Answers are grounded in this paper and cite their sources.
Some theoretical results on the effect of non-normality on the t test of significance, though it is stressed that they are incomplete and not perhaps of much quantitative value, nevertheless tend to agree with the result of sampling investigations in showing that for moderate departures from normality this test may still be used with confidence, particularly for testing differences in means of equal numbers of observations.
No takes yet. Share an insight, caveat, or question.
M. S. Bartlett (1935) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: