Paired Samples T-Test

What did Gosset develop?

the t test and t distribution in 1908

What did Gosset find?

that existing stat techniques using large samples were not useful for small sample sizes he found

What is a t test?

a tool for eval means of 1 or 2 pops using hyp testing

What may be a tt be used to eval? (3)

whether a single grp differs from known value (1 sample tt)

whether 2 grps differ from e/o (idp samples tt)

whether theres a sig dfr in paired measurements (paired s tt)

What does a t statistic use?

sample variance tf we must know the sample mean

What will degrees of freedom determine?

how well the t stat approximates normal distb

Desc the graph abt how to test for dr bt 2 related grps

When are parametric tests preferred?

in terms of their power to provide evidence for sig dfr

What do parametric tests require? (5)

normality

equal variance

independence

no outlier

at least interval level data

What is normality?

data in each grp should be normally distributed

What is equal variance?

data in each grp should have approx equal variance

What is independence?

data in each grp shld be randomly and idpendently samples from the pop

What do paired sample t tests use?

RM/within grps design - is there a dfr in grp bt 2 points in time

What is a strength and disadv of paired s tts?

PVs reduced and less pps needed

order effects and practice effects

What is Ho?

null hyp assumes that the true mean dfr bt the paired samples is 0

What is H1?

all hyp assumes that theres an effect of IV on DV

What are the assumptions for paired samples t test? (4)

DV shld be measured on a continuous scale

IV shld consist of 2 categorical related/matched grps

dfr bt mp shld be approx normally distb

shld be no sig outliers in dfr bt 2 grps

What is the formula for paired samples tt?

How do you calculate degrees of freedom?

n-1

What does effects size? (Cohens d) show?

0.2 = small effect

0.5 = moderate

0.8 = large