%0 Journal Article %T Optimizing technical precision of measurement in computerized psychological assessment on Windows platforms %A JOACHIM H£żUSLER %A MARKUS SOMMER %A STEFAN CHROUST %J Psychology Science %D 2007 %I %X Reaction times and response latencies are required to measure a variety of ability and personality traits. If reaction times are used to measure rather elementary cognitive tasks, the inter-individual variance in the measured reaction times are usually small in the sense that the central 50 percent of a norm population range within less than 100ms. Technical measurement errors therefore have the potential to seriously affect the validity of diagnostic judgments based on such measures. Thus the target of this paper is to investigate the magnitude of possible errors of measurement due to technical reasons and to suggest ways to prevent or at least consider those in the diagnostic process.In Study I a highly precise 'artificial respondent' was applied to simulate reactions corresponding to a given percentile rank on 3 different tests (DG-Lokation CORPORAL, Alertness TAP-M, RT/S9 Vienna Test System) on 11 different computer systems. The result output of the tests was compared to the reaction times, actually provided by the artificial respondent. Results show, that there are detectable errors of measurement - depending on the hardware and software specifications of the computer system used. In the test DG-Lokation these bias caused an offset in the tests main variable of up to 20 percentile ranks.In Study II a self-calibration unit which is part of the Vienna Test System (Version 6.40) was investigated, using the same experimental setup. After calibration, the bias detected can be reduced to the magnitude of about 1 percentile rank on all computer systems tested.It thus can be concluded, that time critical computer based tests typically bear the risk of technical errors of measurement. Depending on how the test is programmed, the errors arising on some computer configurations can cause even severe changes in diagnostic judgment formation. In contrast, self-calibration proved to be an effective tool to permitting the user not only to control but also to ensure the precision of measurement, independent of the properties of the computer system he is administering his test on. %K computerised testing %K technical precision of measurement %K self-calibration %K reaction tests %U http://www.psychologie-aktuell.com/fileadmin/download/PschologyScience/ps_2_2007_116-131.pdf