PSY 350 Lecture Notes - Lecture 48: Psy, Regression Analysis, Standard Scale
Document Summary
The significance test for a correlation is very sensitive to sample size. Adding or subtracting just a couple of people can mean the difference between rejecting and not rejecting h0. Consider your sample size when you interpret the significance of a correlation. If you have a small sample, even a substantial correlation won"t be significant. If you have a large sample, even a tiny correlation will be significant. *need to think about both significance and effect size. Finding confidence intervals around your correlations helps you consider significance and effect size at the same time. Hand calculations are a little complicated, but any good software can estimate these for you. Remember that a confidence interval tells you the plausible range of values for a statistic. If you calculate a 95% confidence interval, 95% of the plausible values are within that interval. Larger samples give you narrower (aka, more precise) confidence intervals. Already on a standard scale of -1 to 1.