STAT 2040 Lecture Notes - Lecture 13: Type I And Type Ii Errors, Count Data, Null Hypothesis
1 Apr 2015
School
Department
Course
Professor
Document Summary
*involves only hypothesis tests, not conf" int" eg. x2tests for one-way tables. In one way, observations are classified according to a single categorical variable. *if the observed counts are very different from the expected counts, the test statistic will be large. *@ given df, the larger the value of the test statistic, the greater the evidence against the null hypothesis. p-value: summarizes the strength of the evidence against the null hypothesis (area to the right of the observed test statistic) *one degree of freedom is lost for every parameter estimated from the data. One way h0: the cell probabilities are equal to some values of interest. Testing goodness-of-fit for specific probability distribution binomial, poisson, or normal distributions. With a small p-value (close to 0), we have very strong evidence against null hypth". Two way h0: there is no relationship between the row and column variables. *multiple samples see if the distribution of the response variable differed between the groups.
