PS390 Chapter Notes - Chapter 11: General Linear Model, Statistical Process Control, Statistical Inference
Document Summary
Traditional statistical procedures in psych (t, f, z tests, anova, multiple regression) are based on the general linear model in statistics. The mathematical assumptions underlying these techniques are: (1) observations come from normal, bell-shaped distribution (which is just one kind of distribution) In reality, it is doubtful that strict normality of observations occurs (2) they are independently and identically distributed. Each datum is a random variable, unrelated to other variables w independent probabilities. Independence occurs when research participants are randomly assigned or randomly sampled but many studies use convenience samples that are neither. For decades, at least some psychologists have made claims to authentic scientific knowledge on the basis of flawed applications of statistics. Criticisms regarding psychology"s dependence on null significnce testing concerned the logic, capacities, limits, and applications of statistical inference. Method of null-hypothesis significance testing discourages replication / encourages one-shot research.
