T-tests

Covers T-tests, Parametric tests and Non Parametric tests

What is a T-test

An inferential stats test to see if there is a difference between two groups

What type of experimental design uses T-tests

Experimental

What is the test output of a T-test

A test statistic and p-value

What does a T-test show

the difference between groups over the variability within groups

What is the observed test statistic in a t test

t - the size of difference between groups

What does a significant result look like for a T-test

A larger T statistic

What two factors do T-tests look at

Cause and effect relationships and variability

What is variability

are the scores in each group consistent W

Why does variability matter

If there is too much noise in a data set difference might just be random

What group difference should be identified

Between groups not within groups

How can you better observe effects of manipulation

Increase between group variance or decrease within group variance

How should T-test data be visualised

Boxplots or bar graphs

What data does a boxplot show

Median, quartiles and variability

What data does a bar graph show

mean, standard deviation and error bars

What are the benefits to a T test

• Identify differences between groups/conditions

• Test hypotheses

• Experimental

• Cause and Effect

• Make predictions about group differences

• Starting point for other experiments

What are the 2 types of T-tests

Parametric and Non parametric

What type of data does the IV need to be for a T-test

Categorical or nominal (ordinal if it can be meaningfully split)

What type of data does the DV need to be for a T-test

Continuous data, ratio or interval

What are the assumptions for a T-test

• Independent observations

• Normality

• Homogeneity of Variance

• Type of data/outliers

What is Homogeneity of Variance

The variance of data should be approximately equal in both groups

How can Homogeneity of Variance be measured

Using Levene’s test

What test should be used if Levene’s test fails

Welch’s T-test

What are Parametric T-tests

used when assumptions are met

What are the 3 types of T-tests

One samples, paired samples and independent samples

What is a one samples t test

One group of participants takes part in the study in one condition

What happens to one samples data

The mean is compared to a known, referenced or hypothesised population

What is a paired samples t test

One group of participants take part in 2 conditions

What happens to paired samples data

Compare the mean of the group when they do condition A against condition B

What is an independent samples t test

2 different groups take part in different conditions

What happens to independent samples data

The means of both groups are compared

What info has to be included in a T-test write up

• Test type (type of t-test)

• Variables (IV + levels and DV)

• Test statistic (t - for parametric)

• Degrees of freedom

• P-value

• Significance

• Direction

• Descriptive (mean/median)

What order is a parametric T-test written in

t(df) = xx.xx, p = .xxx, d = XX.XX

What does the d stand for in the parametric write up

Cohen’s d for effect size

What are non-parametric t tests used for

When assumptions are not met

What are the 2 non parametric t tests

Wilcoxon and Mann Whitney U

When is a Wilcoxon test used

One samples and paired samples T-tests

When is a Mann Whitney U test used

Independent samples t-tests

What is the test statistic letter for the Wilcoxon test

W

What is the test statistic letter for the Mann Whitney U test

U

What is the way to write up a non parametric t-test

W or U = XX.XX, p = .xxx, r_b = XX.XX

What does r_b mean

Rank-biserial for effect size

What changes about the data when using a non parametric t test

Instead of raw scores it is ranked e.g. 1-10