On Testing a Trend in Means in Oneway Layout

A test for a trend in means in oneway layout is discussed with particular emphasis on sample size determination. The proposed method amounts to formulating a linear contrast among sample means and testing it for significance. The resulting test statistic is reduced to a test by Barlow et al. (1972) When sample sizes are equal and to the Armitage‐Kendall correlation test for a trend in proportions when the response variable is dichotomous. The problem of sample size determination is discussed and illustrated with numerical examples.