Term
| As sample size increases, a chance of finding an effect _______. |
|
Definition
|
|
Term
|
Definition
| the ability to detect effects that are really there |
|
|
Term
| Increasing sample size will ______ power. |
|
Definition
|
|
Term
| Increasing alpha will _______ power. |
|
Definition
|
|
Term
| Increasing effect size will ______ power. |
|
Definition
|
|
Term
| Increasing beta will _______ power. |
|
Definition
|
|
Term
| With a large effect size, you need _____ people to detect an effect. |
|
Definition
|
|
Term
As alpha is increased, type _____ error decreased.
As beta is increased, type _____ error is decreased |
|
Definition
As alpha is increased, type II error (β) decreased
As beta is increased, type I error (α) is decreased |
|
|
Term
| For Cohen's D, what are the ranges of effect sizes? |
|
Definition
less than .20 - small effect size
around .50 - medium effect size
over .80 - large effect size |
|
|
Term
| Positive skew: indicated by a positive number, data trails off to the _____ |
|
Definition
|
|
Term
| ________ skew: indicated by a negative number, data trails off to the right |
|
Definition
|
|
Term
| What is the difference between a Leptokurtic and a Platokurtic distribution? |
|
Definition
Leptokurtic kurtosis: data is clustered extremely around the mean
Platokurtic kurtosis: data is flattened out and spread out |
|
|
Term
| When Levene's Test is Significant what should you do and why? |
|
Definition
Transform data and use Tukey's HSD (less conservative)
Use Welch's & Games-Howell (more conservative)
-These do not require the assumption of equal variances |
|
|
Term
What will
-Transforming data and use Tukey's HSD
-Using Welch's & Games-Howel
do to error? |
|
Definition
| They will increase type II error |
|
|
Term
|
Definition
| you have a 14% chance of committing a type 1 error in at least one comparison |
|
|
Term
| How does Tukey's HSD control for error, and what are the consequences of this? |
|
Definition
Tukey’s HSD controls for type I error because it adjusts the original
With Tukey’s you lose power because |
|
|
Term
| How many Orthogonal contrasts can you do? |
|
Definition
| df = number of contrasts you can do |
|
|
Term
| What kind of test should your run when you want to make a complex comparison? |
|
Definition
|
|
Term
| What kind of test should your run when you want to do small number of comparisons? |
|
Definition
|
|
Term
| What kind of test should your run when you need to make a large number of comparisons? |
|
Definition
| regular contrast and use Tukey's HSD |
|
|
Term
| If the lines of an interaction are crossed, should you interpret the main effects? |
|
Definition
you should not interpret the main effects because the main effect is dependent on the interaction
the means differ based on the conditions |
|
|
Term
| What should you do if Mauchley's Test of Sphericity is not significant? |
|
Definition
|
|
Term
| What does it mean when Mauchley's Test of Sphericity is significant? What should you do? |
|
Definition
The assumption of Sphericity (Circularity) was not met.
you should use Multivariate results and (report Lambda)
or use Greenhouse-Geisser (less conservative) |
|
|
Term
| What test do you use to test the assumption of equal Covariance Matrices? |
|
Definition
|
|
Term
| What test should you use to test the assumption of Homogeneity of Variances? |
|
Definition
|
|
Term
| What test should you use to test the assumption of Sphericity of Covariance Matrix? |
|
Definition
|
|
Term
|
Definition
| a co-variate controls for something to remove it’s effect |
|
|
Term
| What are the assumption of ANCOVA? (2) |
|
Definition
homogeneity of regression
linearly related |
|
|
Term
| power is _____ when you add a covariate |
|
Definition
|
|
Term
| effect size is _____ when you add a covariate |
|
Definition
increased
it is now explained by the covariate |
|
|
Term
| When should you use Chi square? |
|
Definition
when your data is:
nominal
ordinal
interval |
|
|
Term
| For a 2x2 Chi square, look at ______ to see the significance |
|
Definition
|
|
Term
| What does it mean when the risk estimate (Chi square) for x is 2.33? |
|
Definition
| if the risk estimate for x is 2.33, then that cell is 2.33 times more likely to also be y |
|
|
Term
|
Definition
| Stating an effect exists when it actually doesn’t exist |
|
|
Term
|
Definition
| Stating an effect doesn’t exist when it actually does exist. |
|
|
Term
| When doing a transformation, you want your skewness and kurtosis values to be close to _______. Why? |
|
Definition
0
the close to 0, the more normal the distribution |
|
|
Term
| For positively skewed data, what transformation do you use? Why? |
|
Definition
use natural log
it compresses positive values closer to 0 |
|
|
Term
| If you have an ANOVA which is a 2x2, how do you control for familywise error? |
|
Definition
|
|
Term
| If you have an ANOVA which is more than 2x2, how do you control for familywise error? |
|
Definition
|
|
Term
| If you are doing more than 1 contrast per family, how do you control for familywise error? |
|
Definition
|
|
Term
| If you are doing more than 2+ contrasts per family, how do you control for familywise error? |
|
Definition
-add the correction into the syntax
-manually correct by dividing alpha by the number of contrasts per family |
|
|
Term
| How do you report the results of a Mauchley's test? |
|
Definition
| X^2 (df) = chis-square, sig |
|
|
Term
| If Mauchley's test is significant, how do you report your hypothesis results? |
|
Definition
look at the Greenhouse-Geisser results (within subjects)
F= GG (dfGG, dfGGerror) p = GG sig |
|
|
Term
| When making a contrast chart, where do you put the condition you are looking in and the things you are comparing? |
|
Definition
condition you are looking in- TOP
things you are comparing- LEFT SIDE |
|
|
Term
| What is familywise error rate? |
|
Definition
| The likelihood of having a type I error in at least one comparison. |
|
|
