Supplemental Materials

Coresidence Duration and Cues of Maternal Investment Regulate Sibling Altruism Across Cultures

by D. Sznycer et al., 2016, Journal of Personality and Social Psychology

http://dx.doi.org/10.1037/pspi0000057

Supplementary Information #1

For Studies 1-5, some subjects reported on more than one sibling, potentially violating the assumption of independence. To determine whether our data indeed violated the assumption of independence, for each study we conducted and then compared the results of two different linear mixed models, one in which we included a level two subject variable and one in which we did not. The former model included a random intercept and subject factor in addition to the three fixed effects (i.e., the main effects of MPA and coresidence and their interaction); the latter had no random effect components, only the three fixed effects, and was thus, effectively, a General Linear Model (GLM). Comparing the -2LogLikelihood values of each model (reported in Table 1S) revealed that the Linear Mixed Model (LMM) that included a level two subject variable improved model fit significantly for all 5 studies. Thus, we report results from LMMs that included a random intercept and subject factor.

Table 1S

Analyses to assess violation of independence of data for Studies 1-5

-2LogLikelihood
Model without random effect / Model with random effect / Chi-square
(df) / p-value
Study 1 / 635.81 / 577.11 / 58.70 (1) / <.00001
Study 2 / 844.78 / 805.64 / 39.14 (1) / <.00001
Study 3 / 823.11 / 779.20 / 43.91 (1) / <.00001
Study 4 / 512.53 / 503.01 / 9.52 (1) / .002032
Study 5 / 517.22 / 482.70 / 34.52 (1) / <.00001

Supplementary Information #2

One of the limitations of the present set of studies is the reliance on self-report. Although we did not collect information on actual behavior, for instance, as measured in the lab, we did ask subjects in Study 1 (California) and Study 5 (Belgium) to report the number of favors they had done for their sibling in the past month (reported behavior item). We also asked subjects how willing they would be to donate a kidney to their sibling (motivation item). In the below analyses we combined data from the two cultures represented by Study 1 (Santa Barbara, California) and Study 5 (Ghent, Belgium) and conducted a linear mixed models analysis identical to that reported in Table 1 in the main text with the only difference being the dependent measure: Here we conducted analyses separately for the reported behavior item and the motivation item. As can be seen, the pattern of results found in Table 1 obtains for both types of items. We also found a significant effect of culture. Participants in Belgium reported more favors (Z-scores: M=0.13; SD=1.11; N=249) than did participants in California (favors: M=-0.11; SD=0.87; N=294). But participants in California reported a greater willingness to donate a kidney (M=0.33; SD=0.65; N=294) than did participants in Belgium (M=-0.39; SD=1.18; N=249).

Studies 1/5 (N=543)
Motivation DV:
Donating a kidney / Studies 1/5 (N=543)
Reported Behavior DV:
Favors in last month
Fixed Effects / F / Parameter Estimate
β (SE) / F / Parameter Estimate
β (SE)
Intercept
MPA x Coresidence
MPA
Coresidence duration
Culture a / 86.92***
24.42***
33.92***
131.63***
33.92*** / -0.70 (.10)
-0.04 (.02)
0.75 (.23)
0.05 (.01)
0.05 (.14) / 40.84***
14.46***
29.11***
37.33***
21.44*** / -0.75 (.10)
-0.06 (.02)
1.19 (.22)
0.04 (.01)
0.41 (.09)
Covariance parameters
Residual
Intercept varianceb / .27 (.02)
.39 (.05) / .54 (.04)
.31 (.06)
R-squared / .33 / .18

a Estimate for Belgian location provided; UCSB estimate set to zero; b Intercept variance is of the Level-2 Subject variable, which includes family from which sibling pairs were sampled; ***p≤.001, **p≤.01, * p≤.05