The Development Of The Computerize Adaptive Test For Depression Assessment Based On Bifactor-model

It will be an important means to use depression assessment tool in assessing the degree of depression.The assessment tool with high quality not only can evaluate person's degree of depression,but also can provide some useful information for screening of depression person and clinical assessment.Now,many depression assessment tools are development based on Classical Test Theory?CTT?,these tools just can provide total score,and can not provide more information.In addition,it will increase the amount of item,if we want to measure more facets.As the development measurement science,Multidimensional Item Response Theory?MIRT?becoming mature,it provides researchers with a new angle.Under the MIRT,the scale can measure more facets,besides,because the difficulty parameter and the latent trait are in the same measuring scale,we can develop the computerized adaptive test?CAT?,so that can abate the burden of the subjects.The aim of this research is to develop a CAT to assess person's depression degree based on bifactor model.This research is made up with two parts.The first study is to develop new methods based the CAT with bifactor model?BICAT?and multidimensional bifactor algorithm for CAT?MBICAT?,in addition,the rationality and scientificity of the new methods also been verified with Monte Carlo simulation,so that we can provide the foundations for developing the CAT of depression assessment based on bifactor model?BICAT-D?.The second study is to construct a BICAT-D base on the first study,and to evaluate the efficiency and validity of BICAT-D with practical measured data.The result shows that:?1?The precision of MBICAT is better than BICAT,but it will improve the precision of BICAT when the BICAT combine the multi-estimate.?2?Although the precision of MBICAT is well,but the test length of MBICAT also longer.In order to control the test length,we can constrain the dimension that have satisfied the termination criterion,it can reduce the number of item but won't reduce the precision.?3?Under the BICAT method,the precision is close to the precision under the MBICAT method when all dimensions have been tested and to re-estimate the potential trait with multi-estimate method.When program run time need to beconsidered,it will be a better way to use BICAT method with re-estimate and multi-estimate.?4?When there is 157 items in item bank,by using BICAT-D can reduce the participants' burden,the mean number of item person need do is 29.4,it reduce 80.8%item.?5?When adopt the score of SDS,CES-D,PHQ-9 as the criterion,the result shows that BICAT-D with a well validity.?6?Using the ROC curve analysis decide the cutting point,when adopt,???_BICAT-D,G=0 as the cutting point can insure the sensitivity specificity satisfied the ideal criterion.?7?BICAT-D not only can provide person's depression degree,but also can provide information about special dimension,for example: mood,cognition,and tell the person with same depression degree apart.