Table 5

Effect of measurement error on sample size required for detecting a significant difference in cord blood levels of total estradiol and SHBG between two populationsa

Mean in population 1Mean in population 2% differencebRequired sample sizec
No measurement errordWith measurement errore
Total estradiol (ng/dl)
 1500172515154193
 150018002091113
 15001875256076
SHBG (nmol/l)
 4450.6158487
 4452.8205051
 4455.0253334
  • a Three sets of hypothetical means are set such that the true means in population 2 are 15, 20, and 25% higher than the true means in population 1, respectively.

  • b 100 × [(mean in population 2) − (mean in population 1)]/(mean in population 1).

  • c Sample size for each population (equal sample size) was calculated for α = 0.05 (two-sided) and 80% power.

  • d Variances used for total estradiol and SHBG were 0.0362 and 0.0198, respectively, based on data on the logarithmic scale (base 10).

  • e Variances used for total estradiol and SHBG were 0.0453 and 0.0204, respectively, based on data on the logarithmic scale (base 10).