0

I'm confused with statsmodels API of the TTestIndPower class.

In the plot_power function the number of observations plotted are the total number of samples or the number of samples of each group?

Emanuel
  • 422
  • 7
  • 12
  • From the comment they have "If the second argument is nobs1, then the number of observations in the plot are those for the first sample." I would assume it's the total number of samples but this would go against the other functions of the class, power and solve_power – Emanuel Jul 13 '21 at 16:07

1 Answers1

0

It's per group.

This is the power_plot I generated with statsmodels

from statsmodels.stats.power import TTestIndPower

panalysis = TTestIndPower()

panalysis.plot_power(
    dep_var="nobs",
    nobs=np.arange(5, 101),
    effect_size=np.arange(0.5, 1.5, .2),
    alpha=0.01,
    ax=ax,
)

enter image description here

And this is the confirmation with R for an effect size (es) of 0.7

> library(pwr)
> pwr.t.test(power = 0.8, sig.level = 0.01, d = 0.7)

 Two-sample t test power calculation 

          n = 49.35336
          d = 0.7
  sig.level = 0.01
      power = 0.8
  alternative = two.sided

 NOTE: n is number in *each* group
Emanuel
  • 422
  • 7
  • 12