Skip to main content
. 2022 May 27;24(6):760. doi: 10.3390/e24060760

Table 1.

Model comparisons based on information criteria for the three ability domains—mathematics, reading and science—in PISA 2009.

Mathematics Reading Science
Model AIC BIC ΔGHP AIC BIC ΔGHP AIC BIC ΔGHP
1PL 217510 217779 0.0059 413555 414317 0.0055 347819 348222 0.0062
1PCL 220022 220291 0.0122 414757 415519 0.0070 348756 349160 0.0077
1PLL 216882 217151 0.0043 416988 417751 0.0098 348984 349388 0.0081
1PGL 216784 217068 0.0041 413369 414146 0.0053 347804 348223 0.0062
2PL 215621 216144 0.0012 410032 411541 0.0011 344597 345389 0.0009
4PGL 215142 216188 0.0000 409163 412182 0.0000 344064 345648 0.0000
3PLQ 215153 215938 0.0000 409327 411591 0.0002 344097 345285 0.0001
3PLRH 215174 215959 0.0001 409275 411539 0.0001 344083 345271 0.0000
3PL 215486 216099 0.0009 409767 411605 0.0008 344420 345362 0.0006
4PL 215179 216060 0.0001 409296 411852 0.0002 344105 345368 0.0001
4PLQ 215168 216102 0.0001 409245 411913 0.0001 344089 345464 0.0000

Note. AIC = Akaike information criterion; BIC = Bayesian information criteria; DGHP = difference in Gilula–Haberman penalty (GHP) between a particular model and the best-fitting model in terms of GHP; For model descriptions see Section 2.1 and Equations (3) to (14). For AIC and BIC, the best-fitting model and models whose information criteria did not deviate from the minimum value by more than 100 are printed in bold. For DGHP, the model with the smallest value and models with DGHP values smaller than 0.0005 are printed in bold.