0

I am trying to run a one way ANOVA on some data. One factor, 4 levels. I only have mean, SD, and sample sizes to work with. Is there a way to do this in R?

  • 1
    If you have the sd's then you have the variances. You don't, however, have the variance of the total group, which would be necessary to give a correct answer. (This is actually not a coding question.) – DWin Feb 20 '15 at 05:12
  • 1
    By "sample sizes", do you mean that you have $n_j$ for each level of the factor? Also, can you clarify "mean, SD"? Do you have the marginal (total) mean & SD, the means & SDs for each level, or both? – gung - Reinstate Monica Feb 20 '15 at 21:59
  • This is an interesting question, but an answer at the linked duplicate gives a strategy that works in any package. R also has some packages that can work with summary data, but the approach given at the link is so fast and simple to do in R that it's hardly worth looking for them. – Glen_b Feb 21 '15 at 00:49
  • Briefly, for each group, generate a random sample of the relevant size (the distribution doesn't actually matter for obtaining estimates and p-values and such). Standardize it to mean 0 var 1 (scale in R), multiply by the group std deviation, add the group mean. Conduct ANOVA as usual. (Don't bother with diagnostics, naturally.) – Glen_b Feb 21 '15 at 00:58

0 Answers0