will be eliminated, as will those cases which
except for a change in the order of variables, duplicate an
analysis already performed.
finally, the user will be asked to enter the dependent
variable. either the variable name (if names have been used) or
the variable number may be used. if the user wishes to
substitute in one at a time all those variables not specified as
independent, he may respond with a "*". the "*" may be used both
as the dependent variable, and as one or more independent
variables in the same analysis.
statpack v4 PAGE 93
example:
which command? regr
enter options separated by commas
list the independent variables?
sex,age,gpa
which is the dependent variable? iq
***** multiple linear regression *****
sample size 575
dependent variable: iq
independent variables: sex age gpa
coefficient of determination 0.00336
multiple corr coeff. 0.05797
estimated constant term 105.10621
standard error of estimate 17.469706
analysis of variance
for the regression
source of variation df s. sq. m.s. f prob
regression 3 587.599 195.866 .6418 0.5923
residuals 571 174264. 305.191
total 574 174851.
regression s. e. of f-value corr.coef.
var. coefficient reg. coef. df (1, 571) prob with iq
sex 1.132208 1.459 .6020 0.4382 0.0318
age -0.2885458 .2626 1.207 0.2724 -0.0443
gpa 0.5975427 1.447 .1705 0.6798 0.0169
statpack v4 PAGE 94
command: sort
--------------
purpose: allow user to sort data into ascending order
description: data may be sorted into ascending order by use of the
"sort" command. when instructed to enter sort keys, the user
types in up to 20 variables separated by commas. variable names
(if names have been given) or variable numbers may be used.
ranges of variables may be specified by typing the extremes of
the range separated by a "-". the sort fields should be listed
from major to minor. when the sort has been completed, each
observation will remain unaltered, only the order in which the
observations occur will have been changed. the major sort key
(first variable in the list) is used to determine which of two
observations is first. if no decision can be made as in the case
of a tie, the next variable in the list of sort keys is used.
the sort proceeds in this manner always checking the next
variable in the list, until it reaches the minor sort key (the
last variable in the list). if no decision can yet be made, they
are left in the same order as they occur.
example:
which command? sort
list sort variables major to minor? 1-3
data sorted by variables: sex , age , weight
statpack v4 PAGE 95
command: srank
---------------
purpose: produce spearman rank-order correlations
reference: "non-parametric statistics", siegel, pages 202-213
description: the "srank" command allows the user to calculate
spearman rank-order correlations between all variables. after
the command is given no other responses are necessary.
example:
which command? srank
***** spearman rank-order corr. matrix *****
iq 1.0000
test1 0.0138 1.0000
test2 0.0341 0.1449 1.0000
test3 0.0229 0.9309 0.4813 1.0000
test4 0.9015 0.4325 0.0903 0.4124 1.0000
iq test1 test2 test3 test4
statpack v4 PAGE 96
command: stepr
---------------
purpose: produce stepwise regressions
reference: "mathematical methods for digital computers", ralston and
wilf, pages 191-203.
description: the "stepr" command allows the user to produce stepwise
regressions for selected variables. when instructed, the user
enters the desired options separated by commas.
possible options to "stepr" are:
"anova"--analysis of variance
"durwt"--durbin-watson test for autocorrelation
"f-val"--indicate f values for entering or omitting a
variable
"force"--indicate variables to be forced into regression
"resid"--store residuals
"toler"--indicate a tolerance other than .0001
if no options are desired, type a <cr>.
if the user has specified the "toler" option he will be
asked to enter a tolerance. the normally assumed tolerance is
.0001.
if "f-val" option was specified, the user will be instructed
to supply the f-value for omitting a variable.
the "resid" option, if used, will cause the residuals to be
stored as a variable. when asked under which variable the
residuals should be stored, the user responds with a variable
name (the variable name must already exist) or a variable number.
once the residuals are stored the name of the variable containing
the residuals will automatically be changed to "resid".
the user will now be instructed to list the independent
variables separated by commas for analysis by the stepwise
regression. up to 19 variables may be listed by variable names
(if names have been defined) or variable numbers. ranges of
variables may be indicated by typing the extremes of the ranges
separated by a "-". one or more "*" may be used when listing the
variables to be analyzed. one at a time each variable not yet
specified in the analysis will be substituted for every "*".
those cases where the same variable would be listed twice in the
same analysis will be eliminated, as will be those cases which,
except for a change in the order of the variables, duplicate an
statpack v4 PAGE 97
analysis already performed.
the user will now be asked to enter the dependent variable,
either the variable name (if names have been defined) or the
variable number may be used. if the user wishes to substitute in
one at a time all those variables not specified as independent,
he may respond with a "*". the "*" may be used both as the
dependent variable and as one or more independent variables in
the same analysis.
if the user specified the "force" option, he will be
instructed to enter those variables, separated by commas to be
forced into the stepwise regression. variables to be forced into
the regression may be entered by either variable names (if names
have been defined) or variable numbers. only variables specified
as independent may be forced into the stepwise regression.
example:
which command? stepr
list the options you wish separated by commas
list the independent variables?
sex,age,height,weight,gpa
which is the dependent variable? iq
***** stepwise regression *****
6 variables; variable: iq is dependent
standard error of y = 17.45335
step no. 1
entering variable: weigh
f-level 2.676 with prob. 0.1024
standard error of estimate 17.43
coefficient of determination = 0.4648738e-02
coefficient of multiple regression = 0.6818165e-01
increase in coefficient of determination = 0.4648738e-02
constant 95.825
variable coefficient std error of coeff
weigh 0.03592 0.02196
statpack v4 PAGE 98
step no. 2
entering variable: age
f-level 1.060 with prob. 0.3037
standard error on estimate 17.43
coefficient of determination = 0.5489679e-02
coefficient of multiple regression = 0.8055854e-01
increase in coefficient of determination = 0.1840942e-02
constant 102.00
variable coefficient std error of coef
age -0.26949 0.26177
weigh 0.03545 0.02196
step no. 3
entering variable: heigh
f-level 0.8806 with prob. 0.3484
standard error of estimate 17.43
coefficient of determination = 0.8019567e-02
coefficient of multiple regression = 0.8955203e-01
increase in coefficient of determination = 0.1529887e-02
constant 122.74
variable coefficient std error of coef
age -0.29100 0.26280
heigh -0.38488 0.41014
weigh 0.06612 0.03937
step no. 4
entering variable: gpa
f-level 0.1101 with prob. 0.7402
standard error of estimate 17.44
coefficient of determination = 0.8211069e-02
coefficient of multiple regression = 0.9061495e-01
increase in coefficient of determination = 0.1915023e-03
constant 121.44
variable coefficient std error of coef
age -0.29444 0.26320
heigh -0.38449 0.41046
weigh 0.06558 0.03944
gpa 0.48003 1.44696
statpack v4 PAGE 99
step no. 5
entering variable: sex
f level 0.4579661e-01 with prob. 0.8306
standard error of estimate = 17.45704
coefficient of determination = 0.8290887e-02
coefficient of multiple regression = 0.9105431e-01
increase in coefficient of determination = 0.7981807e-04
constant 121.2487
std.err. standardized
var. coeff. of coeff. coefficient t-value prob.
sex -0.3844 1.796 -0.1102009e-01 -0.2140 0.831
age -0.2917 0.2637 -0.4645334e-01 -1.106 0.269
heigh -0.3875 0.4110 -0.7074679e-01 -0.9427 0.346
weigh 0.6921e-01 0.4296e-01 0.1313598 1.611 0.108
gpa 0.4800 1.448 0.1388192e-01 0.3315 0.740
statpack v4 PAGE 100
command: stop
--------------
purpose: restart stat pack, alter data size
description: the "stop" command allows the user to restart a run of
stat pack. once the command is given, no additional responses
are necessary.
all files except those created with a "store" command will
be destroyed. output which has been assigned to the line printer
but has not been printed, will remain unchanged, and any further
output assigned to the printer will be added. when the "fini"
command is given all output to that point will be printed. data
which has been entered for analysis will have to be re-entered.
the user will be given the opportunity to restate the limits
for his data when the "stop" command is used. it is possible to
increase, decrease, or leave unchanged the assumed data set;
however, the size is still subject to the original limitations.
example:
which command? stop
maximum number of observations?
statpack v4 PAGE 101
command: store
---------------
purpose: store data on disk
limitations: data stored as floating point numbers, with no option
available to alter output format.
description: the "store" command allows the user to store selected
variables on the disk under his own area. when the file name is
requested, the user types in up to a six character name, and up
to a three character extension separated by a period. next the
user will be asked for the variables to be stored. to indicate
the variables, list either the variable names (if names have been
defined) or the variable numbers on one line separated by commas.
ranges of variables may be indicated by typing the extremes of
the range separated by a "-". if all the variables are to be
stored, a "*" may be used. a standard protection code of 177
will be assigned to the file unless otherwise specified by the
user. data will be stored according to the format (8g15.7).
example:
which command? store
what is the name of the file? song.dat
which variables? 1,3-5
selected variables were stored according to format:
(8g15.7)
what protection would you like? 177
statpack v4 PAGE 102
command: title
---------------
purpose: label output
limitation: the title is limited to one line of 72 characters.
description: the "title" command allows the user to label output with
a line of identification. when instructed, the user types in a
label of up to 72 characters. once the "title" command has been
issued, the output from each command will be labeled. only
another "title" command can modify a header already entered.
example:
which command? title
type in the line of identification
data used is random and not meant for conclusions
which command? desc
data used is random and not meant for conclusions
there are 1 variables and 14 observations
var. means std.dev. variance
1 4.500000 1.911504 3.653846
statpack v4 PAGE 103
command: trans
---------------
purpose: create or modify variables by combining or transforming
existing variables.
description: instructions for transforming variables are structured
in the same manner as a fortran arithmetic statement. the basic
form of each instruction is: the variable to be modified or
created, a "=", and the expression to be evaluated. for each
observation in the data set, the expression to the right of the
"=" is evaluated and its final value is placed in the variable to
the left of the "=". the rules governing the evaluation of the
expression are the same as for fortran. the order in which
operations are executed (hierarchy) is as follows:
order sign explanation of sign
----- ---- -------------------
1 ** exponent
2 * multiply
2 / divide
3 + add
3 - subtract
operations which have the same order of execution are
evaluated as they are encountered proceeding from left to right
through the expression. parentheses inside an expression are
evaluated first, beginning with the innermost. several
predefined functions are available for use in the
transformations. they are:
"abs" - absolute value
"arcsn"- arc sin
"arctn"- arc tangent
"cos" - cosine
"exp" - exponential (e to the x)
"ln" - natural log
"log10"- log base 10
"mean" - mean of variable
"norm" - normal random number generator
"ran" - random number generator (.0-1.)
"sin" - sin
"sqrt" - square root
"std" - standard deviation of variable
functions are evaluated on an equal priority within
parentheses. they are used by typing the abbreviation, and then
the number or expression to be evaluated, enclosed in
parentheses.
statpack v4 PAGE 104
the "trans" command also allows a form of conditional
statement, for which four comparisons to zero are available.
these are:
"ifl" - if less than zero
"ife" - if equal to zero
"ifn" - if not equal to zero
"ifg" - if greater than zero
conditional statements are written in three parts: a
condition to be satisfied, indicated by the three-character code;
an expression enclosed in parentheses to be evaluated and
compared against zero; and the transformation to be executed if
the conditional is satisfied. for each observation, the
expression enclosed in parentheses is evaluated; if it has the
relationship to zero indicated by the conditional code, the
transformation is done for that observation. if the conditional
code is not satisfied, no action is taken and the next
observation is considered.
when issuing instructions to "trans", variable names may be
used or the variable numbers preceded by a "#".
examples:
? total=pts1+pts2+pts3
the variable: total is created or
modified by adding the variables:
pts1, pts2, and pts3 together and
placing the final value in total.
? avgpt=(pts1+pts2+pts3)/3.
the variable: avgpt is created or
modified by adding the variables:
pts1, pts2, pts3 together, dividing
the sum by 3 and placing the final
answer in avgpt.
? iq=mtage/pyage
the variable: iq is modified or
created by dividing the variable:
mtage by the variable: pyage and
placing the answer in iq.
? z=(weigh-mean(weigh))/std(weigh)
the variable: z is created or
transformed by subtracting the mean
of variable: weigh from the
variable: weigh and dividing that
value by the standard deviation of
the variable: weigh.
statpack v4 PAGE 105
? log3=ln(#3)
create or modify the variable: log3
by taking the natural log of
variable number: 3.
? exp=3.14*sex+2.2*weigh+22.
create or modify the variable: exp
by placing the value of 3.14 times
the variable: sex plus 2.2 times
the variable: weigh plus 22 into
the variable: exp.
? ifl(age-24) group=1
create or modify the variable:
group by checking the variable:
age. if the value of variable age
minus 24 is less than 0, put a 1 in
variable: group; otherwise leave
group unchanged.
statpack v4 PAGE 106
command: ttest
---------------
purpose: calculate t tests (significant difference between means)
reference: "statistical methods", snedecor and cochran, pages 104-106
description: the "ttest" command allows the user to calculate t tests
(significant difference between means). when instructed, the
user lists the options he desires separated by commas. possible
options are:
"headr"--suppress initial portion of report, that part containing
means and standard deviations for each sample.
"probs"--output probability (two-tailed) associated with t tests.
"break"--select samples from one variable based on the value of a
second variable. for each observation, the value of the
second variable (breakdown variable) will be used to
determine in which sample the value of variable being
analyzed belongs. this is accomplished by determining which
of a series of ranges the value of the breakdown variable
fits into, and then moving the value of analysis variable to
the corresponding sample. (if this option is not used,
ttest will be calculated between variables.)
note: the following options are used only if "break" has been
used.
"discr"--automatic breakdown. instead of the user entering
ranges, a separate range will be created automatically for
each value in the breakdown variable.
"auto"--automatic breakdown. this option is the same as the
"discr" option. do not enter "auto" with the other options,
it should be entered only when asked to enter the ranges.
note: the following option is only available if automatic
breakdowns are to be used.
"range"--list the ranges to be used for the automatic breakdown.
if the "break" option has not been specified, t tests will
be calculated between all possible pairs of variables. no other
user responses will be necessary in this case.
if the "break" option has been used, it will be necessary
for the user to supply the following information:
statpack v4 PAGE 107
(1) the variables for which t tests are to be calculated (up to
20). the samples for each set of t tests will be selected
from a single variable. variables may be listed using
either variable names (if names have been defined) or
variable numbers. ranges of variables may be specified by
listing the extremes of the range separated by a "-". where
t tests are to be calculated for all variables, a "*" may be
used instead of variable names or numbers.
(2) the variable to be used for the breakdowns. only one
variable may be used, it may be identified by either the
variable name (if name has been defined) or variable number.
all of the variables listed for analysis will be processed
using the same breakdown variable.
(3) ranges for the breakdown variable. if the "discr" option
has been used, this information will be automatically
calculated, and need not be supplied by the user. if the
"discr" option has not been used, the user may still request
the ranges to be automatically calculated by responding with
"auto". to enter ranges, the user types the extremes of the
range, smallest first, separated by a comma. only one range
may be entered per line. up to 50 ranges may be entered.
after the last range has been entered, the user types a
^z(control z).
examples:
which command? ttest
enter options separated by commas
***** t tests *****
analysis run with each variable being used as a treatment
var. size mean std. dev.
iq 559 110.1 14.61
test1 559 41.75 21.74
test2 559 49.68 18.73
test3 559 47.88 12.28
test4 559 64.81 19.70
iq .0000
test1 -61.70 .0000
test2 -60.13 6.537 .0000
test3 -77.09 5,812 -1.896 .0000
test4 -43.66 18.59 13.16 17.24 .0000
iq test1 test2 test3 test4
which command? ttest
statpack v4 PAGE 108
enter options separated by commas
break,discr,range
on what variables are the t-tests to be run? iq,height
what is the variable to be used for the breakdown? sex
ranges for breakdown variable: sex
.0000 , .0000
1.000 , 1.000
***** t tests *****
analysis of variable: iq with treatments determined
by a breakdown on variable: sex
var. size mean std. dev.
1 281 100.4 17.53
2 294 101.5 17.39
1 .0000
2 .7618 .0000
1 2
***** t tests *****
analysis on variable: heigh with treatments determined
by a breakdown on variable: sex
var. size mean std. dev.
1 281 62.45 2.275
2 294 65.42 3.259
1 .0000
2 12.62 .0000
1 2
statpack v4 PAGE 109
command: type
--------------
purpose: type selected variables on the terminal.
description: the "type" command allows the user to display data on
the terminal. when asked for the variables, the user responds by
listing the desired variables on a line separated by commas.
variables may be entered using either variable names (if names
have been defined) or variable numbers. ranges of variables may
be specified by listing the extremes of the range separated by a
"-". where all variables are to be printed, use a "*" instead of
variable names or numbers. the "print" command is also available
in cases where the output to the terminal would be excessive.
example:
which command? type
which variable? sex,age,3,5-6
var
obs sex age 3 5 6
1 1.000 23.00 4.000 9.000 7.000
2 1.000 21.00 6.000 8.000 9.000
3 1.000 25.00 7.000 9.000 6.000
4 2.000 18.00 6.000 5.000 3.000
5 2.000 18.00 5.000 6.000 7.000
6 2.000 19.00 7.000 9.000 6.000
7 1.000 16.00 6.000 7.000 5.000
8 2.000 24.00 5.000 7.000 8.000
9 2.000 21.00 4.000 7.000 8.000
10 2.000 20.00 3.000 4.000 2.000
statpack v4 PAGE 110
command: wilcx
---------------
purpose: use the wilcoxon matched-pairs signed-rank test to calculate
mean, standard deviation, sample size, z-score, and t.
reference: "basic statistical methods", downe and heath, pages 209,
210. "non-parametric statistics", siegel, pages 75-81.
description: the "wilcx" command allows the user to use the wilcoxon
matched-pairs signed-rank test to calculate the associated mean,
standard deviation, sample size, z-score, and t for all possible
pairs of variables. after the command has been given, no
additional user responses are necessary.
example:
which command? wilcx
***** wilcoxon matched-pairs signed-rank test *****
var. vs var. mean s.d. n z t
iq test1 78260.00 3820.40 559 -20.48 0.00
iq test2 78260.00 3820.40 559 -20.48 0.00
iq test3 78260.00 3820.40 559 -20.48 0.00
iq test4 78260.00 3820.40 559 -20.48 0.00
test1 test2 77701.50 3799.94 557 -6.48 53095.00
test1 test3 78260.00 3820.40 559 -10.93 36487.00
test1 test4 76590.50 3759.12 553 -17.75 9882.00
test2 test3 78260.00 3820.40 559 -2.35 69267.00
test2 test4 78260.00 3820.40 559 -11.97 32546.00
test3 test4 78260.00 3820.40 559 -16.26 16128.50
statpack v4 PAGE 111
command: xtab
--------------
purpose: produce cross tabs (output is in the form of ordered pairs)
description: the "xtab" command allows the user to tabulate one or
more cross tabs. the user will first be asked if he desires
percentages, to which the response must be a "yes" or "no".
cross tabs may be produced for the following:
(1) a single variable vs a single variable
(2) a single variable vs all other variables
(3) all variables vs all variables
when instructed to enter the cross tabs, the user lists up
to 20 cross tabs separated by semicolons. each cross tab is
composed of two variables separated by a comma. to indicate the
variables, either variable names (if names have been defined) or
variable numbers may be used. where all variables are to be
used, a "*" may be substituted for the variable names or numbers.
the results will be adjusted in size for output to terminal or
line printer. if a tabled version of the cross tab is desired
see the command "xtab*".
note: positive and negative numbers as well as multiple digit numbers
may be processed with this command.
examples:
which command? xtab
do you also want percentages (yes or no)? yes
list the variables you wish to have cross tabs on
each cross tab separated by a semi-colon and the variables of
each cross tab separated by a comma
sex,age
statpack v4 PAGE 112
cross tab variable: sex vs variable: age
var sex var age freq. percent
-----------------------------------
.0000 18.00 27 4.7%
.0000 19.00 28 4.9%
.0000 20.00 23 4.0%
.0000 21.00 36 6.3%
.0000 22.00 30 5.2%
.0000 23.00 22 3.8%
.0000 24.00 26 4.5%
.0000 25.00 30 5.2%
.0000 26.00 25 4.3%
.0000 27.00 34 5.9%
.1000 18.00 18 3.1%
.1000 19.00 21 3.7%
.1000 20.00 35 6.1%
.1000 21.00 28 4.9%
.1000 22.00 36 6.3%
.1000 23.00 30 5.2%
.1000 24.00 38 6.6%
.1000 25.00 34 5.9%
.1000 26.00 29 5.0%
.1000 27.00 25 4.3%
which command? xtab
do you also want percentages (yes or no)? no
list the variable you wish to have cross tabs on
each cross tab separated by a semi-colon and the variables of
each cross tab separated by a comma
sex,age
cross tab variable: sex vs variable: age
var sex var age freq. var sex var age freq.
-------------------------- --------------------------
.0000 18.00 27 .0000 19.00 28
.0000 20.00 23 .0000 21.00 36
.0000 22.00 30 .0000 23.00 22
.0000 24.00 26 .0000 25.00 25
.0000 26.00 25 .0000 27.00 34
1.000 18.00 18 1.000 19.00 21
1.000 20.00 35 1.000 21.00 28
1.000 22.00 36 1.000 23.00 30
1.000 24.00 38 1.000 25.00 25
1.000 26.00 29 1.000 27.00 25
statpack v4 PAGE 113
command: xtab*
---------------
purpose: produce cross tabs (output is a cross tab table)
limitation: output must be assigned to line printer (see command
"assign").
description: the "xtab*" command allows the user to tabulate one or
more cross tabs. the user will first be asked if he desires
percentages,to which the response must be a "yes" or "no". cross
tabs may be produced for the following:
(1) a single variable vs a single variable
(2) a single variable vs all other variables
(3) all variables vs all variables
when instructed to enter the cross tabs, the user lists up
to 20 cross tabs separated by semicolons. each cross tab is
composed of two variables separated by a comma. to indicate the
variables, either variables names (if names have been defined) or
variable numbers may be used. where all variables are to be
used, a "*" may be substituted for variable names or numbers.
results will be in a tabled form only if output has been
assigned to the line printer. if the "assign" command has not
been used, the output will be in the form of ordered pairs.
note: positive and negative numbers as well as multiple digit numbers
may be processed with "xtab*".
statpack v4 PAGE 114
example:
which command? xtab*
do you also want percentages (yes or no)? yes
list the variables you wish to have cross tabs on
each cross tab separated by a semi-colon and the variables of
each cross tab separated by a comma
age,sex
cross tab variable: age vs variable: sex
variable: age variable: sex
value 0.000 1.00
--------------------
i
18.0 i 27 18
i 4.70% 3.13%
i
19.0 i 28 21
i 4.87% 3.65%
i
20.0 i 23 35
i 4.00% 6.09%
i
21.0 i 36 28
i 6.26% 4.87%
i
22.0 i 30 36
i 5.22% 6.26%
i
23.0 i 22 30
i 3.83% 5.22%
i
24.0 i 26 38
i 4.52% 6.61
i
25.0 i 30 34
i 5.22% 5.91%
i
26.0 i 25 29
i 4.35% 5.04%
i
27.0 i 34 25
i 5.91% 4.35%
statpack v4 PAGE 115
command: zscor
---------------
purpose: calculate z-scores.
reference: "basic statistical methods", downe and heath, pages 60-61.
description: the "zscor" command allows the user to calculate
z-scores for one or more variables. when instructed, the user
lists the variables separated by commas for which z-scores are to
be calculated. up to 40 variables may be listed using either
variable names (if names have been defined) or variable numbers.
ranges of variables may also be indicated by typing the extremes
of the range separated by a "-". where z-scores are to be
calculated for all variables, a "*" may be substituted for the
variable names and numbers. frequencies will be included in the
output for the user's convenience.
example:
which command? zscor
which variable: height
***** z scores for variable: heigh *****
value frequency z-score
58.00000 2 -1.873627
59.00000 45 -1.559810
60.00000 41 -1.245992
61.00000 57 -0.9321746
62.00000 50 -0.6183570
63.00000 70 -0.3045394
64.00000 69 0.9278121e-02
65.00000 63 0.3230957
66.00000 72 0.6369133
67.00000 31 0.9507308
68.00000 19 1.264548
69.00000 21 1.578366
70.00000 15 1.892184
71.00000 8 2.206001
72.00000 10 2.519819
73.00000 2 2.833636
statpack v4 PAGE 116
sample run #1
-------------
.r stp
stat pack v4
western michigan university
data limits are 100 observations and 7 variables
do you wish to change these? (yes or no) no
for a brief program description type "info"
which command? fetch
what is the file name and extension? test.dat
how many input variables? 6
which command? desc
there are 6 variables and 21 observations
var. means std. dev. variance
1 26.57143 9.075084 82.35714
2 30.66667 8.302610 68.93333
3 30.85714 9.408962 88.52857
4 28.00000 8.191459 67.10000
5 27.52381 7.352680 54.06191
6 143.6190 24.29913 590.4476
which command? data
how many input variables? 3
statpack v4 PAGE 117
enter input data
1,2,3
2,3,4
5,4,3
2,4,2
3,4,5
6,5,4
7,6,5
3,5,4
2,4,3
1,3,2
1,4,2
2,5,3
2,5,1
5,7,4
3,4,5
7,6,5
2,3,4
7,6,5
1,2,3
5,4,3
^z
which command? desc
there are 3 variables and 20 observations
var. means std. dev. variance
1 3.350000 2.158825 4.660526
2 4.300000 1.341641 1.800000
3 3.500000 1.192079 1.421053
which command? fini
cpu time: 1.50 elapsed time: 3:0.55
no execution errors detected
exit
.
statpack v4 PAGE 118
sample run #2
-------------
.r stp
stat pack v4
western michigan university
data limits are 100 observations and 7 variables.
do you wish to change these? (yes or no) yes
maximum number of observ.? 600
maximum number of variables? 10
for a brief description description type "info"
which command? acbnk
what bank name and switches? examp
list bank codes separated by commas
1,2,3,4,5,6
which command? estat
there are 6 variables and 575 observations
var. means std.dev. variance
sex 0.5113043 0.5003074 0.2503075
age 22.65391 2.779382 7.724963
heigh 63.97043 3.186565 10.15419
weigh 142.2348 33.12739 1097.424
iq 100.9339 17.45335 304.6193
gpa 2.988052 0.5047344 0.2547569
var. median mode maximum minimum
sex 1.000000 1.000000 1.000000 0.0000000
age 23.00000 22.00000 27.00000 18.00000
heigh 64.00000 66.00000 73.00000 58.00000
weigh 139.0000 112.0000 251.0000 86.00000
iq 101.0000 113.0000 129.0000 70.00000
gpa 3.050000 2.910000 3.790000 1.850000
var. std err of mean skewness coef. of var.
sex 0.2088243e-01 -0.4522899e-01 97.84924
age 0.1160092 -0.3795715e-01 12.26888
heigh 0.1330046 0.4152664 4.981309
weigh 1.382710 0.5287957 23.29064
iq 0.7284885 -0.9540812e-01 17.29185
statpack v4 PAGE 119
gpa 0.2106721e-01 -0.2985082 16.89175
which command? corr
***** correlation matrix *****
var.
sex 1.0000
age 0.0298 1.0000
heigh 0.4663 -0.0661 1.0000
weigh 0.5833 -0.0209 0.8293 1.0000
iq 0.0318 -0.0443 0.0369 0.0682 1.0000
gpa 0.0417 0.0383 0.0533 0.0685 0.0169 1.0000
sex age heigh weigh iq gpa
which command? ttest
enter options separated by commas
break,discr
on what variables are the t-tests to be run? gpa
what is the variable to be used for the breakdown? age
***** t tests *****
analysis on variable: gpa with treatments determined
by a breakdown on variable: age
var. size mean std. dev.
1 45 3.109 0.5089
2 49 2.942 0.4404
3 58 2.971 0.4685
4 64 2.946 0.5188
5 66 2.888 0.5115
6 52 2.902 0.5304
7 64 3.002 0.5373
8 64 3.017 0.5333
9 54 3.096 0.4939
10 59 3.039 0.4715
statpack v4 PAGE 120
1 .0000
2 -1.699 .0000
3 -1.424 .3255 .0000
4 -1.623 .4168e-01 -.2776 .0000
5 -2.239 -.5991 -.9411 -.6459 .0000
6 -1.947 -.4103 -.7216 -.4476 .1512
.0000
7 -1.041 .6344 .3397 .6008 1.244
1.001 .0000
8 -.8984 .7971 .5055 .7644 1.413
1.157 .1585 .0000
9 -.1297 1.656 1.369 1.594 2.249
1.942 .9756 .8224 .0000
10 -.7233 1.092 .7799 1.034 1.710
1.435 .4004 .2374 -.6247 .0000
1 2 3 4 5
6 7 8 9 10
which command? freq
do you also want percentages (yes or no)? no
which variables? age
var. frequency
----- --------------------------------------------------------
age value 18.0 19.0 20.0 21.0 22.0
freq 45 49 58 64 66
value 23.0 24.0 25.0 26.0 27.0
freq 52 64 64 54 59
which command? pcent
which variables? weight,height
type in percentiles you wish to have, separated by commas
dec
statpack v4 PAGE 121
***** percentiles *****
variables
percentile weigh heigh
---------- ----------------------
10.00 102.0 60.00
20.00 112.0 61.00
30.00 120.0 62.00
40.00 131.0 63.00
50.00 139.0 64.00
60.00 148.0 65.00
70.00 157.0 66.00
80.00 170.0 66.00
90.00 188.0 68.00
which command? plot
which is the horizontal variable? weight
which is the vertical variable? height
plot of variable: weigh (horiz.) vs variable: heigh (vert.)
i----------+---------+---------+---------+---------+
73.00 + 1 1
i 221 12 1 1
i
i 1 1 11 11 1 1
69.80 + 1 3 11111111 21
i 1212 121131311 1
i 1 1 1 22 231 1 11111
i
66.60 + 1 1122132521 5 1 21 1
i 1 1215541259545544231
i 1311131268733128513 111
i 12 2554673445575121
63.40 +
i 15616625547834331
i 3 226331545443 131
i241162185526334 211
60.20 + 445213842422
i
i137674663 1 1
i 1 1
57.00 +
i---------+---------+---------+---------+---------+
82.00 162.0 242.0
122.0 202.0 282.0
statpack v4 PAGE 122
which command? fini
cpu time: 12.65 elapsed time: 19:2.27
no execution errors detected
exit
statpack v4 PAGE 123
sample run #3
-------------
.r stp
stat pack v4
western michigan university
data limits are 100 observations and 7 variables.
do you wish to change these? (yes or no) no
for a brief program description type "info"
which command? data
how many input variables? 5
enter input data
12,34,42,32,23
43,43,44,45,43
32,43,54,34,23
34,27,21,28,24
41,23,25,38,32
24,35,34,32,15
24,35,21,28,37
26,28,21,23,34
12,32,41,32,23
23,45,32,12,34
43,25,37,28,21
24,26,27,21,27
23,25,23,21,24
27,28,29,26,25
23,24,21,32,36
31,41,23,25,26
26,25,26,21,27
34,38,29,21,28
21,32,31,45
23,12,24,21,32
12,23,43,23,12
^z
statpack v4 PAGE 124
which command? manip
?? tv5o19
var. obs value
5 19 .000
?? rv5o19
var. obs value new value
5 19 .000 ?32
??
which command? name
var 1? test1
var 2? test2
var 3? test3
var 4? test4
var 5? test5
which command? trans
?total=test1+test2+test3+test4+test5
variable: total has been created
?
which command? anov1
list options separated by commas
which variables? test1,test2,test3,test4,test5
***** 1-way anova *****
tret. size mean std. dev.
test1 21 26.57 9.075084
test2 21 30.67 8.302610
test3 21 30.86 9.408962
test4 21 28.00 8.191459
test5 21 27.52 7.352680
source sum of sq. d.f. mean sq. f prob
between 313.3711 4 78.34 1.085 0.3680
within 7219.619 100 72.20
total 7532.990 104
statpack v4 PAGE 125
which command? estat
there are 6 variables and 21 observations
var. means std.dev. variance
test1 26.57143 9.075084 82.35714
test2 30.66667 8.302610 68.93333
test3 30.85714 9.408962 88.52857
test4 28.00000 8.191459 67.10000
test5 27.52381 7.352680 54.06191
total 143.6190 24.29913 590.4479
var. median mode maximum minimum
test1 24.00000 23.00000 43.00000 12.00000
test2 28.00000 25.00000 45.00000 12.00000
test3 29.00000 21.00000 54.00000 21.00000
test4 28.00000 21.00000 45.00000 12.00000
test5 27.00000 23.00000 43.00000 12.00000
total 140.0000 125.0000 218.0000 112.0000
var. std err of mean skewness coef. of var.
test1 2.029250 0.2228182 34.15354
test2 1.856520 -0.3362000e-02 27.07373
test3 2.103908 0.8388538 30.49201
test4 1.831666 0.4848990 29.25521
test5 1.644109 -0.5015288e-01 26.71389
total 5.433451 1.428420 16.91916
which command? ttest
enter options separated by commas
probs
***** t tests *****
analysis run with each variable being used as a treatment
var. size mean std. dev.
test1 21 26.57 9.075
test2 21 30.67 8.303
test3 21 30.86 9.409
test4 21 28.00 8.191
test5 21 27.52 7.353
total 21 143.6 24.30
statpack v4 PAGE 126
test1 .0000
1.000p
test2 1.526 .0000
.135p 1.000p
test3 1.502 .6956e-01 .0000
.141p .945p 1.000p
test4 .5355 -1.048 -1.050 .0000
.595p .301p .300p 1.000p
test5 .3737 -1.299 -1.279 -.1982 .0000
.711p .202p .208p .844p 1.000p
total 20.68 20.16 19.83 20.66 20.96
.000p -.000p .000p .000p .000p
.0000
1.000p
test1 test2 test3 test4 test5
total
which command? corrt
***** correlated t *****
test1 0.0000
test2 1.656 0.0000
test3 1.447 0.9122e-01 0.0000
test4 0.6143 -1.148 -1.298 0.0000
test5 0.4537 -1.417 -1.109 -0.2229 0.0000
total 26.21 26.20 24.56 26.98 23.92
0.0000
test1 test2 test3 test4 test5
total
which command? corr
***** correlation matrix *****
var.
test1 1.0000
test2 0.1513 1.0000
test3 -0.0780 0.4218 1.0000
test4 0.2415 0.1669 0.3497 1.0000
test5 0.3287 0.1611 -0.3414 0.2100 1.0000
total 0.5758 0.6665 0.5167 0.6833 0.4190 1.0000
test1 test2 test3 test4 test5 total
statpack v4 PAGE 127
which command? hist
which variables? total
***** histogram for variable: total *****
60.00 +
i
i
i ixxxxxi
i ixxxxxi
45.00 + ixxxxxi
i ixxxxxi
i ixxxxxi
i ixxxxxi
i ixxxxxi
30.00 + ixxxxxi
i ixxxxxi
i ixxxxxixxxxxi
i ixxxxxixxxxxi
i ixxxxxixxxxxi
15.00 + ixxxxxixxxxxixxxxxi
i ixxxxxixxxxxixxxxxi
i ixxxxxixxxxxixxxxxi
i ixxxxxixxxxxixxxxxixxxxxi ixxxxxi
i ixxxxxixxxxxixxxxxixxxxxi ixxxxxi
--+-----+-----+-----+-----+-----+-----+
^ 5 ^ 11 ^ 3 ^ 1 ^ 0 ^ 1 ^
^ ^ ^ ^ ^ ^ ^
^ 132. ^ 172. ^ 212. ^
112. 152. 192. 232.
which command? store
what is the name of the file? test.dat
which variables? test1-total
selected variables were stored according to format:
(8g15.7)
what protection would you like? 155
which command? assign
output assigned to printer
which command? estat
statpack v4 PAGE 128
which command? plot
which is the horizontal variable? *
which is the vertical variable? *
which command? corr
which command? fini
cpu time: 17.12 elapsed time: 17:13.53
no execution errors detected
exit
.
statpack v4 PAGE 129
glossary
--------
binary--numbers which are represented in base 2 notation; when
associated with file structures, those files which are
word-oriented rather than ascii.
bivariate--two variable
brackets--[ ], contains the project-programmer number when reading
files from another area.
connect time--amount of time transpired between the user logging in
and logging out.
control z--special character typed by holding down the control button
and typing a z. the control z is the end-of-file mark for the
terminal.
cpu time--amount of time the central processor (computer) actually
spent doing the analysis.
data--any or all facts, numbers, letters, or symbols which can be
processed or produced by a computer; source information.
data set--in the write-up, the data which is presently being analyzed;
i.e., the data in core.
disk--high-speed auxiliary storage device, on which information is
recorded on the magnetizable surface of a rotating disk.
expression--a string of variables and constant values separated by
algebraic operations. a single term or two or more terms
combined, by algebraic operators in accordance with the defined
rules.
extremes--maximum and minimum
file extension (extension)--that portion of the file reference which
follows the period in the name. maximum of three characters,
usually refers to the type of file.
file name (name)--that portion of the files reference which precedes
the period in the name. maximum of six characters.
format--the arrangement of information on a form or in storage.
standard format for input is assumed to be (20f).
hierarchy--the order in which different operations are executed in an
expression.
input--data to be processed; the process of transferring data from an
external storage to an internal storage.
statpack v4 PAGE 130
listing--output printed on line printer.
magtape--abbreviation for magnetic tape, a large capacity storage
medium obtainable from the computer center.
magtape drive--the electro mechanical unit on which a magnetic tape is
mounted prior to accessing it.
major sort key--the variable to which the highest priority is assigned
when ordering.
minor sort key--the variable to which the least priority is assigned
when ordering.
missing data--data where values were not available.
observation--case.
octal--numbers which are represented in base 8 notation.
ordered pairs--a pair of numbers where the order of occurrence
indicates the variable they were chosen from.
output--results or answers written to an output device (line printer,
terminal, or disk). the process of transferring from an internal
storage to an external media.
protection code--a three-digit octal value which defines who may
access, write on, or read a file.
run--in stp, it refers to the time elapsed between typing of "r stp"
and "fini".
scaling--picking beginning points and increments which will fit into
the space available and are easily examined.
scan--to look at each piece of information in the set. to examine
every entry routinely as the first part of a retrieval scheme.
sort--to sequence or order observations according to a key contained
in each observation.
sort key--the variables in the observation which determine or are used
to determine the order in which the observations occur.
terminal--teletype, crt terminal
user area--that portion of the disk allocated to a particular project-
programmer number.
variable name--user specified names for variables.
statpack v4 PAGE 131
<cr>--carriage return.
statpack v4 PAGE 132
index
1wayr . . . . . . . . . . . . . . . . . . . . . . 80
acbnk . . . . . . . . . . . . . . . . . . . . . . 7
anoc1 . . . . . . . . . . . . . . . . . . . . . . 12
anov1 . . . . . . . . . . . . . . . . . . . . . . 16
anov2 . . . . . . . . . . . . . . . . . . . . . . 20
assign . . . . . . . . . . . . . . . . . . . . . . 26
bargr . . . . . . . . . . . . . . . . . . . . . . 27
basic . . . . . . . . . . . . . . . . . . . . . . 29
chisq . . . . . . . . . . . . . . . . . . . . . . 30
copys . . . . . . . . . . . . . . . . . . . . . . 36
corr . . . . . . . . . . . . . . . . . . . . . . . 37
corrt . . . . . . . . . . . . . . . . . . . . . . 38
cvsmt . . . . . . . . . . . . . . . . . . . . . . 39
data . . . . . . . . . . . . . . . . . . . . . . . 42
deass . . . . . . . . . . . . . . . . . . . . . . 43
desc . . . . . . . . . . . . . . . . . . . . . . . 44
erana . . . . . . . . . . . . . . . . . . . . . . 45
estat . . . . . . . . . . . . . . . . . . . . . . 46
facto . . . . . . . . . . . . . . . . . . . . . . 47
fetch . . . . . . . . . . . . . . . . . . . . . . 52
fini . . . . . . . . . . . . . . . . . . . . . . . 53
form . . . . . . . . . . . . . . . . . . . . . . . 54
freq . . . . . . . . . . . . . . . . . . . . . . . 55
glossary . . . . . . . . . . . . . . . . . . . . . 129
help . . . . . . . . . . . . . . . . . . . . . . . 56
hist . . . . . . . . . . . . . . . . . . . . . . . 57
info . . . . . . . . . . . . . . . . . . . . . . . 59
kendl . . . . . . . . . . . . . . . . . . . . . . 60
kolm . . . . . . . . . . . . . . . . . . . . . . . 61
list of commands . . . . . . . . . . . . . . . . . 4
mabnk . . . . . . . . . . . . . . . . . . . . . . 66
make . . . . . . . . . . . . . . . . . . . . . . . 67
manip . . . . . . . . . . . . . . . . . . . . . . 68
mann . . . . . . . . . . . . . . . . . . . . . . . 74
mta/i . . . . . . . . . . . . . . . . . . . . . . 76
name . . . . . . . . . . . . . . . . . . . . . . . 79
statpack v4 PAGE 133
pcent . . . . . . . . . . . . . . . . . . . . . . 81
pcorr . . . . . . . . . . . . . . . . . . . . . . 83
plot . . . . . . . . . . . . . . . . . . . . . . . 84
print . . . . . . . . . . . . . . . . . . . . . . 86
prob . . . . . . . . . . . . . . . . . . . . . . . 87
program transfer . . . . . . . . . . . . . . . . . 6
ptbis . . . . . . . . . . . . . . . . . . . . . . 90
regr . . . . . . . . . . . . . . . . . . . . . . . 92
sample run #1 . . . . . . . . . . . . . . . . . . 116
sample run #2 . . . . . . . . . . . . . . . . . . 118
sample run #3 . . . . . . . . . . . . . . . . . . 123
sort . . . . . . . . . . . . . . . . . . . . . . . 94
srank . . . . . . . . . . . . . . . . . . . . . . 95
stepr . . . . . . . . . . . . . . . . . . . . . . 96
stop . . . . . . . . . . . . . . . . . . . . . . . 100
store . . . . . . . . . . . . . . . . . . . . . . 101
table of variable-observation combinations . . . . 3
title . . . . . . . . . . . . . . . . . . . . . . 102
trans . . . . . . . . . . . . . . . . . . . . . . 103
ttest . . . . . . . . . . . . . . . . . . . . . . 106
type . . . . . . . . . . . . . . . . . . . . . . . 109
wilcx . . . . . . . . . . . . . . . . . . . . . . 110
xtab . . . . . . . . . . . . . . . . . . . . . . . 111
xtab* . . . . . . . . . . . . . . . . . . . . . . 113
zscor . . . . . . . . . . . . . . . . . . . . . . 115