Lecture 4 - between subjects ANOVA Flashcards
What does the ANOVA test do?
Tests for differences between sample means
F = t2
What is the main difference between the t-test and ANOVA tests?
T-tests can only compare two group means at a time, whereas the ANOVA can simultaneously compare more than two means
What does the significance level of p
Reflects the overall risk we are prepared to take of making a Type 1 error (false positive)
Why is ANOVA good for keeping the error rate low?
Each time we run a statistical test the statistical test carries an additional 5% of conducting a type 1 error so if you run the test three times it would be 15% or p
When did you use a between - subjects ANOVA?
When there is one independent variable with two or more levels, each produced by a different group of people
When you want to know if the means are really different or if the differences can be explained by error variability
Why is ANOVA a good test?
One test
Looks at whether the variance we observe between conditions represents a real difference or whether it could be explained completely by error variance
What are the three steps to hypothesis testing?
Work out the null distribution
Compare obtained difference with the difference due to error
If the probability of the result of the null distribution is 5% or less we can reject the null hypothesis and infer support for the experimental hypothesis
In ANOVA we use the variability within the conditions to estimate error variability. True or false?
True
When are groups of subjects likely to differ significantly?
If the variability between the group means is much larger than the variability within the groups
ANOVA makes a direct comparison between what?
The amount sample means vary from each other and the amount scores in each sample vary around the sample mean
(Observed variability divided by variability due to error)
Between treatments variance divided by within treatments variance is that same as saying what?
Observed variance divided by variance expected due to error
Where do we find between treatment groups variance?
Treatment effects + error variability
Where do you find within treatment groups variance?
Error variability
In a between-subjects design, where is the error variability found?
Experimental error + individual differences
How do you work out an f ratio?
Obtained variability (variability between conditions) divided by Error variability (variability within conditions)
Treatment effects + error variability divided by what is how you work out an f ratio?
Error variability
What does it mean when the ratio is close to one?
The les likely it is that the differences between the treatment conditions are due to treatment effects
What does Factor mean in ANOVA?
Variable - single factor ANOVA - there is one independent variable
What does treatment mean in ANOVA?
Condition
What is the mean square in ANOVA?
Variance
What do you need to calculate an ANOVA?
Need two measurements of variance: between treatment and within treatment variance
How do you calculate variance for an ANOVA?
Variance (mean square) = SS/df (sum of squares divided by degrees of freedom)
Steps of one way ANOVA between subjects:
MS between: sum of squares divided by degrees of freedom to get that variance
MS error: sum of squares divided by degrees of freedom to get that variance
MS between divided by MS error = f ratio
What are some important things to remember about F?
F can never be less than 1
Assuming the null hypothesis is true, the majority of scores you be near one
The F distribution is different with different values for the df’s for both numerator and denominator
What is the numerator?
Between treatments df
What is the denominator?
Within treatments df
