Table of sample sizes required for tests of non-zero Kendall, Spearman and Pearson correlations
We assume 90% power and a Type I error of 5%
Null hypothesis : correlation = 0;
Alternative: correlation = non-zero value
If you know the sign of the non-zero correlation the test is one-tailed otherwise it is two-tailed.
Computations using methods in Kraemer, HC & Thiemann, S (1987) How Many Subjects? Statistical Power Analysis in Research. Sage:London.
Power calculator also available for the Pearson correlation.
- Dunlap WP and Myers L (1997) show that for a Pearson correlation, r, 8/$$r^text{2}$$ gives a total sample size with at least 80% power.
Correlations of 0.1, 0.3 and 0.5 correspond to small, medium and high correlations using rules of thumb.
|
1-tail |
2-tail |
||||||||||
correlation |
Kendall |
Spearman |
Pearson |
Kendall |
Spearman |
Pearson |
||||||
0.1 |
1041 |
1013 |
854 |
1277 |
1111 |
1047 |
||||||
0.2 |
224 |
250 |
212 |
274 |
307 |
259 |
||||||
0.3 |
106 |
107 |
93 |
129 |
130 |
113 |
||||||
0.4 |
58 |
62 |
51 |
70 |
75 |
62 |
||||||
0.5 |
37 |
39 |
32 |
44 |
46 |
38 |
||||||
0.6 |
25 |
26 |
21 |
29 |
30 |
25 |
||||||
0.7 |
18 |
19 |
15 |
21 |
21 |
17 |
||||||
0.8 |
13 |
<14 |
<10 |
15 |
15 |
12 |
||||||
0.9 |
9 |
<10 |
<10 |
11 |
11 |
<10 |
||||||
Reference
Dunlap WP and Myers L (1997) Approximating Power for significance tests with one degree of freedom. Psychological Methods 2(2) 186-191.