PCjs Machines
Home of the original IBM PC emulator for browsers.
PC-SIG Diskette Library (Disk #639)
[PCjs Machine "ibm5170"]
Waiting for machine "ibm5170" to load....
Information about “SUPERSTAT 2 OF 2 (ALSO 638)”
SuperSTat Share is a shareware Statistical Package for analyzing the
results of Market surveys and poles. The statistical analysis is very
complete. The program is operated by choosing items from menus and
filling in forms on the screen. This version allows 25 variables
and 50 data columns with up to nine values to the variable.
This is a two disk set that requires both to function.
System Requirements: 128K, two disk drives, Hercules graphics display,
Epson-compatible printer.
How to Start: After consulting the NOTES638.TXT and SST.DOC for
information, initiate the program by entering SSTMENU at the DOS
prompt.
Suggested Registration: $35.00; $20.00 for students.
File Descriptions:
The First Disk Contains:
SCATTER EXE Scatter plot program
SSTMENU EXE Main menu and system management program
FREQ EXE Frequency distribution analysis program
CROSS EXE Cross tabulate program
SST DEF System configuration file
TEST TTL Analysis Module
TEST DTA Analysis Module
TEST APD Analysis Module
TEST VRN Analysis Module
INT APD Analysis Module
INT DTA Analysis Module
INT TTL Analysis Module
INT VRN Analysis Module
The Second Disk Contains:
SSTMENU EXE The menu and system management program
SST DEF Configuration file
DATAMAN EXE The data management module
FILEMAN EXE The file management module
CODEMAN EXE The code entry module
SST DOC The short form of the SST Manual
FILES639.TXT
----------------------------------------------------------------------------
Disk No 639 SST Version 1.00 (Disk 2 of 2) V1 DS2
----------------------------------------------------------------------------
SuperSTat Share is a two disk, shareware Statistical Package for analyzing
the results of Market surveys and poles. The statistical analysis is very
complete. The program is operated by choosing items from menus and filling in
forms on the screen. This version allows 25 variables and 50 data columns
with up to nine values to the variable.
SSTMENU EXE The menu and system management program
SST DEF Configuration file
DATAMAN EXE The data management module
CODEMAN EXE The code entry module
FILEMAN EXE The file management module
SST DOC The short form of the SST Manual
PC Software Interest Group (PC-SIG)
1030D E. Duane Ave.
Sunnyvale, CA 94086
(408) 730-9291
(c) Copyright 1986
GO.TXT
This disk contains the program: SUPERSTAT Share
To find out about this program: "PRINT" or "TYPE" these files:
SST.DOC The user's Manual
to print the Manual type: PRINT SST.DOC and press ENTER or
type: COPY SST.DOC PRT1 and press ENTER
To print the other documents, use the same commands with the appropriate
file name.
PC Software Interest Group (PC-SIG)
1030D E. Duane Ave.
Sunnyvale, CA 94086
(408) 730-9291
(c) Copyright 1986
NOTES639.TXT
Program name: SST v. 1.00
Author name:(C) 1986 Derby Micro-Computer Services, Inc.
Address: P.O. Box 220566
Charlotte, NC 28222
Telephone Number: (704) 536-7721
Program Classification: Statistics for Market Surveys
Registration Fee: $35.00; students $20.00
A full commercial version is available.
Program Description: This is a two disk set.
Superstat Share is a shareware Statistical Package for analyzing the results
of Market surveys and other questioneers (poles). It allows weighting the
data set, encryption of your data and automating analysis by using special
batch file commands. The statistical analysis is very complete. The program
is operated by choosing items from menus and filling in forms on the screen.
This version allows 25 variables and 50 data columns with up to nine values
to the variable. The commercial version allows 250 variables and 250 data
columns with up to 9 values each. Both versions handle an unlimited number
of records.
This version of SST runs under MS/PC/DOS 2.0 or higher on any IBM PC/XT/AT or
compatible with a color monitor, a monochrome monitor, or a Hercules
monochrome graphics card. It needs a hard disk or two floppies and an EPSON
compatible printer.
There are three analysis modules; FREQ, CROSS, and SCATTER. Graphics is
provided in the form of bar and scatter charts.
There are measures for nominal to nominal association and significance,
ordinal to ordinal association and significance, and nominal to ordinal
association and significance. Cramer's V is provided as well, along with the
degrees of freedom, N, and missing values.
You are also provided row, column, and total percentages and the cells counts
as well as complete information on row and column totals and missing values.
You are provide stats for bi-variate correlation, regression, and one-way
ANOVA, as well as descriptive stats on the individual variables.
SST.DOC
SuperStat Short Documentation
(C) 1986 Derby Micro-Computer Services, Inc.
P.O. Box 220566
Charlotte, NC 28222
(704) 536-7721
ACKNOWLEDGEMENTS
Dr. Gary R. Rassel, University of North Carolina at
Charlotte political science department. His comprehensive
and insightful knowledge of research and statistics have
guided the development of this software from its conception
to its documentation. It is with heartfelt appreciation that
I thank you for all you have done.
Dr. Theodore S. Arrington, University of North Carolina at
Charlotte political science department. My senior research
professor, Dr. Arrington's advice has been consistently
excellent. Without his guidance I am sure I would never have
become proficient with research or stats.
Ray Twery, University of North Carolina at Charlotte
mathmatics department. If the others have inspired me to
write theoretically correct research code, Mr. Twery has made
sure the calculations were correct.
Sherry Hicks, Charlotte Market Research Group. Sherry has
used SuperStat in all its various forms for the past two
years. She has done everything from small focus group
tabulations to massive public opinion survey analysis with
the package. She has supported a growing and deservedly very
busy research business with SuperStat alone when other
packages would have, in comparison to SuperStat's earlier
versions, been easier on her and done a better job. At the
same time she has regularly let me know what the problems
were and how it SHOULD as well as could be improved. For
this I am eternally grateful. I could never have duplicated
the extensive real world testing she has done for me. Due to
her work, the package is easier to use, more straightforward
in approach, and fast enough to meet tight deadlines.
Paula Schneiderman, Charlotte Market Research Group. For
trusting and believing in the product. Paula gave me the
chance to put this package to work and never backed away.
Mike Derby
DISCLAIMERS
Epson is a trademark of Epson America, Inc.
IBM is a trademark of International Business Machines Corp.
Introduction
Welcome to SuperStat Share, the shareware version of
SuperStat 1.00. We hope you will find this program useful.
If you do, we request a $35 contribution, students are asked
to contribute $20 and send a copy of their student ID. This
version of SuperStat doubles as a demo for the larger and
more powerful commercial version. When you register, you are
entitled to a 35$ discount on the purchase of commercial
SuperStat.
The essential difference between the two versions is this:
the shareware version allows 25 variables and 50 data columns
with up to nine values to the variable. The commercial
version allows 250 variables and 250 data columns with up to
9 values each. Both versions handle an unlimited number of
records.
As with all shareware, you are allowed to copy and share the
program (in unmodified form) with anyone. You may not charge
a fee for the software. You should provide the entire
package of eight modules and the documentation.
CONFIGURING YOUR SYSTEM
To start SuperStat, place your Utilities disk in the default
drive, type SSTMENU and press <ENTER> (or return as it is
also known). The menu will allow you to configure your
system and call the other modules of SuperStat.
If you are using a color monitor, you may set colors for use
with all modules. If you are using a monochrome monitor, you
can tell the computer it is a color monitor and use reverse
video or bright white as your foreground color. If you are
using a Hercules monochrome graphics card, set it to DIAG and
use it as if it were a regular monochrome monitor, or set the
system to Hercules mode and all color statements will be
skipped. If you are using a hard disk, you may specify drive
A: as the default DATA drive and use drive C: etc. as your
PROGRAM drive.
If you are using a serial printer you may specify it as
the default printer along with the baud rate and etc.
SuperStat will use any of the five print/com ports you
specify for output.
SuperStat uses EPSON compatible printer commands to compress
print in large cross-tabulations (over 5 x 5) and also to
squeeze 100 x 100 scatter plots (squeezing is vertical
compression). Check your printer manual for the appropriate
printer codes and use SSTMENU to change them if necessary.
If you wish to change the defaults to the "factory setting,"
a menu choice is there to reset the original defaults.
You may use SSTMENU to call the other modules or you may type
the name of the module at the DOS prompt and press <ENTER>.
The names are self-explanatory.
ex. A>SCATTER <ENTER>
This command will load and execute the scatter plot
module.
SYSTEM REQUIREMENTS
You need two disk drives, 256 K main memory, and an ASCII
compatible printer. SuperStat uses the Epson standard
command set for compressing print.
SSTMENU creates a file called SST.DEF. If it is not present
when you run other modules, the "factory" defaults will be in
effect.
COMPATIBLE COMPUTERS
SuperStat has been tested successfully on a wide range of
compatible computers, including some piecemeal systems with
very obscure BIOS origins. However, the TANDY line of
compatible computers produce strange results when the
<NUMLOCK> or function keys are pressed. To avoid these
problems on TANDY computers, load DOS, press the <NUMLOCK>
key, then load SuperStat, and DO NOT press the <NUMLOCK> or
function keys again until you return to DOS.
CODEBOOK ENTRY
Start with your codebook. This way, you can enter the
codebook filename in the data entry module and it will set
the number of columns you need and keep track of your maximum
and minimum values (if you have interval level variables in
your data set).
CODEMAN is the name of the module. It stands for codebook
manager.
FILENAME ENTRY
It first asks you for your codebook filename. Follow
standard file naming conventions (explained in your DOS
manual) but do not enter an extension. Actually, SuperStat
won't let you use a period (.) here. You may use directory
naming conventions (paths) if you need them but remember, the
default drive is always your filename prefix UNLESS you enter
one explicitly. To do this, enter b:\dir\filename. When the
second character of the filename string is a colon (:), the
default drive is over ridden. This is true of all modules.
Now the menu appears. You may now choose to enter a new
codebook file.
NUMBER OF VARIABLES
You are then asked to enter the number of variables. This is
the number of questions if you are analyzing a questionnaire.
Be careful, researchers often number the questions and then
alphabetize the other sub-questions. The sub-questions count
as variables in SuperStat. If you skip a variable, create a
one variable codebook and either merge it to the end of your
larger codebook, or split the larger codebook, merge the
single variable, and finally, merge the latter half of the
original codebook. Merging and splitting is accomplished
with FILEMAN, which will be explained later.
TITLE OF THE VARIABLE
Now you will begin to enter the codebook. You are first
prompted for the title of the variable (up to 32 characters),
it can be any combination of characters except the carat (^),
the ampersand (&), the underline (_), the tilde (~) and the
at sign (@).
INSERT AND DELETE
The carat and the ampersand are reserved as additional delete
and insert keys respectively. This is so because SuperStat
is set up to allow the use of the ten-key pad for ten-key
punch. When <NUMLOCK> is not active you may use the <INS>
and <DEL> keys on the ten-key pad.
CURSOR MOVEMENT
The backspace key will move the cursor from the right to the
left one space at a time but will not erase what is there.
The <TAB> key moves the cursor from the left to the right one
space at a time. When the <NUMLOCK> key is not active you
may use the <RIGHT> and <LEFT> cursor keys on the ten-key
pad.
CAUTION: When you backspace over a character it is no
longer a part of the input string unless otherwise noted.
Every character from the cursor right is ignored. The
character remains on the screen so you can tab back over and
re-include it.
TYPE OF VARIABLE
After you enter the name of the variable you are prompted for
the type of variable, either nominal/(O)rdinal, (I)nterval,
or (M)ultiple response. The capital letters in parentheses
indicate the key you press to specify the type of variable.
INTERVAL LEVEL VARIABLES
If you specify an interval level variable, you are then asked
for the number of data columns this variable will require.
You may use from one to twelve columns for the variable.
SuperStat Share does calculations to the 32nd decimal place
in floating point but if you enter too many large numbers you
CAN cause an overflow.
MULTIPLE RESPONSE VARIABLES
There are two types of multiple response variables (MRV), a
stand alone multiple response variable and a multiple
tabulation of a single value of other variables. A MRV may
have up to nine values and use up to ten data columns with
its required missing value switch. The number of columns
required by a MRV may be computed by adding the number of
values to the variable (1-9) plus one.
If you specify a MRV you will then be prompted for the number
of columns the variable will require. This is the number of
values to the variable plus one.
THE MULTIPLE RESPONSE MISSING VALUE SWITCH
Multiple response variables use the first columns of their
field in the data set as a missing value switch. When the
first column of a MRV is a one, then the case is present. If
the first column contains a zero, the case is not present and
is a missing value. Missing values may be included or not
included in the table calculations depending upon your needs.
However, if the first column is zero, the case will always be
considered a missing value, even if other columns of the
field contain responses.
STAND ALONE MULTIPLE RESPONSE VARIABLES
Stand Alone MRV's are coded one or zero only. A one
indicates a response, a zero represents a non-response.
TAB-OF-OTHER MULTIPLE RESPONSE VARIABLES
Tab-of-Other MRV's have a single column missing value switch
but may tabulate up to nine other variables and will allow
the tabulation of any value of the other variables in the
range of 1-9. As the missing value switch is always in the
first column of the MRV field, and all fields in the data set
must be contiguous, the other variables must immediately
follow the switch.
MULTIPLE RESPONSE VALUE LABELS
After you specify the number of columns, you are then
prompted for the value labels of the variable, in turn, one
at a time. If the MRV stands alone, this is usually some
abbreviation of the response category. If this MRV is a tab
of other variables, the labels are usually some abbreviation
of the other variable's titles.
As you enter the labels, you are told which column each label
is associated with. Notice that the first label is
associated with the second column of the MRV field.
NOMINAL-ORDINAL VARIABLES
Nominal and ordinal (NO) variables are treated the same by
the analysis modules so it saves time and programming to
group them in the type of variable prompt.
If you specify a NO variable you will then be prompted for
the number of values to the variable. Then you will be
prompted to enter the value labels, in turn, one at a time.
If this is an ordinal variable, be sure to enter the codes in
their proper order.
The labels may contain any combination of characters except
the carat and ampersand.
LINE ENTRY
"Any combination of characters" means any ASCII character
with a code between 32 and 126 except the ^, the &, the ~,
the _, and the @. Periods are sometimes dis-allowed, as are
commas, depending upon the error prevention needs of the
program at a given place.
All data entry in SuperStat is line entry. One piece of
information at a time, as it is needed, only if it is needed,
with no backing up. All prompts have error checks to ensure
valid entries, although the program cannot, and does not
check for every possible error, just the most common, or the
only ones possible at a given place.
Full screen editing is nice and we are sure you have a
favorite word processor. Just learn the format of a
SuperStat codebook file and then use an ASCII word processor
to enter and edit the codes. We fully expect experienced
users to eventually by-pass our "hand holding" entry modules.
Go ahead, we've made it easy for you.
FORMAT OF CODEBOOK FILES
Two files are created by CODEMAN, ".TTL" files and ".VRN"
files.
The VRN file contains:
Ex. 5,22,5
| | |
| | -> total variables
| -> first column of next variable
-> next variable number
When the first and third number are equal, the codebook has
been completed.
The TTL file contains the codes. The first line contains the
specifications for the variable to follow.
----------> Type of variable (must be in double quotes)
| ------> Number of first column in data field
| | ----> Number of columns used by this variable
| | | --> Number of labels for the variable
| | | |
| | | |
"NO",1,1,4
Title of the variable
label-1
label-2
label-3
label-4
The format for MRV's is functionally equal.
"MSA",2,5,4 ("MSA means stand alone)
Title of the variable ("MTO" means a Tab-of-Others)
label-1
label-2
label-3
label-4
The format for interval variables is slightly different.
|-> DIFFERENCE = values always equal 2
"I",7,3,2
Title of the variable
Maximum value (always initialized to zero)
Minimum value (always initialized to nines, if three
columns, then 3 nines)
Ex. "I",7,3,2
Age of Respondent
0
999
TAKING A BREAK
If you want to take a break from code entry, just press
<ESC>ape at the title of the variable prompt. This will
close the file and return you to the menu. Information is
always saved after each variable is entered so you don't need
to worry about explicitly saving files.
ADDING TO INCOMPLETE CODEBOOKS
When you come back to your data entry, choose the second menu
item, "add to existing codebook." The routine is the same as
above.
PRINTING A CODEBOOK
You may print your codebook to a disk or printer. It comes
up on screen as you do so. Press <CTRL> and <NUMLOCK>
together to slow the process down. Press any key to resume
printing.
If you print to disk, you will be prompted for an output
filename. Follow the same conventions as always. Remember
your default drive specification is still in effect.
If you print to a printer other than LPT1:, be sure to change
the system defaults with SSTMENU. Any ASCII compatible
printer should work fine.
EDITING A CODEBOOK
Press (4) when you see the CODEMAN menu. First the variables
that have been entered so far are displayed. You are then
prompted for the NUMBER of the VARIABLE you wish to edit.
After you have chosen a variable the screen clears and the
variable title and labels are displayed.
A prompt line allows you to edit the titles, labels, level,
columns, maximum and minimum, save the changes or <ESC> to
the menu.
If you do not change anything, the program will not save the
codebook to disk, even if you tell it to. You must
explicitly save the changes to disk at the prompt.
If you change something and then change your mind, press
<ESC>ape. This will take you back to the menu without saving
the changes. All changes are lost.
DATA ENTRY
Data entry is accomplished with the DATAMAN module which,
you might have guessed, stands for "data manager." You may
call the module from the main menu or you may type the module
name at the DOS prompt.
First you will be asked for a data filename. Use standard
file naming conventions and remember that your default DATA
drive is in effect. Next you are prompted to enter your
codebook filename. The data manager will look at your VRN
file to determine the number of columns required, keep track
of maximum and minimum values of interval level variables and
save the changes (if any) when you leave the input routine.
The menu now appears. Press (1) to begin the entry of a new
data set. If you have not entered the codebook filename you
will first be prompted to specify the number of data columns
required. This is the middle number of the VRN file minus
one.
Next you will see a data entry record line. It specifies the
record number and provides a marked field to ease data entry.
Above it is a column counter and below it lies several help
statements.
Ex.
Column 1
Record #1 ____+____|__
Fill the field completely and press <ENTER>.
Missing Values (blanks) are zero (0) for Nominal/Ordinal
and asterisks (*) for Interval/Ratio variables.
Press <ESC> when finished or to return to menu.
Use (^) for delete and (&) for insert.
INSERT AND DELETE
The carat (^) and the ampersand (&) keys have been changed
respectively to delete and insert to allow full use of the
ten-key pad.
MISSING VALUES
Missing Values (blanks) are zero (0) for Nominal/Ordinal and
asterisks (*) for Interval/Ratio variables. No blank spaces
are allowed. Missing values are always zero for
nominal/ordinal variables. If you forget and enter
asterisks, or just prefer them, don't worry. They are
functionally equivalent to zeroes in SuperStat WHERE IT
MATTERS. Namely, in FREQ and CROSS. In SCATTER, where
interval level variables are processed, zeroes are important
numbers. Therefore, asterisks are considered missing values
and are automatically not included.
DATA FILES
DATAMAN creates two files, the ".DTA" file, which holds your
raw data, and the ".APD" file.
Example APD file:
250,12
| |
| -> number of columns
-----> number of records
Example record in DTA file:
67864236181298367432
SAVING RECORDS
Each record is automatically saved each time you press
<ENTER> and the APD file is updated. This is done to prevent
any loss of data. It will be a bit slow for you if you never
have records over 10 to 15 columns long. But losing data is
no fun and we prefer the extra second it takes to save one
record and move on the next.
Each record is saved in ASCII format so you can transfer it
to different programs, or from your favorite word processor
to SuperStat. Just use a fixed format in the manner of
DATAMAN, and the analysis modules should have no trouble
picking it up.
CAUTION: Some word processors do not allow non-document
processing and inbed spaces in front of your records. Some
word processors imbed characters invisible to you but
apparent to the computer. These problems can cause
unpredictable results.
SuperStat uses fixed field format to save space on the disks
for large data sets. Commas take up a lot of room.
SuperStat writes data to the disk SEQUENTIALLY. That is, one
record at a time, one after the other, from the first to the
last. It reads data the same way. Except when editing, this
is the faster way to manipulate the data.
TAKING A BREAK
Just press <ESC>ape at the beginning of the next line and you
will return to the menu. When you come back to your data
entry, choose menu item (2) to add more data. DATAMAN will
pick up where you left off and work just as it is explained
above.
PRINTING DATA
Choose menu item (3) to print a copy of your data set. You
will print to the default printer.
EDITING DATA
Choose menu item (4) to edit the data. It is best to keep a
list of records with errors so that you can go through the
data set sequentially. You can't back up so if you skip a
record you need to edit, you'll have to begin again. This is
when sequential file manipulation gets slow. You have to
organize your editing to overcome it.
The edit routine will ask you for the record you wish to edit
and then load it. It will put the old record on that
familiar data entry line and position the cursor at column
one. You need only correct the mistake and press enter.
Unlike other line entry in SuperStat, this one routine allows
you to press <ENTER> from anywhere on the line (except column
one) without ignoring all characters from the cursor, right,
to the end of the line prompt.
MANAGING FILES
After you have completed your codebook and data entry, there
may be some manipulations you would like to do to your files
as a whole. The FILEMAN module provides a number of options
for working with your data to whittle it down or build it up.
You may, for instance, wish to have more than one person
enter your data to facilitate processing. FILEMAN will join
the various files together into a single whole for further
processing and analysis.
In general, you are first asked for a first source filename
and if needed, a second source filename. You are then
prompted for a new output filename. It is assumed that you
do NOT want your original data altered (except when
encrypting).
ENCRYPTING DATA
The first menu option allows you to encrypt your data based
on a single letter encryption key in the range of 1-9, a-z,
and A-Z. It is the ASCII code that is operative. Therefore,
you have 62 possible keys. You are also prompted for a
password. The password is also encrypted so that snoops
cannot decipher it.
If you forget either your key or your password, your data is
lost. The analysis modules cannot work with the encrypted
data and we cannot decrypt your data for you. If you keep an
un-encrypted version of the data, you defeat the purpose of
encryption. The source file is destroyed but could probably
be un-deleted with one of the utilities available on the
market. Therefore, if you really wish to hide the data,
encryt the data, then back up the encrypted file, and
finally, reformat the original data disk and any backup data
disks. Before you reformat, be sure any other important
files on the disk are backed up. Consult your DOS disk for
more information on formatting diskettes.
APPENDING DATA FILES
The second menu option allows you to append data files (add
records from one file to another). This is useful when more
than two people are entering data on different machines. The
records must be the same length. The files may have
different numbers of records in them. If you are working
with two large files, read the data from one disk and write
to another. SuperStat will write the data wherever you wish
so if you have three disks, you can read one file from A:,
another from B:, and write to C:. If your data fills both
source disks, you will have no choice in this matter.
Ex. How data files are appended.
123123123123123 1st source data file
/|\
|
987987987987987 2nd source data file
MERGING DATA AND CODEBOOK FILES
The third menu option allows you to merge data and codebook
files (add variables). If you only wish to do one or the
other (presumably because one set of files, either data or
codebook, are not yet entered), a prompt gives you that
option. If you have both sets of files entered, merge them
at the same time. SuperStat does most of the gritty work of
keeping track of specifications for you and prevents
mistakes if it can. Your data files must contain the same
number of records to use this feature.
Ex. How data files are merged.
1st source data record 2nd source data record
123123123123123123 <------ 987987987987987987
Since you are adding variables to the data set, your codebook
(if it exists) must be updated to allow the analysis modules
to use the added data. The second source codebook is merged
to the end of the first source codebook. Don't let this
confuse you. The action of the codebook merge resembles the
action of the data append. This is due to the diference in
the structure of the data and codebook files.
Ex. How codebook files are merged.
"NO",1,1,2
Variable Title 1st source codebook
Label 1
Label 2
/|\
|
|
"NO",2,1,3
Variable title
Label 1
Label 2
Label 3
FILEMAN keeps track of the number of columns you are using
and prevents you from creating a codebook (or data file)
larger than SuperStat can handle and re-numbers the merged
variables to reflect their new position in the new file.
SPLITTING DATA AND CODEBOOK FILES
The fourth menu option allows you to separate variables away
from each other in the data set. It has an action opposite
to that of merging with regard to both data and codebook
files. Your original file is not destroyed, two files are
simply created where there was only one. Again, the program
keeps track of the position of the variables in the new data
set.
PERMANENT RECODES
The fifth menu option allows you to permanently recode your
data. If you will be using a variable in its recoded form on
a regular basis throughout your analysis, it may be
advantageous to recode it permanently and put it, along with
all of the other variables, into a separate data file. This
will save time and keystrokes later. You must recode both
the data file and its codebook for the analysis modules to
operate correctly.
You are prompted for the number of variables you wish to
recode (up to ten at a time), the number of the variable
(each in turn), and for each variable, the number of
nominal/ordinal categories you wish to create. Then you are
prompted for the high and low values of the variable that are
to fit into the new categories and in turn, the value label
for each new value.
You are free to recode any variable except MRV's. You are
free to recode zeroes to nines and fives to threes if that is
what you need to do.
When you complete the recode specifications the new recoded
data set is saved to disk under a the new filename you are
prompted for.
RECODES IN THE ANALYSIS MODULES
The operation of all other recode routines is the same with
one exception. When recoding a variable in the analysis
modules, the recode is transient, that is, it is NOT saved.
PERMANENT SELECTIONS
The sixth menu option allows you to permanently select a
group of records based on one value of up to nine different
variables. A new data file is created which uses the
original codebook. No data is changed and no variables or
records are lost. You simply state your criteria for the new
data set and the records of the source file are searched for
cases that match the criteria. The matching records are then
written to a new file under a filename you are prompted for.
You are allowed to make up to 9 selections. The actual
selections you make are of the values to variables. You may
not select values of Interval or Multiple Response variables.
You may, however, recode an interval level variable and then
select one of the new categories.
You are prompted for the operator on which to base the
selection. The operators are AND, OR, and NOT. If you
choose AND, all of the selections must match the
corresponding values of the case for the case to be selected.
Ex. If selected value 1 = corresponding case value AND
selected value 2 = corresponding case value AND
.
.
selected value 10 = corresponding case value THEN
select the case and write it to the new file.
If you choose OR, at least one of the selections must match
the corresponding value of the case for the case to be
selected.
Ex. If selected value 1 = corresponding case value OR
selected value 2 = corresponding case value OR
.
.
selected value 10 = corresponding case value THEN
select the case and write it to the new file.
If you choose NOT, if any of the selections of
match the corresponding values of the case, the case will not
be selected. The action for NOT is actually AND NOT.
Ex.
If selected value 1 does NOT = corresponding case value AND
selected value 2 does NOT = corresponding case value AND
.
.
selected value 10 does NOT = corresponding case value THEN
select the case and write it to the new file.
SELECTIONS IN THE ANALYSIS MODULES
The operation of all other selection routines is the same with
one exception. When selecting a value of a variable in the
analysis modules, the selection is transient, that is, it is
NOT saved.
WEIGHTING THE DATA SET
The seventh menu option allows you to weight the data set
based on the selection of three weighting criteria. The
selection operator is always AND. In other words:
If selected value 1 = corresponding case value AND
selected value 2 = corresponding case value AND
selected value 3 = corresponding case value THEN
add another identical case to the data set the number times
specified as the weighting factor.
You are prompted for the weighting factor. It can be in the
range of 0.1 to 9.9. As cases are found that match the
weighting criteria, the case is duplicated the number of
times indicated by the integer part of the weighting factor.
The fractional parts are added together each time a match is
found until they are greater than or equal to 1, at which
time 1 is added to the weighting factor, and the fractional
part is reduced by one. The addition of 1 to the weighting
factor is a transient addition. You will not end up with a
weighting factor of 45 through repeated addition.
The weighted data set is written to a new file under a new
filename which you are prompted for.
ANALYSIS
There are three analysis modules supplied with SuperStat
Share; FREQ, CROSS, and SCATTER. You may run the programs
from SSTMENU or you may type the filename at the DOS prompt.
Ex. A>FREQ
FREQUENCY DISTRIBUTIONS
The frequency distribution module is called FREQ. It will
analyze from 1 to 250 variables, display a bar chart, and
provide:
Value counts
Value percentages
Mean Ordinal Rank
Mode
Median
Standard Error at 95% confidence level
Some of you will wonder why there are no interval level
statistics in the module. It is simply because it is set up
to handle CODES, not numbers (although the codes themselves
are numbers). It is basically a nominal and ordinal level
variable analyzer, and based on that fact, higher level
statistics are simply inappropriate. If you insist on them
in spite of this, use SCATTER, it is the real number
analyzer.
First you are prompted for the data filename and then the
codebook filename. The reason you have to enter both is
simple; if you ever repeat a study, you can use your old
codebook. The filenames can be the same because the data
entry modules give them different filename extentions. FREQ,
CROSS, and SCATTER can tell them apart.
After your filenames are entered the menu appears. If you
wish to make selections or recodes, do so first. The
routines are identical to those in FILEMAN except that the
changes you make are not saved.
Next you may choose to run all of the frequencies or just
one. Either way, all of the variables are tabulated, you
just have some options about which variables are to be
printed and where if you choose to print only one.
If you choose to tabulate a single variable, you are first
given a display of your variables to choose from. Press
<ENTER> to skip the display or the <SPACEBAR> to see
subsequent pages. You are then prompted for the NUMBER of
the first variable you wish to see.
Either way, you are then asked for the value you wish to be
used to tabulate multiple responses. If you only have
MSA's, this is always one (1). If you have MTO's, your
choices are in the range of values common to the variables
you specified to be included in each MTO, ie., if the
variables each have 6 values, you may choose to tabulate
value 4.
NOTE: You are only prompted for the multiple response value
if you have one in the data set.
Either way, you are then asked whether you want missing
values included in the calculations. If you do, every
calculation will be affected by the higher N. If you do
not, every calculation will be affected by the lower N. You
are reminded on the output what your choice was.
The variables are now tabulated. As FREQ works, it checks
for out of range values and if it finds any, it warns you on
screen and sets the value to zero (0). If you wish to get a
listing of the error messages, use DOS piping to redirect
output to a disk file and press <ESC> at the next prompt
(consult your DOS manual for more information on DOS piping).
If you chose to tabulate a single variable, the variable you
chose is displayed first. You are then asked where you want
the out put to go; the printer, a disk file, both, or
neither. You may also skip a page from the prompt. The
program then lets you choose again.
If you choose the printer, the output will go to the default
printer. If you choose the disk or both, you will be asked
for an output filename and the file is then written. If you
choose both, it is also printed to the default printer. If
you choose neither, you are then asked which variable you
wish to see next. If none, press <ESC>. Otherwise, type the
number of the variable and press <ENTER>.
If you chose to print all the variables, the first variable
is displayed and then the print prompt described above. Its
operation is the same. You are NOT asked which variable you
wish to next. Each variable is displayed, in turn, as it is
printed. If you wish to see all the variables but don't want
them printed, press (N) for neither and be ready with the
<CTRL><NUMLOCK> combination to slow things down. It works so
fast you will need to stop it to see anything.
PIPING
When you have learned the prompts it will be easier for you
to get your work done if you use piping to redirect input.
Piping fools the modules such that when they look for
keystrokes, DOS gives them characters you have typed in a
file on disk. Using this method you can fully automate the
analysis process. By combining batch files and piped input
files, you can set your machine and it will do the work while
you do something else. The modules have all been set to
allow this to take place and streamline the process. You can
literally run thousands of frequencies, crosstabs, and
scatter plots on any number of data sets WHILE YOU SLEEP.
Just make sure you have the keystrokes correct, and enough
paper and disk space to handle the job.
For purposes of piping, the following keys have functional
equivalents:
<ENTER> = tilde (~)
<SPACEBAR> = underline (_)
<ESC>ape = at sign (@)
CAUTION: ALWAYS end your input file with the appropriate
"return to DOS" character. Otherwise, the "big red button"
will be required to get control of the computer back (also
known as the "on/off" switch).
CROSS TABULATIONS
The crosstabs module is named CROSS. Just as in FREQ, you
are first asked for your data and codebook filenames. Then
you see the menu.
Select and recode works just as in the other modules. You
can recode all the values of one variable into a single value
and then select that value as an independent variable.
You are then shown the menu again and you may then choose to
run a crosstab.
You are first shown a list of all your variables. You may
skip the display by pressing <ENTER> or see other pages by
pressing <SPACEBAR>. You are then asked for and independent
and dependent variable.
Next you must tell the system whether you want missing values
included in the calculations.
Now CROSS runs through the data, does the calculations and
displays a bar chart. You may skip the display of the charts
by pressing <ENTER> or you may see the others by pressing the
<SPACEBAR>. There are as many charts as there are values of
the independent variable.
Next you are asked if you want to print the bar charts to the
line printer, a file on disk, or not at all. Press the
letter in parentheses that corresponds to your choice. If
you choose to print to disk you will be prompted for a
filename.
Next you are asked if you want the table printed to the line
printer, a file on disk, or not at all. Press the letter of
your choice. You will be asked for a filename if you choose
to print to disk.
After printing you will return to the menu.
You are provided stats for bi-variate association,
significance, and chi-square appropriate for nominal and
ordinal level variables.
Goodman and Kruskals Tau is provided in place of Lambda
because it is a more sensitive measure. Where you can SEE
some association and Lambda shows none, G & K's Tau will pick
it up.
There are measures for nominal to nominal association and
significance (G & K's Tau and Chi-Square), ordinal to ordinal
association and significance (Gamma and Z-Score Gamma), and
nominal to ordinal association and significance (Wilcoxons
Theta and contingency coefficient). Cramer's V is provided
as well, along with the degrees of freedom, N, and missing
values.
You are also provided row, column, and total percentages and
the cells counts as well as complete information on row and
column totals and missing values.
SCATTERPLOTS
The scatterplot module is named SCATTER. It is structured to
work very much like FREQ and CROSS. You are asked for your
filenames and then you see a menu. You can select and recode
just as in the other modules.
When you choose to run a scatterplot you are shown your
variables and prompted for the independent and dependent
variables.
Next you choose to consider zero as an absolute minimum or
not. If you choose to do so, the minimum value of both
variables becomes zeroe (0), thus making your interval level
variables ratio variables.
Missing values are automatically not included because they
are not numbers and SCATTER only works with numbers.
Remember that asterisks (*) are missing values in SCATTER.
Next SCATTER runs through the data, does the calculations,
and displays a 20x50 scatterplot on the screen. Press a key
to see the statistics.
After viewing the stats you are asked if you want to print a
30x50 scatterplot or a 100x100 scatterplot to the line
printer, disk, or not at all.
The 30x50 scatterplot is appropriate for daisy wheel printers
or disk files. The 100x100 scatterplot is appropriate for
dot matrix printers compatible with the Epson standard
command set for sqeezing and compressing print. Both formats
come very close to being a square plot on their appropriate
mediums.
You are provide stats for bi-variate correlation, regression,
and one-way ANOVA, as well as descriptive stats on the
individual variables.
REFERENCES
Evans, James D., Invitation to Psychological Research. Holt,
Rinehart, and Winston. New York. 1985
Manheim, Jarol B. and Rich, Richard C. Empirical Political
Analysis, Research Methods in Political Science.
Prentice Hall, Inc., Englewood Cliffs, NJ. 1981.
Watson, George and McGaw, Dickinson. Statistical Inquiry.
John Wiley and Sons. New York. 1980.
McClave, James T. and Dietrich, Frank H., II. Statistics
Dellen Publishing Co. San Francisco. 1982.
Welch, Susan and Comer, John C. Quantitative Methods for
Public Administration, Techniques and Applications. The
Dorsey Press. Homewood, IL. 1983
Babbie, Earl R. The Practice of Social Research. Wadsworth
Publishing Co. Belmont, CA. 1979.
Phillips, John L. Statistical Thinking. W.H. Freeman and
Co. San Francisco. 1973.
Freeman, Linton C. Elementary Applied Statistics. John
Wiley and Sons. New York. 1965.
Jones, E. Terrence. Conducting Political Research. Harper
and Row. New York. 1984. pp. 166 - 167.
Directory of PC-SIG Library Disk #0639
Volume in drive A has no label
Directory of A:\
CODEMAN EXE 65946 8-23-86 5:56p
DATAMAN EXE 52234 8-23-86 3:30p
FILEMAN EXE 73898 8-23-86 3:33p
FILES639 TXT 959 1-04-80 5:01p
GO BAT 28 11-27-86 12:31p
GO TXT 490 1-04-80 5:00p
NOTES639 TXT 2008 1-04-80 5:04p
SST DEF 158 8-23-86 3:14p
SST DOC 49589 12-08-86 11:03a
SSTMENU EXE 52106 8-23-86 3:05p
10 file(s) 297416 bytes
19456 bytes free