| As a test development technology that effectively solves the response bias in traditional Likert personality assessment,multidimensional forced-choice test(MFC)has shown great prospects in the high-stake talent selection environment since its emergence.Recent research results show that MFC not only has similar or higher construct validity and criterion-related validity to Likert personality scale,but also effectively guarantees the fairness and accuracy of the personnel selection and recruitment process.With the rapid development of modern psychometric technology,the forced-choice test has changed from the traditional scoring method that only reflects the relative level of individual internal personality traits to the multidimensional forcedchoice item response theory(MFC-IRT)scoring model,which can realize the comparison of personality traits between individuals,and a large number of studies have also confirmed the superior performance of the MFC-IRT model in terms of parameter estimation accuracy and its anti-fake strength.With the increasing requirements of recruitment organizations in terms of measurement accuracy and test efficiency of forced-choice tests,the multidimensional forced-choice computerized adaptive testing(MFC-CAT)that accompanies MFC-IRT has gradually been promoted and provided a new perspective for personality assessment.Due to the advantages of anti-fake,high efficiency and high measurement accuracy,MFC-CAT has been actively concerned by psychometric scholars and other experts in related fields since entering the 21st century.However,in previous MFC-CAT studies,blocks(or called testlet)in forced-choice tests were preassembled with manpower,and researchers only considered nonstatistical constraints during the assembly process,therefore neglecting the effect of the combination relationships among statements on the measurement accuracy.From the beginning of the assembly to the completion of the item bank construction,once the combinational relationship between the statements is fixed,it cannot be changed again during the test.Therefore,this assembly method does not seem to consider the impact of the different combinational relationships between statements on measurement accuracy.In addition,this traditional assembly method also has the disadvantages of consuming a lot of time and economic cost in operation.Therefore,this research aims to propose an MFC-CAT algorithm that can assemble blocks online.The algorithm can not only make the combinational relationship between statements dynamically change with the changes in the candidates’ performance in the current measurement of personality traits,but also consider the basic principles of assembling FC blocks while ensuring high measurement accuracy,for example,the proximity of social desirability,the proportion of opposite polarity blocks,etc.Two carefully designed experiments are carried out to explore the proposed method’s performance in terms of measurement accuracy and its violations of the principle of assembling FC blocks.In addition,this research also applies the online-assembled method to the measurement of two personality traits including empathizing quotient and systemizing quotient in a highly simulated environment.The results of the study found that the proposed method has been well verified in the experiments,and is superior to the traditional preassembled method in terms of measurement accuracy and the degree of violations of the principle of assembling FC blocks,which can further promote the application of MFC-CAT in the actual environment. |
PDF Full Text Request