Results Processing in GuineaPig System

Results created by the GuineaPig system are in a format different from what the usual statistical sofware packages expect. A small converter program is provided to convert result data into a text file.

The ResPrint tool will convert result files from the GuineaPig system's internal format to a tabulated text file. It will read the session results files given as arguments, scans their contents and prints out a text file to standard output. For each entry (test item), it will print (by default):

Item ID - the test item ID.
Subject ID - the subject ID of the person who gave the answers.
Session ID - the session ID of the session.
Duration of ihem - how much time was spent for this item from starting of this item to the time the subject was done. The time is shows as seconds. For tests that enforce a time limit for answering this field may read TIMEOUT which tells that the subject exceeded the time limit.
Number of switches - the number of times the subject has switched samples during an item (free sequence test) or number of samples played to the subject (fixed sequence tests).
Item parameters - the test item's parameters, usually the names of the samples played (sample IDs of the samples, actually).
Answers - answers to the questions. If an answer for a particular question is missing, it's field will be left blank.

In the beginning of the printout, a comment will be printed that shows which column is which. For result files that contain session information (serialized result files from tests), comments are printed that contain the session ID, starting and ending time of the session and the MCL level of the session.

Running the converter

First go to the test directory where the result files are kept. Then run the converter program and give the file names of the session result files as agruments, for example:

    gpResPrint results_*

if your result file names begin with 'results_'. The converter then scans in the files given as arguments and prints out the results in a tabulated table.

For more details about ResPrint tool, see ResPrint manual page.

Program output

Program output format depends slighty on the test type (different parameters and questions). For example output of the ResPrint program processing a A/B/C-test (actually a Ref/A/B) could look like this:

#item   subject session time    switch  A       B       Ref     qB      qA
item4   ville   zappa   8.7     7       pirr11  pirr16  pirr16  7.0     2.0
item3   ville   zappa   14.2    12      pirr8   pirr11  pirr8   2.0     8.0
item2   ville   zappa   8.5     9       pirr22  pirr32  pirr32  2.6     7.6
item1   ville   zappa   10.0    7       pirr44  pirr32  pirr44  7.8     9.4

Here is another example of a A/B scale test and session information. Also shows an example of automatically generated session and subject IDs if IDs were not set.

#session id: S14837061
#session start time: Wed Mar 18 14:21:55 GMT+02:00 1998
#session end time: Wed Mar 18 14:25:12 GMT+02:00 1998
#session MCLL: -3.999999999999999dB
#item   subject         session         time    switch  A       B       q2      q1
item4   S14837061a      S14837061       130.2   25      pirr11  pirr16  5.3     5.2
item3   S14837061a      S14837061       27.6    13      pirr8   pirr11  6.0     3.0
item2   S14837061a      S14837061       4.9     3       pirr22  pirr32  7.0     3.0
item1   S14837061a      S14837061       24.8    11      pirr44  pirr32  6.0     4.0

Output can be directed to a file or piped to another program with normal unix syntax.

Comment lines start with the '#' character. Each field will be separated with a single ASCII TAB-character. White space other than TAB-characters are not considered field separators.

The output of the tool can be customized with result options file(s) and tools options.

Note!

You might have noticed that the order of the answers (question IDs) in the result file may not be the same as they appear on the test configuration files. Also, the order of the answers (for example: [qA, qB]) in the result file produced from a session1 may be different from the result file produced by another session2 ([qB, qA]). For example:

   gpResPrint session1.ser >res
   gpResPrint session2.ser >>res

The order of answers may be different in session1's result from session2's results. This is due to internals of the system where results are stored in an item as a hash table which is not ordered.

The problem occurs only when you haven't defined manually the order of fields to be printed by the tool (either with command line option or result options file). To avoid this problem, define the fields to be printed and their order in the results config file with the fields property or with the -fields option in ResPrint.

· Results Processing Index · Manual index ·

Last modified: Sun Aug 1 19:09:42 EEST 1999