DOI: 10.1016/J.INTELL.2012.08.002
关键词:
摘要: InWoodley (2009) an analysis was performed in which the ecological relationship between national IQ and percentage of cousin marriages (consanguinity) investigated. Bivariate correlations revealed a strong association two measures (r=− .61; Pb .01; N=72), stability significance robust to use number different procedures, including data quality controls, trichotomization partial correlation analysis. Multiple Regressions were also used three additional independent variables considered (the UN's Education Index, Economist Intelligence Unit's Index Democracy GDP per capita). It reported at time based on conducted Excel that consanguinity zero magnitude predictor when other controlled, however recently reanalyzed using SPSS (a much more sophisticated stats package doesn't require manual standardization variables), somewhat set results obtained. Presented Table 1 is overview both sets findings. In Woodley log-transformation employed, skewness this generally inappropriate as all but one exhibited insufficient warrant (skewness statistic −1). Only education marginal (−1.6), log transformation reduced (− .34), it boosted kurtosis substantially (2.31 6.43). Furthermore none assumptions multiple regression actually (Allison, 1998, p. 130). The re-analysis (using non-log transformed variables) indicates basic pattern inWoodley remains unchanged, variable still non-significant once controlling are taken into consideration. This means conclusion — ergo whilst bivariate