Significance Test - Within T-test

Question 1

Second decision

Answer 1

What type of data has the DV measured.
Nominal or ordinal- non parametric test, because not real numbers.
Interval- can use more powerful parametric test.

Question 2

Third decision

Answer 2

Between or within participants.

Question 3

Fourth decision

Answer 3

Normality.
Kolmogorov-smirnov compares data set with normally distributed set.
Don’t want significant difference.
Want p>.05.

Question 4

Parametric tests/distribution dependent tests

Answer 4

Estimating population parameters. 
More powerful; detailed analysis. 
Estimates population based on sample stats. 
Restrictions- need interval data. 
- should have normal distribution.

Question 5

Choosing test

Answer 5

Correlation parametric- pearson’s.
Correlation non- spearman’s.

Parametric differences:
Between subjects- between t test.
Within subjects- within subjects t test.

Non parametric differences:
Between subjects- Mann Whitney.
Within subjects- wilcoxon.

Question 6

Confidence intervals

Answer 6

Pointe estimate- single value to represent estimate of the parameter; little about accuracy.
Confidence intervals- two values.

Question 7

Effect size

Answer 7

Quotes alongside significance statement.
Tells how much effect IV has.
Cohen’s d is a standardised way of expressing effect size.
Not reported if no significant effect.
d= mean1-mean2/mean sd.

Question 8

Formal notation for within t test

Answer 8

t(degrees of freedom) = t value; p>0.001; tail?; effect size (d=?)

Question 9

First decision

Answer 9

Differences or correlations.
Difference- measure two groups.
Correlations- no IV or DV; relationship.