6. Analysing Continuous and Categorical Data Flashcards
What are the different type of t-tests and what do they measure?
- Single mean: 1-sample t-test
- Difference between 2 means: 2-sample t-test
- Difference between paired measures: Paired t-test
What is the null hypothesis of the One Sample T-Test?
Is that the population mean = fixed hypothesised value.
What makes a one sample t test valid?
- Approx. Normal Distribution
2. Sample size is 20 or more
What is the null hypothesis of the two sample t test?
Two population means are equal.
The mean difference = 0.
What makes a two sample t test valid?
Valid if…
- Samples are approximately normally distributed
- At least 20 in each group
- Standard deviations are approximately equal
What is the null hypothesis of a paired t test?
Within-pair differences are normally distributed around zero.
Tests whether the mean within a pair is larger than would be expected to have occurred by chance.
What is the null hypothesis of ANOVA?
All of the samples that are randomly selected from a given population all have the same mean.
It simultaneously compares the means of more than 2 groups
What are the requirements for ANOVA to be valid?
- 20 samples
- All samples approx. normally distributed
- Approximately equal variances
What is the one sample t test non-parametric cousin?
Wilcoxon sign test
What is the two sample t test non-parametric cousin?
Mann Whitney U test
What is the paired t test non-parametric cousin?
Wilcoxon paired test
What is the ANOVA test non-parametric cousin?
Kruskall Wallis Anova
What are we looking for when asking whether correct test was allocated.
- Normal distribution
- SD not twice one another
- Approx equal variances (ANOVA)
What are the remits of parametric?
Has assumptions that must be met for results to be valid
Uses more of the data so more powerful
What are the remits of non-parametric?
Assumes nothing of the data
Always valid but less powerful than its parametric equivalent
If parametric and nonparametric disagree, nonparametric should be trusted
