The monthly meeting of the CODAP Users Group was held September 14, 1995, at the Air Force Occupational Measurement Squadron conference room, Randolph AFB, TX . Ms Elena Weber chaired the meeting in the absence of SqLdr Herlihy, AL/HRMJ (RAAF Exchange), who was unavailable. Those in attendance were:

Elena Weber          AFOMS/OMYX     652-5811
Johnny Weissmuller   Metrica        340-8211
Cindy Luster         AFOMS/OMYO     652-5811
Jimmy Mitchell       IJOA           349-8525
Darryl Hand          Metrica        340-8211

1. The Metrica version of FACEXT was enhanced. The output factors are written with two decimals of accuracy (instead of four), and a blank is included between fields. This reduces file size and increases readability. Also, if the maximum factor value is 100 (as is sometimes the case for PMP data), the extra space is not included. This does not detract from the readability. Further, it provides for a more standard format when similar factors are extracted across multiple studies. In addition, the generated layout was modified to better describe its internal format. This information is contained as an additional heading card, and denotes which field contains the starting character (SC), ending character (EC) and number of characters (NC) , as well as the name and description for each extracted variable. OMS was amenable to these improvements, and will receive a version in the development library as soon as the documentation has been updated.

2. The IMPORTTF problem was diagnosed and corrected. Only the RISC version was in error. The use of EXPORT/IMPORT programs led to a general discussion regarding data transfer from the Unisys. OMS has software for performing electronic transfer to their 930, but they have been experiencing some problems. Metrica uses an attached modem and KERMIT software, but the data requirements for Metrica are much smaller in scope. As a last resource, data can be read from ODDBALL foreign output tapes.

3. The JOBTYP/DIAGRM problem was resolved. OMS analysts had noticed that some job groups selected by JOBTYP were not appearing on the DIAGRM. This occurred because different cutoffs were used. Both runs used PMP cutoffs of 50.0, but the Average Between cutoff was set to 25.0 for DIAGRM, but the default of 20.0 was used in JOBTYP. Therefore, any JOBTYP groups that had an average-between value that was less than 25.0 and more than 20.0, would not appear on the DIAGRM. The standard OMS JOBTYP sample has been modified to use the 25.0 cutoff. In addition, the JOBTYP block header now reports this cutoff value.

4. Additional discussion of JOBTYP usage followed. The OMS default is to use the G option, with no Case Title file. This option produces a graphic listing of the relationships between the primary, secondary, and tertiary groups. This report has been made obsolete by the specification of a Case Title file. The relationships are still displayed, but now can include information that can help the analyst determine which group is better. Metrica and McDonnell-Douglas in their TDS/TIDES work typically display the following variables: Grade, Duty AFSC, Number of Tasks Performed, Number Supervised, Base, Organization, and Job Title. Essentially, this report is a PRTVAR with Primary, Secondary, and Tertiary group information included. It was agreed that an analysts meeting should be set up to show what is available.

5. Mr. Hand recommended an additional statistic to be added to JOBTYP - the number (and percentage) of cases covered by the JOBTYP selections. This would provide an overall estimate of how well the JOBTYP groups describe the career field, and could also be used to evaluate modifications to the various cutoff values. OMS agreed to this change. Dr. Mitchell suggested that a similar modification be made to MODTYP. Since MODTYP is the companion program to JOBTYP, this is an excellent idea.

6. The Metrica version of PRTFAC has been modified to include an L option, which allows the user to include the relative task number (instead of the actual number), when using task masks and/or task suppression. This value is contained in the documented pseudo-factor L0000. OMS expressed no interest in obtaining this update.

7. In regard to the CODAP front-end, it is necessary for Mr. Hand to first learn the Perl programming language, which is the most common technique for processing the output from HTML forms (as CGI scripts). Perl is freely available on the Internet, and exists on the OMS and Metrica RISC machines.

8. Ms. Weber reported a problem with AUDITR. The program detects duplicate cases only if the file is sorted. Since the case ID is not restricted to be numeric, it is prohibitive for AUDITR to check for non-sorted duplicates. For now, a disclaimer will be added to the documentation that says the input must be sorted. The VARSXX program can be run for V0001on the output Case Data file to show if any duplicates got through.

9. Olga Velez reported a problem with IMPORTCD. (Note: After the meeting, this problem was resolved. It turned out that the input file was missing the first two lines. This may have been caused by the technique used to transfer data from the Sperry. The two lines were added back to the file, and the program worked fine.)

10. Elena Weber and Johnny Weissmuller discussed the possibility of standardizing the biographical portion of CODAP surveys, regardless of the type of survey (officer, enlisted, civilian, or combination of these). This creates some wasted space (for example, Civilian pay grade does not apply to officer and enlisted studies), but this cost is outweighed by the benefits of standard data locations.

11. The next meeting of the group is scheduled for 1400 hours, 26 October 1995, at the Metrica, Inc. conference room, 10010 San Pedro Avenue, 4th floor.