1. The status of controls in the birth-order literature is not always straightforward, so I have adopted a four-step scheme for such controls in my own meta-analysis: (0) studies without any mention of controls; (1) studies in which controls, although mentioned as variables that were included in the study, are not employed in any of the relevant statistical tests; (2) studies in which sibship size and social class are “partially controlled” (for example, by the presence of tests for birth-order effects in sibship sizes of two versus sibship sizes of three or more); and (3) studies that are “fully controlled” by the inclusion of control variables in a formal analysis of variance or multiple regression model, or by multilevel stratification of data according to control variables. Only studies coded “2” or “3” are considered by me as “controlled” for a given variable in my analysis ( Sulloway , 2002b). By contrast, Ernst and Angst (1983), and following them, Townsend, sometimes consider studies coded as “1” as being controlled. The meta-analytic results in my Table 3 include only those studies for which Ernst and Angst (1983) report the presence of controls. Anyone who carefully checks the publications in Townsend's (2000) Appendix B, which lists 26 studies in which controls for sibship size or social class were supposedly overlooked by Ernst and Angst (1983), will find that more than two-thirds of the codable outcomes have the status of their controls or findings incorrectly reported by Townsend, thus undermining the validity of his claims. Elsewhere I analyze studies with overlooked controls ( Sulloway , 2002b).
2. For the data summarized in Table 4, the mean correlation for sex differences in the Big Five personality dimension scores is .12 (N=4,716), which is similar to effect sizes reported elsewhere ( Costa and McCrae , 1992; Feingold , 1994; Costa , Terracciano , and McCrae , 2001).
3. To their credit, Ernst and Angst (1983:ix) explicitly warn readers about some of these problems, such as the omission of some published findings from their discussions. Although I corrected errors and identified many overlooked findings in my earlier meta-analysis ( Sulloway , 1995), I did not detect all such instances. These additional corrections (and findings) do not in any way affect the general conclusions that I reached in my earlier vote-counting meta-analysis.
4. For dominance and birth order among spouses, pr=-.07 (N=780, p<.05; this and subsequent effects are controlled for age, sex, sibship size, and social class). For dominance and birth order among roommates, pr=-.16 (N=165, p<.05). For friends, the same partial correlation is -.05 (N=1,063, p<.14). For the relationship between birth order and dominance with respect to a female friend, pr=-.09 (N=630, p<.03).