A study was undertaken of the test-retest reliability of a skills audit questionnaire by administering a retest questionnaire to a sub-population of 103 out of a group of 130 administrative workers, six months after they had completed the original questionnaire. For responses relating to possession/non-possession of skill the overall extent of change was 11.96%. Some questions achieved little or no discrimination between individuals. Exclusion of these questions increased the extent of change to 15.33%. The extent to which individual respondents changed between first and second testing ranged from 2% up to 32%. For some individuals the change was entirely in the direction of a reported skills gain, while for others the change reported was entirely a skills loss. Response changes to individual questions and to question groups were examined. The analysis illustrates that computation of reliability figures from gross change data may mask large net changes at the level of individual respondents and individual questions. The importance of judging the validity and reliability of a questionnaire with respect to the purpose for which information is used is discussed. The thesis concludes that the design of skills audit questionnaires requires attention to their psychometric properties, in particular, content validity, discriminatory power, and reliability over time.
History
Year
1992
Thesis type
Doctoral thesis
Faculty/School
Department of Management
Language
English
Disclaimer
Unless otherwise indicated, the views expressed in this thesis are those of the author and do not necessarily represent the views of the University of Wollongong.