PS280 Chapter Notes - Chapter 4: Smoking And Pregnancy, Longitudinal Study, Statistical Significance
Document Summary
Non-invasive: behaviour not manipulated, just measured quantitatively and then analyzed statistically. Resulting correlation coefficient statistic describing relationship b/w two variables is analyzed by a test of statistical significance to determine whether it is likely that the observed relationship could have occurred simply by chance. Psychologists generally interested in correlation coefficients > or = +-. 30. Frequently used when experimental manipulation is impossible or unethical (ex studying effects of maternal smoking on developing fetus) Significant correlation interpretations: (1) variable a caused variable b, (2) b caused a, (3) c is responsible for both a and b. One way to reduce ambiguity regarding direction of relationship is to conduct longitudinal studies which permit examination of early factors that precede onset of a disorder. Longitudinal studies are very demanding in terms of resources, especially when disorder or outcome of interest has low base rate in general population.