Table 3.

Type 1 error and power for the estimated RR using three control sampling methods

True RRProportion exposedType 1 error or power*
Full cohort designCase-control designs
Method 1: incidence density sampling without replacementMethod 2: incidence density sampling with replacementMethod 3: pure control sampling
1.0010%5%5%3%5%
30%5%5%3%6%
50%5%5%3%5%
1.5010%72%40%38%50%
30%97%74%75%87%
50%98%83%84%92%
2.0010%>99%88%89%95%
30%>99%>99%>99%>99%
50%>99%>99%>99%>99%
  • * Under the null of a true RR = 1, the type 1 error is given; under the alternatives of a true RR = 1.5 or 2, the power is given.