Chapter 22 Multi-Level Analysis (625-635) Flashcards
Analysis of Variance (ANOVA) (3)
- Statistical technique that
- Compares variances within and between samples
- In order to estimate the significance of differences between a set of means
Capitalising on Chance (4)
- Making too many tests
- With alpha set at .05
- On the same data,
- Hence increasing the likelihood of a Type I error
Within Groups Sum of Squares (4)
- Sum of squares
- of deviations of scores
- around their sample means.
- Also: error SS.
Within Groups Variance (3)
- Total variance of scores
- around sample means.
- Also. error variance.
Between Groups Variance (2)
- Variance of sample means
2. around grand mean.
Grand Mean (2)
- Mean of all scores in a data set,
2. irrespective of conditions or groups
Total Variance (2)
- Variance of all scores in a set
2. around their grand mean
Error Variance (3)
- Total variance of all scores
- from their group means.
- Also: within groups variance.
F test/ratio (1)
- Statistic giving ratio of between groups to within groups variance
Sum of Squares (1)
- Addition of the squares of deviations around a mean
Variance Ratio Test (1)
- Full name for the test producing the F statistic
Mean Sum of Squares (1)
- Sum of squares divided by df
Between Group Sum of Squares (1)
- Sum of squares of deviations of sample means from the grand mean
Error Sum of Squares (2)
- Sum of squares of deviations of each score from its own group mean.
- Also: within group SS.
Pairwise Comparison (2)
- Comparison of just two means
2. From a set of means
Post Hoc Comparisons/Tests (2)
- Tests between means, or groups of means
2. Conducted after inspection of data from initial analysis
A Priori Comparisons/Planned Comparisons (2)
- Tests of differences between selected means, or sets of means,
- Which, from prior theory, were predicted to differ
Family-Wise Error Rate (3)
- The probability of making at least one Type I error
- In all the tests made on a set of data,
- Assuming H0 is true
Error Rate per Comparison (3)
- Given the significance level set,
- The likelihood of a Type I error in each test made on the data
- If H0 is true
Bonferroni t Tests (3)
- Procedure for testing means pairwise,
- Which involves raising the critical values of t
- To lower the family-wise error rate
Linear Contrasts (2)
- Procedure for testing between individual pairs of means or combinations of means,
- a priori (i.e. predicted)
Linear Coefficients (2)
- Values to be entered into an equation
2. For calculating linear contrasts
Newman-Keuls Post Hoc Test (2)
- Post hoc test of means pairwise
2. Safe so long as number of means is relatively low
Tukey’s (HSD) Post Hoc Test (3)
- Post hoc test of all possible pairwise combinations
- Appropriate analysis choice with a large number of means
- Considered conservative.