  Other's On-Line Resources (Chapter 9)

 Small rs Can Be Significant Description: This interactive on-line resource allows you to quickly determine the smallest r that will ne statistically significant for any given sample size. All you do is enter the value of n; you will then be given the critical values for the .05, .01, and .001 levels of significance. What to Do: Click on the colored title of this on-line resource: "Small rs Can Be Significant." Enter, in the "Sample size (n)" box, the number 10. Click on the gray bar that says "Minimum coefficient." After a few seconds, the critical values (for alphas of .05, .01, and .001) will appear as "Results" beneath the gray bar you just clicked. Each critical value will be referred to as the "minimum r significant at p=alpha." Try to memorize the size of r that would have p = .05. Change the value of the sample size from 10 to 25, and then click the "minimum coefficient" bar. Compare this critical value with the one you saw when the sample size was 10. The comparison you made of the results in Steps 4 and 5 should show you that there's an inverse relationship, at any given alpha level, between the critical value of r and the sample size. If n is large enough, even a small r will turn out to be statistically significant. Sky Huck's Puzzle Question: Can you show that an r of 0.012 would be significant at the .05 level of significance if n is large enough? Comparing 2 Correlations Description: This interactive resource allows you to see if there is a statistically significant difference between two correlation coefficients. By exerting control over (1) the size of the difference between the two rs and (2) the sample sizes, you'll be able to see that large sample sizes can turn a small difference into something "significant." What to Do: Click on the colored title of this on-line resource: "Comparing 2 Correlations." Make 2 private decisions: (a) your level of significance and (b) whether you want to do a one-tailed test or a two-tailed test. Enter .50 and .55 as the r values for Samples A and B. Enter 100 for each sample size. Click "Calculate" and then look to see if you have a statistically significant finding; this will be the case if the computed p is smaller that the alpha level you selected in Step #2. If you don't have a significant finding, repeat Steps #4 and #5 again and again (each time increasing each n by 100) until you finally can cross over into "The Wonderful Land of Significance." Sky Huck's Puzzle Question: What do you think will happen to the computed value of p if you (1) set each n equal to 100, (2) make the two values of r different by .20, and (3) change the "location" of your two r values on the "correlation continuum" (that extends from -1.00 to +1.00)?