New Features

Home Page	Overview
New Features	Downloads & Prices
P-STAT documentation	Technical Support
Contact P-STAT	Unistat for Windows

P-STAT version 2.23 revision 11 is now available for Windows, Unix and Linux.

The major reason for this release was to solve problems with the help file formats on PC's running Vista. The only portable way to support help files from the menus is by using HTML for browser access. The text version continues to work on all supported systems. The HTML version is also supported everywhere.

On PC/Windows the html files are stored in a folder named PSTATHELP directly below the directory where the P-STAT module is installed. It can be accessed from within P-STAT by clicking on "Help" from the menus.
On Linux and Solaris the directory is named "helpfiles". Bring up your browser and locate helpfiles below the P-STAT install directory. Then select "index.html" for a complete list of helpfile entries.
SPREADSHEET was new in release 10. There were some interactions with the page change routines that caused problems when multiple tables were produced. These problems have been fixed.
While the menu system was not designed to do complex PPL, some changes have been made to simplify the process. A problem with subcommands entered from the menu system has been fixed.

P-STAT version 2.23 revision 10

LIBRARIES: the use of libraries of P-STAT system files is going to be discontinued. Libraries were useful in the 1970s and 1980s, but the availability of directories and folders has made them unnecessary, and the supporting commands (SAVE, FIND, etc) are going to be dropped.
MAP: an error message used to occur when neither USE nor MR.GROUPS was used, because there was no specification of which variables were to be mapped. This has been changed; when no variables are selected, ALL of the character variables will be mapped.
RENAME.VARIABLES: this is a new command which uses a LABELS file to rename some or all of the variables in an input file. The new names look like extended labels in the labels file.

For example, a labels file record of var22 'total.income' renames var22 into total.income. Use HELP RENAME.VARIABLES $ for details.
PPL: the TRIM function now has an optional argument that limits the number of characters being trimmed from a given end of a value. For example LRTRIM ( 'XXabcXX', 1, 'X' ) produces 'XabcX', whereas LRTRIM ( 'XXabcXX', 'X' ) produces 'abc'.
SURVEY: a new identifier SPREADSHEET provides enhancements for users who export survey printout to spreadsheets such as EXCEL. The usage is identical to the TABS identifiers. A major difference in the output is that strings such as variable names and value labels do not get folded across multiple lines to fit narrow column widths.

A new subcommand can be used with SPREADSHEET which causes character fields to be enclosed in quotes. It takes the form: "QUOTES ON" or "QUOTES OFF". the default is OFF.

The SURVEY command will now MAP up to 24 characters of a character variable and will report any variable that has data values too large to map. An enhancement to the MAP command will automatically map all character variables if no variables are specified.
SPSS.IN has been changed to check that changes made to make variable names legal in P-STAT do not result in duplicate variable names. If such a situation is encountered an extra period is added to the end of one of the names and a report is added to the printout.
PC users: Vista does not support Windows HLP files. The program WinHlp32.exe was no longer felt to meet MicroSoft standards and was dropped from Vista. The result is that ANY file that has the ".HLP" extension causes an error condition

The main P-STAT help file has used the extension HLP for many years even though it is not written as a windows help file. It has been renamed PSTATHELP.DAT.

The subset of the help file for commands named in the menu system was written as a windows help file and will be unavailable (Vista users only) until it can be rewritten as HTML.

P-STAT 2.23 Revision 9

September 1, 2007

TURF:

A change has been The format of the SIZE.AND.RANK values in the REACH.RESULTS and FREQ.RESULTS files has been changed from 3:59 (or such) to 3_59.

Why ? When these files are moved into EXCEL, as people sometimes do, EXCEL wants to be helpful. It sees 3:1, 3:2, etc and decides it is a TIME value. When 3:60 comes along, it interprets that as 4 hours and displays 0.1666666667, 4 hours being one sixth of a full day.

Using an underscore instead of a colon avoids this.
Support for the ISO 8859-1 standards

There is a new identifier, CHARACTER.SET (or CH.SET), in the PostScript, PostScript.Setup and SURVEY commands. "CH.SET german" has the same character set as the older "UMLAUTE.ANSI". "CH.SET ISO" adds most of the special characters defined in the ISO 8859-1 standards.
TAB delimited output in SURVEY

SURVEY has an expanded print buffer of 5000 characters. An output width of 1500 should be sufficient to produce a tabbed output file of the current maximum width of 91 columns. This limit will probably be expanded in the next release.

P-STAT 2.23 Revision 8

April 1, 2007

TURF:
1. A TURF run can now be done in a stepwise manner. This is indicated by using STEP instead of SIZE.
  
  SIZE, followed by one or more sizes, causes a separate, independent run for each given size. This usage is unchanged.
  
  STEP, followed by two or more sizes, processes the sizes (which must ascend) in a stepwise manner. The first size is done normally, then the items in its best combination for reach (or optionally for frequency) are fed into the second size, and so on.
  
  STEP 8 12, does a size 8 run, and then forces the items from the best size 8 combo into a size 12 run.
  
  STEP 1 TO 9, creates a cascading effect in the specified results file. This takes very little time.
  
  A STEP run is usually much faster than a SIZE run, but the final result will probably not be as good as the full SIZE run.
  
  In general, the larger the initial step, the better. For example, STEP 8 12 should provide a better result than STEP 6 12. However, the larger initial step would make it slower.
2. The REACH.RESULTS file has been revised and extended. Three additional lines have been added; one provides the UNIQUE reach provided by each item in the combination. This is the reach that would be lost if that item were dropped. The others are CUMULATIVE.FREQ and ADDITIONAL.FREQ.
3. A SHOW option has been added to control which results file summary variables are in fact shown. OMIT can also be used.
PPL: MONTH.DAY.YEAR, DAY.MONTH.YEAR, types of functions. These 6 functions now allow a character input argument like '12/25/2006' or '06.12.25' or '12252006'. Non-digits are simply treated as separators between the date values.
CLEANUPS:
1. PUT and TEXTWRITER had erroneously not allowed a quoted string to follow @BEFORE. This is now working properly.
2. POSTSCRIPT.SETUP now allows numeric arguments after the COLOR identifier. POSTSCRIPT.CLOSE now supports longer filenames.
3. On the PC, I/O system error message text has been added to many of P-STAT's error messages when I/O problems occur. Also on the PC, using command line arguments had in some situations prevented use of PSFLAG. This is now fixed.
4. The TOL test in COLLATE had succeeded only when the TOL value was greater than the absolute difference between the final values of the two cases being compared. A greater than or equal test is now being used.
5. The internal workspace used by all P-STAT commands has been considerably increased. For example, the whopper/2 workspace was 2.4 megabytes, it is now 4.0 megabytes.

P-STAT 2.23 Revision 7

December 11, 2006

PPL: IF-OR-AND phrases.

The code for processing IF-OR-AND tests in PPL has been completely re-written. This fixes a problem which could occur when several AND and OR tests were done and the phrases were not parenthesized.
ERROR MESSAGES:

Long error messages (which are framed with EEE) will now report whatever it can about where the run was when the error occurred. For example, if a transfer file is in use, its most recently read record number will be reported.
USING THE F2 KEY ON THE PC:

Use of the F2 key causes a TURF command to immediately end; the results of any already completed sizes are preserved. Using the F2 key is now recognized from within other commands such as MAKE or SORT; these now simply end with an error message.
MAXIMUM SIZE OF A PRINT LINE:

This has been increased from 1200 to 5000 characters. Thus
PRINTER.SETTINGS filename, OW 5000 $
would be supported.

P-STAT version 2.23 revision 6

September 17, 2006

SPSS.OUT and extended variable labels

When there are multiple extended variable labels in P-STAT they are combined into a single label for SPSS. This has been changed so that there is always one AND only one blank between the pieces.
SPSS.OUT and missing values.

When the identifiers M1, M2, and M3 are used to convert P-STAT's missing values into values for SPSS any missing type that was not specified was converted to missing type 3. Thus if M2 and M3 were used and M1 was not provided it was processed as M3. This has been changed and any P-STAT missing value that is not converted is written out as the SPSS system missing.
SPSS.IN: new identifier SYSTEM.MISSING.

The identifier SYSTEM.MISSING requires an argument of 1, 2, or 3 to indicate which of the 3 P-STAT missing data values is to represent the SPSS system missing value.
TURF:

This release completely changes the way the incremental reach and freq scores are calculated. The earlier code was fine in the simple case but unsatisfactory when the reach threshold was other than one.

FIRST PHASE: Find the smallest number of items (taken from the combination now being re-ordered) that gets a a nonzero reach, and determine the best item or group of items at that size. In the simple case, that will be just one item. When the reach threshold is 3 and there are no item or response weights, for example, we will not get a nonzero reach until groups of three are tried. We take the best of all possible groups of three.

SECOND PHASE: We now have one or more items as the beginning items for the reordered combination. If more remain, try each one with the beginning items, and add the item that adds the most to the reach count. Repeat this step until all items are added.

Suppose we are reordering a combination of 6 items, AA, BB, CC, DD, EE and FF. When we started by trying each one by itself and that produced nonzero reaches, we have a single best item and we have a reach score for it. That is simple and obvious to show in the reach results file. For example:
```
     DD    CC    EE    AA    FF    BB
     43    62    71    78    81    82
```
However, suppose nonzero reaches did not begin until groups of three were tried. If BB, DD and EE were the best such group, they are shown in input order, with the reach score for the group under the last one. The symbol --> appears under the initial items in that group. For example:
```
     BB    DD    EE    CC    FF    AA
     -->   -->   37    55    61    64
```
This shows that no reaches occurred until groups of three items were tried, and that BB-DD-EE were the best of all 3-item groups. Since AA, CC and FF remained, BB-DD-EE-AA was tried, then BB-DD-EE-CC, then BB-DD-EE-FF. The output shows that CC added more than the others, bringing the reach for those four items up to 55, and so on.

For complete documentation on the TURF command click on TURF documentation

P-STAT version 2.23 release 5

MAKE:

Two options have been added to help in situations where an input character like a quote is intended to be placed within a quote-delimited string. See OPTION 1 and OPTION 2 in the MAKE help section.
TEXTFILE.IN (aka TABFILE.IN):

An identifier, DECIMAL.COMMA, can be used when the input uses commas rather than periods for decimal separators.
TEXTFILE.OUT (aka TABFILE.OUT):

An identifier, DECIMAL.COMMA, will cause the command to write commas rather than periods for decimal separators. For example, it writes 12,34 instead of 12.34 .
SURVEY:

An identifier, DECIMAL.COMMA, will cause the command to write commas rather than periods for decimal separators. For example, it writes 12,34 instead of 12.34 .
SPSS.IN:

The enhancement in revision 4 that allowed importing longer names and labels had a problem with the labels file that was created. This has been fixed.
PC Menu System

A problem with the menu system which inserted a blank between a pathname and an external output file name has been fixed.

P-STAT version 2.23 release 4

TURF:

The cumulative results were incorrect when size was greater than 20 and no weighting was in use. This was corrected in the June 1, 2006 version.
MAKE:

MAKE reads data in which the values are separated by delimiters (like commas) and creates a P-STAT system file. The input may be entered interactively, or it may be read from an external ascii input file.

The names and types of the variables can be provided in subcommands. They can also be read from a separate DEFINITIONS text file, or taken from an existing P-STAT system file. Identifiers NV and VARS can also be used.

The input file will usually have an end-of-case indicator ending each case; carriage return with line feed is common, as is just a line feed. These are identified by the program.

Non-standard delimiters and end-of-case fields are supported. For example, using END.OF.CASE '~~~~~' would indicate that 5 tildes are being used to end each case.

Values can be missing, and cases can be incomplete. When reading 3 variables using a comma delimiter and a slash end-of-case, abc,,def/ would get a missing second variable, and abc/ would cause variables 2 and 3 to be missing.

A summary file can be requested. It shows the name, type, and the number of missing and good cases for each variable.

The input characters can be changed as they are being read by using CHANGE statements in the command.
MAKE.FIXED

MAKE.FIXED is a new command which substantially replaces the fixed format capabilities of the older BUILD command. It reads fixed format data records from an external ascii file and creates a new P-STAT system file.

There are no delimiters between the values in the input records. Therefore, a given variable must use the same number of characters from the same positions in each of the cases in the input file.

The final variable, however, may be open-ended. This is useful when a survey ends with a variable like OTHER.COMMENTS, which might range from zero characters in some cases to as many as 50,000 characters in other cases.

An open-ended variable requires an end-of-case field like CR-LF (carriage return-line feed) as its delimiter.

The names and types of the variables, and their locations within each case, are either provided as subcommands or are found in a separate DEFINITIONS text file.

The input file will usually have an end-of-case indicator ending each case; carriage return with line feed is common, as is just a line feed. These are identified by the program.

Non-standard end-of-case fields are also supported. For example, using END.OF.CASE '~~~~~' would indicate that 5 tildes are being used to end each case.

A summary file can be requested. It shows the name, type, the starting and ending byte locations in the input record, and the number of missing and good cases for each variable.

The input file may have extra characters after the last complete case. If so, the initial 20 of them are reported.

The input characters can be changed as they are being read by using CHANGE statements in the command.
UNMAKE

UNMAKE is a command which reverses the MAKE command. It replaces the UNBUILD command.

UNMAKE reads a P-STAT system file and writes two ascii files: a COMMAND file and a DATA file. These can be used to re-create the system file on some other computer.

The COMMAND file contains a MAKE command; this has the various options that were used to write the DATA file. It also provides the names and datatypes of the variables.

The COMMAND file can be text-edited; it will usually be necessary to change the name that it has for the data file. The command file can be transferred to in a P-STAT run.

The DATA file contains every value for every case. One option is to place each value within quotes. Another is to use long delimiters to terminate values.
New Date Functions
1. DAY.MONTH.YEAR (integer arg)
2. DAY.YEAR.MONTH (integer arg)
3. MONTH.DAY.YEAR (integer arg)
4. MONTH.YEAR.DAY (integer arg)
5. YEAR.DAY.MONTH (integer arg)
6. YEAR.MONTH.DAY (integer arg)

Version 2.23: previous releases

Release 2 is a minor housekeeping upgrade to release 1. Release 3 has just one change which is in the TURF command. This change improves the reordering of items by their contribution in the reach.results and freq.results files when the reach threshold exceeds one.
TURF:
The items of each group in the REACH.RESULTS file are now placed in the order of their reach contribution. Thus, the first named item had the best standalone reach. Each of the remaining items is paired with the first one in turn; the best becomes the second named item, and so on.

Additional lines may be written below each group. They contain: cumulative reach, separate reach, cumulative percent of sample reached, and separate percent of sample reached. The default is to add just the cumulative reach line.

The FREQ.RESULTS file is treated in a similar manner. PCT.OF.MAX.REACH had been dropped from the REACH.RESULTS and FREQ.RESULTS files; this statistic is now restored. The maximum reach is more fully described in the report.

An OMIT option has been added; this allows the user to supress some (or all) of the numeric values which precede the item names in the RESULTS files.

For full details see: TURF command: identifiers (July 8, 2005)

Examples of TURF input, the TURF report and the REACH.RESULTS output.
MAXIMUM SIZE OF CHARACTER VALUES:

Character values can now be as large as 50,000 characters. The previous maximum was 999 characters. This includes all use of character data in the p-stat system.
DEFAULT SIZE OF CHARACTER VALUES:
The default size of a character value has been increased from 16 to 40. For example

GENERATE ##weekday:c = 'monday' $ GENERATE ##comment:50000 = 'many blanks' $

##weekday will be size 40, ending with 34 blanks.
##comment will be size 50000, ending with 49,989 blanks.
LIST:
A new identifier, BLANK.MISSING, has been added. It causes blanks to appear where one or more dashes would otherwise be used to show a missing value. This may be helpful in printing a REACH.RESULTS file.

Also, consider a size 50,000 value containing just 'x'. If the 'x' is left justified, LIST prints x, as before. If the 'x' is right justified, LIST prints literally (49,999 blanks) x.
EXAMINE 'external.file', eoc crlf, out work1 $

This is a new command that reads an external file and counts how many times each type of ascii byte occurs. It writes a report and, if requested, a P-STAT output file.

The report summarizes what was found. It also gives the number of CR-LF combinations, and displays the first ten and last ten bytes in the input file.

An end-of-case (EOC) field like CR LF can be supplied. If so, the report is extended to show the number of cases in the file, the smallest and largest record lengths, and the number of characters, if any, after the final EOC.

The optional output file has a row for each different ascii character found in the input file. Each row has:

(1) the ascii value of the character (0 to 255).
(2) the character, like A or CR or COMMA.
(3) how many times that character was found in the file.
(4) the location in the file where it was first found.
SURVEY
It is now possible to print percents smaller than .05 rather than having them replaced by the special character "*". Combine the subcommands:

CHARACTER PERCENT '%' '%' '?' and
PLACES PERCENTS n, where n is between 2 and 5.

Download Version 2.23 release 8

P-STAT modules - Updates for many current customers can be downloaded from the web.

Features of earlier releases.

P-STAT Version 2.22 release 1-10 and version 2.21 releases 1-10