# Factor Analysis

Páginas: 4 (804 palabras) Publicado: 10 de marzo de 2013
GVPT 722: Factor Analysis in Stata
Example of a single underlying variable: religious commitment (2000 NES) (1) Correlation matrix: . corr prayer chattend guidance bibread (obs=1712) | prayerchattend guidance bibread -------------+-----------------------------------prayer | 1.0000 chattend | 0.5605 1.0000 guidance | 0.6682 0.5907 1.0000 bibread | 0.5822 0.5876 0.5532 1.0000 (2) Factoranalysis, with principal components factors . factor guidance prayer chattend bibread, pcf (obs=1712) (principal component factors; 1 factor retained) Factor Eigenvalue Difference ProportionCumulative -----------------------------------------------------------------1 2.77196 2.28764 0.6930 0.6930 2 0.48432 0.06267 0.1211 0.8141 3 0.42164 0.09956 0.1054 0.9195 4 0.32208 . 0.0805 1.0000Factor Loadings Variable | 1 Uniqueness -------------+--------------------guidance | 0.84722 0.28222 prayer | 0.84675 0.28301 chattend | 0.82049 0.32679 bibread | 0.81484 0.33603 (3) Creating ameasure of religious commitment with a factor score: . score commitfac (based on unrotated factors) Scoring Coefficients Variable | 1 -------------+---------guidance | 0.30564 prayer | 0.30547 chattend| 0.29600 bibread | 0.29396 . summ commitfac Variable | Obs Mean Std. Dev. Min Max -------------+-------------------------------------------------------commitfac | 1712 -2.27e-09 1 -1.619581.904733

(4) Regression model with the commitment factor score: . reg bushft partyid ideol7b educ income south race2 age sex commitfac, beta Source | SS df MS-------------+-----------------------------Model | 165477.859 9 18386.4288 Residual | 312335.494 781 399.917405 -------------+-----------------------------Total | 477813.353 790 604.827029 Number of obs F( 9, 781) Prob > F R-squaredAdj R-squared Root MSE = = = = = = 791 45.98 0.0000 0.3463 0.3388 19.998

-----------------------------------------------------------------------------bushft | Coef. Std. Err. t P>|t| Beta...

Regístrate para leer el documento completo.