Lecture 8B: Comparing Means T-Test

Purpose

To compare means between 2 groups

u2-u1= 0

Overlapping confidence intervals: population mean is not significantly difference
Different in sample

T-Test

It is right because of the Cumbersome method whereby you calculate several CI

T-distribution: looks like a normal distribution so long as you have a wide range of data

Test on 2 group means: is the difference 0 or not

Test whether difference beween group means is statistically different

2 different groups: repeated test for the same observation, and 2 test for different tes

Formula: check lecture slides

Independent sample: ask one group of people independent from the other group of peeps


Dependent: asking the same peeps

Dependent sampes share characteristics that affect the variace

Dependent sample formula: check lecture slide
Independent sample formula: check lecture slide

Dependent T test assumptions:


Variables measured at interval/ratio level

Sampling distribution is normally distrubted
Parametric test

N>30, variables not clearly non-normally distrubuted

Independent T-test assumptions:

Variables measured at interval/ratio level

Sampling distribution is normally distrubted
Parametric test

N>30, variables not clearly non-normally distrubuted

Homegeneity of Variance: variances of both samples are the same

Levene's Test

Independent T-test: variances assumed equal

H0= variances equal

Outcome Levene's test: not significant means variances equal
Sig. > 0.5
Look up on SPSS and hull hyothesis is true

Significant: variances are not equal Sig < 0.05
Look down on SPSS and null hypothesis is false

Effect Size

Cohen's d for dependent (paired sample)
You compare the difference to standard deviation of difference

0.2 is small and 0.5 is medium and 0.8 is large

Big: it is significant
Small: it is not significant

Independent : which S do you pick, you pick a control group variance (as a baseline)

Pooled variance: variance of one group and another group