T TESTS ( DIFFERENCES)

Design

Repeated

need less p

ie. morning vs evening

same particpants

Independant

different participants

ie. sex differences

Variables

Dependant

Independent

outcome

manipulation

Assumptions

data type

ratio

parametric

interval

independent observations

data

Normally distributed

VIOLATES ASSUMPTIONS

NON PARAMETRIC

SKEW

Negitively

Positively

low s overrep

high s overrep

tail +ive

tail -ve

homogeneity of varience

consistent diff between groups

no significant difference

variance of data to be similar in all groups

LEVENES TEST

Independent

Research Q

HYPOTHESIS

Do M & F differ in enjoyment of TV

F and M will report different enjoyment of three TV Shows

EastEnders, Football and news

TWO TAILED

Variables

independant

dependant

SEX

Enjoyment

ANALYSIS OUTPUT

HOMOGENIETY

WHERE?

INDEPENDANT SAMPLES TEST TABLE

LEVENES #

Not signifigant

(p.>.050) ASSUMPTION MET

READ numbers of the Equal variences assumed

Signifigant

Read numbers of equal variences not assumed

(p< or equal to <.050) ASSUMPTION VIOLATED

F=0.4, p= .535 #

HOW TO REPORT

T TEST STATISTIC

t, df, sig

t(10) =2.5, p=.031 #

t(df)= t statistic, p value

There is a sig differnce

Group Statistics

females =sig higher

Descriptive stats per group

males sig lower

(M=6.8, SD=1.5)

(M= 4.2, SD=2.1)

GRAPHING

BAR GRAPH

ERROR BARS

(+-)1SD

REPEATED

Research question

Does increased time to learn words improve memory performance?

Variables

Indpendant

Dependant

Time to learn

2 levels

4 minutes

2 minutes

words recalled (15)

Hypothesis

The longer time to learn the better remembered

ONE TAILED

ANALYSIS

PAIRED SAMPLES STATISTICS

t (DF) = T statistic, signifigance

IE> t(5)= 7.9, p =.001

descriptive stats

higher mean = significant increase in memory performance (t test stat) with increasing time to learn the list of words. #

interpret the means

with 2 minutes recall time (m=7, SD=1.5), 4 minutes = (M=13, SD=1.6)

CREATE A BAR GRAPH WITH ERROR BARS

repeated test

assumption that independence of observations

THIS NOT POSSIBLE IN REPEATED MEASURES

cant avoid as same participants- no individual differences

random variance reduced

DIFFERENT ASSUMPTION - SPHERICITY

ONE SAMPLE

COMPARES from single sample to REFERENCE

ANALYSIS

ie average height in uk is 164cm - is our sample above below or average

EG sample mean = 37 test value= 50

look at one sample stats table for the mean to compare to TEST VALUE in other table

one sample t test table for test value

then use this for t value df, and sig

t(9)=5.6, p<.001

height of sample smaller than the average