6 Oct 2016

School

Department

Course

Professor

For unlimited access to Class Notes, a Class+ subscription is required.

Lecture 4 – Analysis of Categorical

Data I

-parametric versus non parametric tests

-what happens when normality cannot be assumed?

-based on characteristics of what we’re studying

-minimum entry level data

-normal distribution (known areas in curve)

when we do not have normal distribution

-very few assumptions of characteristics of pop, less powerful than parametric

how do we measure association for nominal variables?

-categories instead of mathematical numbers

-relationships between variables

-chi-square test of independence

the basic idea of chi-square

-are two variables related to one another?

-one-way chi-square is similar to one sample t test

-looking for relationship between two categorical variables

-null hypothesis – two variables are independent

-not influencing or affecting each other

-if we reject the null, we are stating that the two variables are dependent

-chi-square independence vs dependence

the logic of chi-square

-only require frequency data

-observed and expected counts

-observed: actual data collected from your sample (actual numbers in each category of

variables in dataset)

-expected counts: info we expect to have on variables if the data was independence

-reject the null if the observed counts are sufficiently different from the expected counts

**chi-square formula x2

how do we conduct a chi-square?

-1. Calculate marginal

-2. Calculate expected counts

-3. (observed – expected)2 / expected

-4. Sum across all cells

find more resources at oneclass.com

find more resources at oneclass.com