Psychological research skills Flashcards
(29 cards)
One way analysis of variance has what kind of a design?
One IV with three of more levels (conditions) or 2 or more IVs
Assumptions for one way ANOVA are
normally distributed data
homogeneity of variance
independence of observations
Interval or ratio data
if parametric assumptions for One way ANOVA are not met, you should use the corrected factors?
Between subjects, Kruskal-Wallis
Within subjects, Friedman
Which test is used to check for normality?
Shapiro-Wilk, Kolmogorov-Smirnof has too few observations most of the time.
Which test is used to check for equality of variances?
Levene’s.
Description in the ANOVA needs to include?
Decribe the statistical test, the variables, the purpose of the statistical analysis.
In APA format presenting the means?
a sentence for 3 or fewer, a table for 4-20 and 20+ is a graph.
There is an extra assumption in the within-participants ANOVA, which is? And it is tested with?
Sphericity. Mauchly’s test of Sphericity, only for factors with three or more levels.
When writing up Shapiro-Wilk versus when wiriting up Levene’s?
W(value) vs F(df1,df2)= …
Partial Eta Squared is? And it’s values indicate?
Measure of practical importance, 0.01 small effect, 0.06 medium effect, 0.14 large effect.
In a two-way between participants ANOVA there needs to be
Two or more IVs with one or more levels.
There is an interaction when…
The scores of a condition differ across factors.
What is a typer 1 error?
Rejecting the null hypothesis when it is in fact true. alfa is significance.
What is a type 2 error?
Failing to reject null hypothesis when it is false. B is significance.
Disadvantages to within-subjects design are… and can be fixed?
practice-effects and carry over effects. Fixed by complete counterbalancing.
In writing up a significant sphericity…
The Chi Square statistic for Maulchy’s test is significant, indicating that teh assumption of sphericity is not met for the effect of - X2(juu)….
Whixh corrector factor to use for violates sphericity assumption?
Greenhouse Geisser when epsilon <0.75, Huyn-Feldt if >0.75
What does power depend on?
Type 1 error Typer 2 error Sample size Statisctical method used Experimental design effect-size
How to increase power?
by reducing overlap between two distributions, aim to create low variablitiy = LOW ERROR
- increasing sample size, minimises sampling error
- eliminating poor methodology or poor tevhonolgy, minimises experimental error
- within-part design, no individual differences error
Partial Eta Squared and effect size,
Partial Eta Squared is the proportion of variance that is due to the effect
If data non normal…
Can remove outliers, or transform e.g. logarithmic transfromarion, Arcsine…
Sphericity violation indicates an
Increased chance of type 2 error and decreased power.
When violations of aaumptions are problematic?
small samples
data-samples are non-normal in different ways
one variance is more than 4 times the other
unequal sample size in each condition
If interactions are found, need to follow up with…
Test for simple main effects
