I've just run a series of point biserial correlation tests in R between whether or not characters were assigned national identities, and attributions given to their behaviours - results shown in the attached screen shot file. Whilst the results show no strong correlations, it is the two middle p-values that have thrown me as I've never had a result so low before (n = 483)
Any guidance or insight would be greatly appreciated.
Hello Alex,
A Pearson correlation of +/- .3 would be statistically significant at the .05 level with 44 cases. So, an N of 483 just yields a much lower probability (that population value is zero, under the null hypothesis). The computed values appear to be in the correct ballpark (the rounding of r to a single digit makes it impossible to be exact).
Good luck with your work.
This is equivalent to an independent t test (the equivalent between point-biserial and r is well known). So would it help to realise this is a standardised mean difference of about .63 and (in this sample size assuming roughly equal groups) and a t statistic of about 7?
2.272e-11 means 2.272 times 10 to the power of -11. This is a very small number, which is significant at the 0.001 level, which I would interpret as very strong evidence of a non-zero correlation.
Other than the Internal Attribution r value all other "r" values show no correlation. The "r" value of -.14 show non significance even with a high n as is for this study.
- Badges
- Science topic
- Similar topics
- Psychology