The preliminary research on unifying different neuropsychological scales under the guideline of item response theory
10.3760/cma.j.issn.0254-9026.2018.08.008
- VernacularTitle:项目反应理论指导下的不同神经心理学量表评分一致性探讨
- Author:
Xiaodong PAN
1
;
Chen ZHOU
;
Yiran HE
Author Information
1. 江苏省老年医学研究所
- Keywords:
Cognition disorders;
Neuropsychological scale;
Item response theory
- From:
Chinese Journal of Geriatrics
2018;37(8):870-874
- CountryChina
- Language:Chinese
-
Abstract:
Objective To explore the probability to unify the scores of different neuropsychological scales under the guideline of item response theory (IRT).Methods Two hundred and three old individuals with a mean age of (80.86±9.52) years were enrolled in this study,including 101 normal individuals and 102 patients with mild cognitive impairment (MCI).The data of cognitive ability assessed using MoCA and MMSE scales were collected and analyzed according to IRT with Stata 14.0 software.The population,qualitative data,and scaled data were analyzed by Hybrid model,2PL model,and GRM model,respectively.Finally,model fitting was conducted in these data.We used test characteristic curves(TCC)to calculate corresponding theta values of the scores of MMSE and MoCA,and draw cartograms after fitting the model.Results The data of MoCA and MMSE conformed to the characteristics of unidimension.The scores of MoCA and MMSE were 26.37 ±2.63 and 29.06±1.32,respectively,in the normal group,and those were 21.57±3.47 and 26.62± 2.89,respectively,in the MCI group.The Log likelihood for the fitting model of MoCA and MMSE was-1 670.9012 and-773.9797,respectively.The TCC indicated that the score of MoCA was lower than that of MMSE at the same level of theta value.The TCC of the two scales were both steep at first,and then became flat.In evaluating the cognition ability,the MoCA score was more precise than MMSE score,and MMSE score was easier to show ceiling effect than MoCA score.Conclusions Under the guideline of item response theory,we can unify the scores of two scales in order to make the scores of two scales fit with each other and be in common use during study and research.
