Week 3: Correlations and Chi-Square

3 types of variables for correlations

cont vs cont, cont vs ordinal, ordinal vs ordinal

covarianace

how much each score deviates from mean

pearsons correlation coefficient is represented as

r

pearsons us used for

parametric testing to assess linear relationship between 2 variables

characteristics of both variables in pearsons

continuous and normally distributed

null for pearsons

no linear association between teh 2

if p is less than 0.05

reject the null and conclude there is significant coorelation

interpreting r

closer to r is to 1 or -1 (only range), the stronger the association

positive r

positive relationship (both increase, both decrease)

negative r

negative relationship( one up, one down)

what is the effect size for pearsons

r

nonparametric coorelation test

spearman rho (rs)

spearman Rho function

ranks data

what contstitutes using spearman

if at least 1 variable does not follow normal distribution or 1 or both are ordinal

correlation does not indicate

causality (r only is for linear relationships)

chi squared assesses

2 categorical variables (establishes equivalence, primary research question)

in a 2x2 table, how many variables are present

4

pearsons chi square statistic (x2)

tests whether or not the 2 categorical variables are related, compares frequencies yu observed vs expected

null hypotheses for pearsons chi squared

variables are independent of one another (not related)

dof for chi squared

(#rows-1) x (#columns-1)

for 2x2 tables, dof is

1

fisher test

used instead of chi square when any cells have expected counts less than 5 (most common in 2x2 tables)

mcnemar test

used in 2x2 tables when they are correlated samples (within subjects/pre-post design where DV is categorical)

3 main measures of effect size for chi square

phi, odds ratio, cramers V

phi

only recommended for 2x2 tables, max is 1

odds ratio

represents likelihood of event occuring, only for 2x2

cramers v

2x2 table or larger, max is 1

odds ratio calculation EXAMPLE

odds walking after A/Odds walking after B

-number you get is the "times" they are more likely to do something