Confirmatory Factor Analysis SPSS/AMOS The WISC, Verbal IQ • INFOrmation – general knowledge questions • COMPrehension – of social situations and common concepts • ARITHmetic • SIMILarities – how are two words similar • VOCABulary • DIGITspan – repeating strings of digits after hearing them The WISC, Performance IQ • PICTureCOMPletion – identify the missing part • PictureARrANGement – arrange pictures to tell a story. • BLOCK design – arrange blocks to match model. • OBJECT assembly – puzzles involvement arrangement into a whole • CODING – associate simple shapes with symbols coding them Download From BlackBoard • CFA-WISC.sav • CFA-Wisc.amw • CFA-Wisc2.amw • CFA-Wisc-Amos-Output.doc Bring CFA-WISC into SPSS Analyze, AMOS AMOS • Open, CFA-Wisc.amw • Select Data Files, Working, View Data, OK • Analysis Properties – Minimization history – Standardized estimates – Squared multiple correlations – Residual moments – Modification indices • • • • Calculate Estimates View the output path diagram Standardized View Text (Output) Text Output • • • • Chi-square = 70.236 Degrees of freedom = 43 Probability level = .005 Poor fit or just too much power? Standardized Residual Covariances CODING OBJECT BLOCK PARANG PICTCOMP DIGIT CODING .000 OBJECT .159 .000 BLOCK .758 .156 .000 PARANG .049 -.180 .358 .000 -1.513 .331 -.301 -.414 .000 2.062 -1.248 -1.098 .519 -.805 .000 .886 -.908 -.152 -1.058 .199 -.079 SIMIL -.931 .443 -.272 1.327 1.573 -.191 ARITH .872 -1.884 .576 .905 -.553 .623 COMP .413 1.186 1.159 -.073 2.112 -.431 -.333 -.872 -.965 -.127 -.464 .618 PICTCOMP DIGIT VOCAB INFO Large residuals for Comp-Pictcomp and Digit-Coding. Fit • • • • GFI = .931 CFI = .941 RMSEA = .06 Fit is not bad. Modification Indices M.I. 4.509 5.513 4.194 4.608 4.065 4.276 Par Change .171 -.194 -.138 .143 -.122 -.109 CODING OBJECT PICTCOMP DIGIT VOCAB ARITH <--<--<--<--<--<--- DIGIT ARITH CODING CODING PARANG OBJECT COMP <--- Performance IQ 4.569 .438 COMP COMP <--<--- OBJECT PICTCOMP 5.317 7.109 .142 .159 I am going to add a path from Performance to COMP. CFA-Wisc2.amw has this path diagram. 2 Dropped Significantly • • • • Chi-square = 60.295 Degrees of freedom = 42 Probability level = .033 Change in 2 = (70.326 – 20.295) = 9.94 – On 1 df, is significant. Better Fit • GFI = .942, had been .931 • CFI = .960, had been .941 • RMSEA = .050, had been .060 Regression Weights COMP INFO PICTCOMP VOCAB SIMIL ARITH CODING OBJECT BLOCK PARANG <--<--<--<--<--<--<--<--<--<--- Verbal IQ Verbal IQ Performance IQ Verbal IQ Verbal IQ Verbal IQ Performance IQ Performance IQ Performance IQ Performance IQ Estimat e 1.491 2.256 1.790 2.273 2.205 1.307 .200 1.633 1.823 1.189 S.E. C.R. P .254 .200 .239 .201 .227 .173 .253 .234 .219 .224 5.860 11.286 7.474 11.297 9.721 7.562 .791 6.990 8.310 5.302 *** *** *** *** *** *** .429 *** *** *** • The path to CODING is not significant. I am going to eliminate CODING. CFA-Wisc3.amw has this path diagram. 2 No Longer Significant • Chi-square = 45.018 • Degrees of freedom = 33 • Probability level = .079 Yet Better Fit • GFI = .952, had been .942 • CFI = .974, had been .960 • RMSEA = .046, had been .050