Dear list,
I encountered some strange difference between stata 9 and stata 8.2:
When I use a factor analysis, pcf:
-- factor grata-gratn gratr if s_func!=4, pcf
And then do a rotation:
-- rotate, varimax
My results differ between Stata 9 and 8.2. I am aware that factor
analysis might be differing between different software, but this seems
to be a bit strange. My results in Stata 8.2 are:
. rotate, varimax
(varimax rotation)
Rotated Factor Loadings
Variable | 1 2 3 4 Uniqueness
-------------+------------------------------------------------------
grata | 0.07200 0.83750 -0.14724 0.01761 0.27142
gratb | 0.68471 -0.25914 -0.01918 0.06472 0.45947
gratc | 0.77708 0.03638 -0.08072 0.05893 0.38483
gratd | -0.03944 0.83761 0.03883 -0.11732 0.28159
grate | 0.77688 -0.08293 -0.12157 0.19759 0.33576
gratf | 0.73725 0.01383 -0.21993 -0.12543 0.39217
gratg | 0.18574 0.03545 -0.80388 -0.05362 0.31514
grath | 0.33253 0.06513 -0.56911 0.32161 0.45787
grati | 0.04308 0.17309 -0.82092 -0.02904 0.29343
gratj | 0.26887 0.10299 -0.47618 -0.34600 0.57064
gratk | 0.55214 -0.02347 -0.24318 0.41838 0.46041
gratl | 0.64118 0.23528 -0.12570 0.09820 0.50809
gratm | -0.07864 0.81732 -0.14104 0.02735 0.30516
gratn | 0.57284 0.04382 -0.29150 -0.07797 0.57888
gratr | 0.09351 -0.04578 0.05855 0.85439 0.25575
The results in Stata 9 are:
Rotated factor loadings (pattern matrix) and unique variances
---------------------------------------------------------------------
Variable | Factor1 Factor2 Factor3 Factor4 | Uniqueness
-------------+----------------------------------------+--------------
grata | 0.0720 0.8375 0.1472 0.0176 | 0.2714
gratb | 0.6847 -0.2591 0.0192 0.0647 | 0.4595
gratc | 0.7771 0.0364 0.0807 0.0589 | 0.3848
gratd | -0.0394 0.8376 -0.0388 -0.1173 | 0.2816
grate | 0.7769 -0.0829 0.1216 0.1976 | 0.3358
gratf | 0.7372 0.0138 0.2199 -0.1254 | 0.3922
gratg | 0.1857 0.0354 0.8039 -0.0536 | 0.3151
grath | 0.3325 0.0651 0.5691 0.3216 | 0.4579
grati | 0.0431 0.1731 0.8209 -0.0290 | 0.2934
gratj | 0.2689 0.1030 0.4762 -0.3460 | 0.5706
gratk | 0.5521 -0.0235 0.2432 0.4184 | 0.4604
gratl | 0.6412 0.2353 0.1257 0.0982 | 0.5081
gratm | -0.0786 0.8173 0.1410 0.0274 | 0.3052
gratn | 0.5728 0.0438 0.2915 -0.0780 | 0.5789
gratr | 0.0935 -0.0458 -0.0585 0.8544 | 0.2557
---------------------------------------------------------------------
The difference occurs AFTER the VARIMAX-Rotation, the unrotated results
are identical. Especially look at Factor3 and the signs of the loadings
in that factor -- it seems like the whole factor is inverted.
Any hints what I am missing?
Daniel Schneider
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
