PSYC 305 Lecture Notes - Lecture 8: Null Hypothesis, Normal Distribution, Chi-Squared Distribution
Document Summary
It is possible for this to happen, in theory: when using sheffe"s test, where things are too conservative where overall null hypothesis is rejected but you do not know where the differences lie. One-way anova: assumptions: the population distribution of the dv is normal within each group, homogeneity of variance (variance within each group is equal amongst the groups, independence of observations. The null hypothesis and assumptions are what determine the shape of the sampling distribution (z, t, f, chi square, etc. ) If assumptions do not hold, or if there are gross violations, it may mean that you are continually making the wrong conclusion. Distributional assumptions: has to do with the scores within a population: best thing to do is look at what is going on in the samples to see what"s up with the population in terms of assumption violation. Descriptive/inferential statistics: tests for skewness, k-s/shapiro-wilk tests.