Applied statistics - Chapter 9: Two-sample tests correlation

Treatment effects

Systematic differences between the two groups

Random effects (sources of error)

Measurement errors

e.g. Data entry error, measurement unreliability, etc.

Random effects (sources of error) [con’t]

Individual differences within the study

e.g. There are differences between individuals, so not all people are the same at the start of the study

Correlated groups design

When there’s a direct relationship between groups

Repeated measures

A within-subjects, pre-post design where the same subject is tested on two or more occasions

Matched-group design (matched-pairs design)

When participants are paired up on some variable that one thinks is related to the DV

Matched-group design (matched-pairs design) [con’t]

Pairs of individuals are formed based on similar scores on the third variable

Matched-group design (matched-pairs design) [III]

The individuals who have been paired are randomly assigned, one to each of two conditions/group

Benefits of repeated-measures and matched-pairs designs

• It ensures two groups are more equivalent in terms of some relevant characteristic

• Removes a source of error from measurements (data points)

• Lowers standard error of differences, which results in more power by reducing individual differences

What does a correlated-samples t test do for the difference between two means

It obtains data through either a repeated-measures or a matched-pairs design

What does the direct difference method do?

It finds the differences (D) between the scores obtained by each pair of correlated scores and treat each D as a raw score

Direct difference method (con’t)

It conducts a one-sample t test on the mean of the differences