Sex and Location as Determinants of Handedness: Reply to Vuoksimaa and Kaprio (2010)

Acknowledgement:

Vuoksimaa and Kaprio (2010) provided evidence regarding the influences of sex and geographical location on the distribution of handedness, supplementing the evidence presented in our meta-analysis (Papadatou-Pastou, Martin, Munafò, & Jones, 2008). Our meta-analysis integrated the results from a wide range of studies that had reported both female and male handedness distributions. Through analysis of 199 data sets drawn from 144 studies, an overall best estimate of the male-to-female ratio for left- to right-handedness odds was 1.23, with a 95% confidence interval = 1.19, 1.27. To be specific, we note that this is the odds ratio that would be yielded by female and male left-handedness rates of 10% and 12%, respectively. However, there was significant heterogeneity over data sets in the values of the overall odds ratio, with significant moderation by geographical location (i.e., nation) and study date.

With regard to location as a moderator, the differences in handedness sex odds ratios were found to be linked to cultural indices, introduced by Hofstede (1983, 2001), that differ across nations. In particular, the odds ratio increased with higher masculinity (i.e., relating to higher levels of differentiation between social sex roles) and with higher uncertainty avoidance (i.e., relating to higher levels of the promotion of conformity); it decreased with higher levels of individualism (i.e., relating to higher levels of reliance on the individual). According to the norms of Hofstede (2001), Scandinavian (Nordic) nations have low levels of cultural masculinity, with Sweden and Norway occupying the first and second ranks, with index scores of 5 and 8 (on an approximately 100-point scale), respectively, and Finland (often grouped with these countries) occupying the seventh position, with an index score of 26. Indeed, examination of data sets from these countries indicated an overall sex difference odds ratio of 0.88. Because the 95% confidence interval (CI) for this group (0.67, 1.17) includes zero, there is the distinct possibility of there being no handedness sex effect in these nations (see Papadatou-Pastou et al., 2008).

Vuoksimaa and Kaprio (2010) suggested that these results would change if certain investigations were both subtracted from and added to the Scandinavian studies analyzed by Papadatou-Pastou et al. (2008). In terms of investigations that they believe should not have been included, we contend that their three proposed deletions are not appropriate. Specifically, they questioned the inclusion of (a) the Norwegian study reported by Götestam (1990), because the proportions of female and male students participants contributed by different academic areas were not identical and (b) the Swedish study reported by Beckman and Elston (1962), because the method of establishing handedness was not described. Yet, neither of these features satisfies the criteria for exclusion in the report of Papadatou-Pastou et al. (2008), which involved factors that would increase the proportion of left-handed participants selected for the study, the testing of persons up to the age of 16, and the testing of twins. Vuoksimaa and Kaprio (2010) also questioned the numerical accuracy of the contribution to the meta-analysis from (c) the Finnish study reported by Kauranen and Vanharanta (1996). Yet, we contend that Vuoksimaa and Kaprio misread Table 1 of our study (Papadatou-Pastou et al., 2008). Specifically, the odds ratio on the first of the five lines of the Kauranen and Vanharanta entry represented only one of their five data sets, not the overall odds ratio assumed by Vuoksimaa and Kaprio (2010).

In terms of investigations they believe should be added, Vuoksimaa and Kaprio (2010) estimated the combined odds ratio for four studies not included in Papadatou-Pastou et al. (2008). Of these, Salvesen, Vatten, Eik-Nes, Hugdahl, and Bakketeig (1993) and Kieler, Axelsson, Haglund, Nilsson, and Salvesen (1998) studied children and were therefore excluded by the method of our meta-analysis (Papadatou-Pastou et al., 2008). The remaining two are Finnish studies, which we acknowledge as eligible for inclusion within the method of Papadatou-Pastou et al. (2008): Pekkarinen, Salminen, and Järvelin (2003) and a sample described by Vuoksimaa and Kaprio (2010) themselves, which assessed the prevalence of adult left-handed writers. We thus carried out a meta-analysis on a combination of these two new studies and the five original studies in the Scandinavian group from Papadatou-Pastou et al. (2008): Götestam (1990), Beckman and Elston (1962), Kauranen and Vanharanta (1996), Levander and Schalling (1998), and Ofte (2002).

Over the seven data sets, a fixed effects analysis gave an overall odds ratio for sex difference in handedness of 1.24 (95% CI = 1.09, 1.41). The low level of heterogeneity among studies, I2 = 25.55, did not attain significance, Q(6) = 8.06, p = .23. Nevertheless, there was a significant moderating contrast between the three Finnish studies and the four Swedish/Norwegian studies, Q(1) = 7.06, p = .008, with the odds ratio higher for the Finnish studies (OR = 1.35; 95% CI = 1.17, 1.56) than for the non-Finnish studies (OR = 0.88; 95% CI = 0.66, 1.16). Note that the handedness sex effect was significant for the Finnish studies (Z = 4.07, p < .001) but not for the non-Finnish studies (Z = 0.92, p = .36).

Before concluding that the difference between the two groups of studies has its origin in a factor directly linked to geographical location, however, the theoretical underpinnings of the comparison process should also be examined. The present use of the odds ratio is associated implicitly with an essentially multiplicative model of the sex difference in handedness. Whereas, as noted previously, a female rate of left-handedness of 10% and a male rate of 12% correspond to an odds ratio of 1.23, a multiplicative process such as halving the female and male rates to 5% and 6%, respectively, leaves the odds ratio almost unchanged at 1.21. However, an alternative possibility, in principle, would be to hypothesize an additive model, congruent with analysis of variance or more generally with processes that are exclusive (Jones, 1987). In this case, an additive process such as lowering the rates by 5% would lead to female and male rates of left-handedness of 5% and 7%, respectively, yielding a much higher odds ratio (specifically, 1.43). It would be premature to identify substrates of such processes in the present case, but they could relate to different ways in which environmental factors may interact with genetic factors in the determination of handedness.

It is noteworthy that the overall levels of left-handedness in each of the additional Finnish studies here were relatively low, with female and male rates of left-handedness of 3.7% and 5.3%, respectively, in Vuoksimaa and Kaprio (2010) and 6.2% and 8.1% in Pekkarinen et al. (2003). In contrast, Corballis (1997) noted that aggregate values of 9.8% and 11.6% were found for female and male rates of left-handedness, respectively, in the wide-ranging survey of McManus and Bryden (1992). For mathematical reasons alone, a fixed additive contribution to the handedness sex effect would give rise to higher odds ratios at low overall rates of left-handedness than it would at high rates of left-handedness. Therefore the relatively low rates of left-handedness in the Finnish studies may have contributed to their observed high level of the sex-difference odds ratio.

Is there direct evidence for an inverse relation between sex-related odds ratio and overall rate of left-handedness, as an additive handedness sex effect would produce? One possible indication is the significant moderating effect of publication date on the handedness sex effect that was observed by Papadatou-Pastou et al. (2008). Their meta-regression showed that as the date went back in time from 2007 to 1927, the estimated sex effect odds ratio rose from 1.24 to 1.46. On an additive basis, such an increase would be the expected consequence of the lower overall levels of left-handedness observed in earlier studies. Mean rates of left-handedness can be calculated as 13.3% and 7.3% for births in the periods 1955–1979 and 1880–1939, respectively, in the tabulation of McManus and Bryden (1992), demonstrating an inverse relation with the sex-effect odds ratio.

In summary, we agree with Vuoksimaa and Kaprio (2010) that there is evidence against the hypothesis that the sex handedness effect is completely absent in Scandinavian (Nordic) studies. However, evidence also remains for a partial absence of the effect. As we have argued herein, understanding the variation in its magnitude across location (nation) and time may require consideration of a range of variables that includes cultural factors (e.g., Hofstede, 2001) and overall handedness levels.

Beckman, L., & Elston, R. (1962). Data on bilateral variation in man: Handedness, hand clasping and arm folding in Swedes. Human Biology, 34, 99–103.

Corballis, M. C. (1997). The genetics and evolution of handedness. Psychological Review, 104, 714–727.

Götestam, K. O. (1990). Lefthandedness among students of architecture and music. Perceptual and Motor Skills, 70, 1323–1327.

Hofstede, G. (1983). National cultures revisited. Behavior Science Research, 18, 285–305.

Hofstede, G. (2001). Culture's consequences: Comparing values, behaviors, institutions, and organizations across nations (2nd ed.). Thousand Oaks, CA: Sage.

Jones, G. V. (1987). Independence and exclusivity among psychological processes: Implications for the structure of recall. Psychological Review, 94, 229–235.

Kauranen, K., & Vanharanta, H. (1996). Influences of aging, gender, and handedness on motor performance of upper and lower extremities. Perceptual and Motor Skills, 82, 515–525.

Kieler, H., Axelsson, O., Haglund, B., Nilsson, S., & Salvesen, K. Å. (1998). Routine ultrasound screening in pregnancy and the children's subsequent handedness. Early Human Development, 50, 233–245.

Levander, M., & Schalling, D. (1998). Hand preference in a population of Swedish college students. Cortex, 24, 149–156.

McManus, I. C., & Bryden, M. P. (1992). The genetics of handedness, cerebral dominance and lateralization. In I.Rapin & S. J.Segalowitz (Eds.), Handbook of neuropsychology (Vol. 6, pp. 115–144). Amsterdam: Elsevier.

Ofte, S. H. (2002). Right–left discrimination: Effects of handedness and educational background. Scandinavian Journal of Psychology, 43, 213–219.

Papadatou-Pastou, M., Martin, M., Munafò, M. R., & Jones, G. V. (2008). Sex differences in left-handedness: A meta-analysis of 144 studies. Psychological Bulletin, 134, 677–699.

Pekkarinen, A., Salminen, S., & Järvelin, M.-R. (2003). Hand preference and risk of injury among the Northern Finland birth cohort at the age of 30. Laterality, 8, 339–346.

Salvesen, K. Å., Vatten, L. J., Eik-Nes, S. H., Hugdahl, K., & Bakketeig, L. S. (1993). Routine ultrasonography in utero and subsequent handedness and neurological development. British Medical Journal, 307, 159–164.

Vuoksimaa, E., & Kaprio, J. (2010). Sex differences in left-handedness are also evident in Scandinavia and in twins: Comment on Papadatou-Pastou, Martin, Munafò, and Jones (2008). Psychological Bulletin, 136, 344–347.

Submitted: January 27, 2010 Revised: February 1, 2010 Accepted: February 3, 2010