Comparing Means
PAIRED SAMPLES T-TEST
Assumptions
● N > 30 (central limit theorem)
● pairs of observations (dependent)
● DV: interval or ratio
● IV: categorical level
Null hypothesis
In the population there is no difference between the mean/average … scores on …(condition
1) and …(condition 2) (compare means within ONE group).
Example: In the population, there is no difference between the mean hygiene scores on day
1 and day 3.
Effect size
Cohen’s d (Mean/Std.Deviation)
SPSS
Analyze – Compare Means – Paired samples t-test (Options: 95% CI)
, INDEPENDENT SAMPLES T-TEST
Assumptions
● N > 30 for both groups (central limit theorem)
● IV: categorical consisting of 2 independent groups
(not more than 2 groups)
● DV: interval or ratio level
● The variances in both groups are equal
Null hypothesis
In the population, the mean (DV) for (group1) is not different from the mean (DV) for
(group2).
Equality of variances
Levene’s test
If not significant- ‘’equal variances assumed’’.
If significant- ‘’equal variances cannot be assumed’’.
Effect size
Cohen’s d
SPSS
Analyze – Compare Means:
● Test Variables: select numeric variable
● Grouping Variable: 2 groups we want to compare (example: sex) – Define Groups –
use specified values (example: group1: 1=female, group 2: 0=male) – Options (95%)
ONE WAY ANOVA
Assumptions
● IV: categorical (more than 2 independent groups)
● DV: continuous
● Groups are of approximately equal size
(largest group not more than 10% larger than smallest group)
OR
● equal variances for IV in the population
(Levene’s F‐test: NOT check when if N equal for all groups)
Null hypothesis
In the population, the mean DV does not differ between groups (g1, g2, g3,..)
Post Hoc tests
for variables with more than 2 groups
Effect size
Eta squared (SSbetween/SStotal) <.3 weak, .3-.5 moderate, > .5 strong.
SPSS
Analyze-Compare means- One way Anova
● Post Hoc-Bonferroni
● Options: Descriptives, Means plot, Homogeneity of variances
PAIRED SAMPLES T-TEST
Assumptions
● N > 30 (central limit theorem)
● pairs of observations (dependent)
● DV: interval or ratio
● IV: categorical level
Null hypothesis
In the population there is no difference between the mean/average … scores on …(condition
1) and …(condition 2) (compare means within ONE group).
Example: In the population, there is no difference between the mean hygiene scores on day
1 and day 3.
Effect size
Cohen’s d (Mean/Std.Deviation)
SPSS
Analyze – Compare Means – Paired samples t-test (Options: 95% CI)
, INDEPENDENT SAMPLES T-TEST
Assumptions
● N > 30 for both groups (central limit theorem)
● IV: categorical consisting of 2 independent groups
(not more than 2 groups)
● DV: interval or ratio level
● The variances in both groups are equal
Null hypothesis
In the population, the mean (DV) for (group1) is not different from the mean (DV) for
(group2).
Equality of variances
Levene’s test
If not significant- ‘’equal variances assumed’’.
If significant- ‘’equal variances cannot be assumed’’.
Effect size
Cohen’s d
SPSS
Analyze – Compare Means:
● Test Variables: select numeric variable
● Grouping Variable: 2 groups we want to compare (example: sex) – Define Groups –
use specified values (example: group1: 1=female, group 2: 0=male) – Options (95%)
ONE WAY ANOVA
Assumptions
● IV: categorical (more than 2 independent groups)
● DV: continuous
● Groups are of approximately equal size
(largest group not more than 10% larger than smallest group)
OR
● equal variances for IV in the population
(Levene’s F‐test: NOT check when if N equal for all groups)
Null hypothesis
In the population, the mean DV does not differ between groups (g1, g2, g3,..)
Post Hoc tests
for variables with more than 2 groups
Effect size
Eta squared (SSbetween/SStotal) <.3 weak, .3-.5 moderate, > .5 strong.
SPSS
Analyze-Compare means- One way Anova
● Post Hoc-Bonferroni
● Options: Descriptives, Means plot, Homogeneity of variances
