W2 - One Way ANOVA Flashcards
The first assumption of ANOVA is
Homogeneity of variance: That the variance for each mean is similar. V1=V2=V3
The second assumption of ANOVA is
Normality: normally distributed data
The third assumption of ANOVA is
Independence of observations: each observation is unique
How do we test that means are normally distributed?
Levene’s F test
If F test is significant, what do we do?
The null hypothesis of variance being equal in all populations is rejected
We represent whole populations with ___ letters and subpopulations with ___ letters
Greek, normal
ANOVA provides variance information about 2 things
Variance due to treatment
Variance unrelated to treatment
SStotal refers to
Sum of squares of all observations regardless of treatment
SStreat refers to
Sum of squares of treatment means around the grand mean multiplied by n
SSerror refers to
SStotal -SStreat
df refers to
The total independent pieces of data - at least 1
k refers to
The number of treatment groups
dftreat is
k-1
dferror is
N-k
dftotal is
N-1
Mean Square refers to
Variance
Sum of squares (SS2) equation is
(mean-GM)squared
MS=
SS/df
F test tells us the statistical significance. What is missing?
Psychological significance
The underlying model of one way ANOVA is:
X = GM + Tcondition + e
X = GM + Treatment Variance? + SD?
Cohen’s d is what kind of measure?
Effect size
How is Cohen’s d flawed?
It is positively biased
What’s a less biased measure to use instead of Cohen’s d?
Hedges g
What’s the basic effect size measure for three or more means?
Cohen’s f
What’s the basic measure of strength of association between the dependent and independent variables?
Eta squared (n thingy with a funny long second leg)
What is the bias of Eta2?
It is only a sample of a larger population
What is a less biased measurement of strength of association?
Omega-squared (woolworths w2)
