Short Answer Practice Questions Flashcards
Write the formula for the independent samples t-test
t = “x bar 1” - “x bar 2” / standard error
Write the generic t-test formula
t = sample data - hypothesized pop. parameter / estimated standard error
List the three properties of a dataset that contribute to the size of an obtained t-test statistic in the independent samples t-test
- Sample size
- Dispersion of means (sd and SE)
- Effect size
Write the formula for the paired samples t-test
t = “D bar” - “Mu D” / “S D bar”
**Can remove “Mu D” because this is the mean difference scores of population which is hypothesized to be 0
Write the formula for Cohen’s d for the independent samples t-test
d”s” = “x bar 1” - “x bar 2” / pooled standard deviation
Write the formula for Cohen’s d for the paired samples t-test
d”av” = “D bar” / “Avg. S”
List the three factors that influence power
- Alpha
- Sample size
- Effect size
List the two factors that influence effect size (Cohen’s d)
- Difference in means
2. Standard deviation
Write the formula for the ANOVA F-ratio
F = between variability/ within variability
Write the formula for the ANOVA F-ratio with regard to sources of variance
F = treatment effects + individual diffs. + experimental error / individual diffs. + experimental error
What are the two analyses that make up analysis of variance for ANOVA?
- Sums of squares (SS) analysis
2. Degrees of freedom (df) analysis
What is the term for variance in ANOVA? What is the formula?
Mean squares
MS = SS/df
Write the formula for the ANOVA F-ratio expressed using mean squares
F = MS”between” / MS”within”
List the 3 statistical assumptions
- Groups are independent of each other
- Sample means are normally distributed
- Homogeneity of variance between groups
What are the 2 effect size indices for ANOVA and their corresponding formula?
Eta squared and Cohen’s f
Eta squared = SSbetween / SSwithin
Cohen’s f = squareroot of SSeffect / SSerror
Contrast standard deviation and standard error
Standard deviation is a measure of variability within a sample - the variability between each mean score and the actual mean
Standard error is a measure of variability between samples in a given population - the variability between each sample mean
What is an omnibus test?
An omnibus test is one that evaluates a general research question. The ANOVA f test is considered an omnibus test because it provides vague results (a mean difference occurs but not where or how many).
Why does ANOVA need follow-up tests?
Since ANOVA is an omnibus test, it needs follow-up tests to identify a more specific research question (to figure out where the mean difference occurs and if there is more than one).
Explain why power decreases when alpha becomes more strict
Alpha is the likelihood of finding an effect when no such effect exists, and as alpha decreases and becomes more strict the likelihood of committing this type I error decreases. Essentially, the likelihood of finding an effect decreases and considering that power is the likelihood of finding an existing effect, it will also decrease.
What is family-wise error? What tests control for family-wise error?
Family-wise error is the cumulative likelihood of making a type I error. Essentially type I error is inflated.
Both Tukey’s HSD and Bonferroni adjustment control for family-wise error.
When can using raw effect size be problematic?
- When measurement units are difficult to interpret
2. When measurement units differ between samples being compared