The Study Of Psychological Measurement Model Of Forced-Choice Personality Test

One of the most important is the accuracy of measurement for personality test which are used in industrial organization as tools.The most popular self –report personality inventory is in the Likert type,which could not work well in high-stakes selection,be-cause participants can easily make response bias in personality measures that rely on Likert rating scales,which could do much good for themselves,the forced-choice(FC)item format was introduced in an attempt to minimize response biases as found with Likert-type rating scales,In forced-choice questionnaires,participants have to make choices between two or more items presented at the same time,However,if scored with traditional methodology,FC instruments produce ipsative data,the limitation of ipsative data in the statistical analysis which hindered the application in the field of personality test.Luckily,with the progress of modern measuring technology,researchers find the probability of combining the item responses theory(IRT)with FC format,which maybe solve the ipsative of traditional test score,several models based IRT that could deal with the FC data have been developed for now.In this article,we introduce the Thurstonian IRT model which is a normal ogive model,and this model could be straightforwardly estimated using structural equation modeling(SEM)software Mplus.A simulation study and an empirical research are performed to prove and apply the model,in simulation study,we manipulate three variables which are: a,the direction of statements in pairs;b,the correlation of traits that measured by FC questionnaires;c,the number of items that measure per trait.The results show that Thurstonian IRT model could deal with the data produced by questionnaires that measure different personality dimensions at the same time,and would produce a good fitting,the quality of FC questionnaires is not only affected by the direction of statement in blocks but also by the length of the test,in empirical research part of this article,we apply the NPI-40 to get data,which was structured in forced-choice form initially.However,in practical application,many researchers change the NPI forced-choice format to single stimulus format for the limiting of the traditional method of scoring,which maybe not appropriate because of the structure could be changed with the format presented,we compared the data from two format: 5-point rating scale and the forced-choice presentation,result shows that the forcedchoice format has a better fitting index.Research shows: 1,Thurstonian IRT model can directly fits the data from FC,and has a better index;2,the length of questionnaires and the direction of statement would affect the quality of FC questionnaires;3,transform the FC format to single stimulus would change the quality of NPI-40.