Do sex differences in a faceted model of fluid and crystallized intelligence depend on the method applied? Ricarda Steinmayr a, ⁎, André Beauducel b , Birgit Spinath a a University of Heidelberg, Department of Psychology, Hauptstrasse 47-51, D-69117 Heidelberg, Germany b University of the Federal Armed Forces, Personality Psychology, Holstenhofweg 85, D-22043 Hamburg, Germany article info abstract Article history: Received 6 May 2009 Received in revised form 11 August 2009 Accepted 17 August 2009 Available online 9 September 2009 Recently, different methodological approaches have been discussed as an explanation for inconsistencies in studies investigating sex differences in different intelligences. The present study investigates sex differences in manifest sum scores, factor score estimates, and latent verbal, numerical, figural intelligence, as well as fluid and crystallized intelligence as measured by the German Intelligence-Structure-Test 2000-R (IST 2000-R; Liepmann, Beauducel, Brocke, & Amthauer, 2007). The not population-representative sample consisted of 977 German 11th and 12th graders enrolled in a “Gymnasium” (551 female; mean age: M= 16.70; SD = 0.65) who completed the IST 2000-R. Sex differences in fluid and crystallized intelligence were not influenced by the method applied with men performing better than women. However, extent and direction of sex differences in verbal, numerical, and figural intelligence differed by the method applied. Whereas there was a male advantage in all three factors measured as manifest sum scores, women performed better in verbal intelligence as measured by factor scores or as latent variables. Effect sizes of sex differences in numerical and figural intelligence were also greatly reduced when applying the latter two methods. Results are discussed with regard to their theoretical and practical implications. © 2009 Elsevier Inc. All rights reserved. Keywords: Sex differences Latent cognitive variables Factor scores 1. Introduction Sex differences in intelligence have been extensively investigated ever since the first intelligence tests have been introduced. Despite the substantial attention the topic has received, it is far from being thoroughly illuminated as re- search on sex differences in intelligence has partly produced inconsistent results concerning the presence, magnitude, and direction of the effects. Possible explanations for these in- consistencies are developmental effects (cf. Lynn, 1999), selective samples (cf. Dykiert, Gale, & Deary, 2009), and the measures used (cf. Lynn, 1999). A further explanation for the inconsistencies might be the methodological approach applied to the investigation of group differences (cf., e.g., Keith, Reynolds, Patel, & Ridley, 2008). Recently sex differ- ences in intelligence have been investigated by multivariate latent variable approaches rather than comparing manifest intelligence test scores. An advantage of a latent variables approach is that it allows conclusions about sex differences in underlying, pure intelligence factors. Contrary to this, comparing manifest intelligence test scores might yield misleading results about the true nature of sex differences. Nevertheless, it is important to investigate sex differences in measured manifest intelligence because real world decisions, such as selection for jobs, are based on manifest intelligence test scores. The present study examines sex differences in manifest sum scores, factor score estimates, and latent verbal, numerical, figural intelligence, as well as fluid and crystallized intelligence as measured by the German Intelligence-Struc- ture-Test 2000-R (Liepmann, Beauducel, Brocke, & Amthauer, 2007). Thus, the present study aims to compare the three methodological approaches and their impact on the emer- gence of sex differences. Intelligence 38 (2010) 101–110 ⁎ Corresponding author. E-mail addresses: ricarda.steinmayr@psychologie.uni-heidelberg.de (R. Steinmayr), beauduce@hsu-hh.de (A. Beauducel), birgit.spinath@psychologie.uni-heidelberg.de (B. Spinath). 0160-2896/$ – see front matter © 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.intell.2009.08.001 Contents lists available at ScienceDirect Intelligence