W6 Flashcards
Do p values tell us anything about how important or large a difference between a group might be?
NO
What does an effect size tell us?
How likely an effect is to have occurred by chance
Effect size is a standardised way to tell us what?
How large the effect of our experimental manipulation was
What type of study is effect size especially important for?
Meta-analysis
To standardise results across all different types of studies.
In the context of ANOVA, how is the effect size usually reported?
partial eta squared
Effect size of 0.45
Is this a lot
Yes
How do you calculate effect size?
SS between / (SS between + SS error)
Now known as SStotal or Total variance.
What can Eta^2 range from?
0-1
Small —> Large effect
How do you layout the report of an effect size?
Between p value and MSE (mean squared error)
In the context of ANOVA, what will post hoc tests allow you to determine?
Which means sig differ from which
Which test directly counteracts inflated familywise error
Bonferroni
What does the Bonferroni test do?
Calculates a new alpha
Divides the original alpha by no. of tests conducted.
What do some argue about the Bonferroni test?
That it overcorrects for inflated type 1 error
Sacrifices too much power
When should you avoid using the Bonferroni test?
When you have lots of conditions
What does the TUKEY HSD (Honest Sig difference) do?
Calculates a new critical value
Compare TUKEY HSD with Bonferroni
TUKEY HSD is less conservative when comparing larger numbers of means
What is the LSD test equivalent to?
Perf multiple t tests
== Does NOT red our familywise error rate = Best to avoid
Which test should be used when you have unequal variances between your groups?
Games-Howell test
When Levines test p value is less than 0.05
Differences between the 1-way independent ANOVA and the 1-way repeated measures ANOVA
1 -way independent:
- Compare several independent groups w/ each other.
1-way repeated:
- Compare same/related groups
SSt (Sum of squares total) / Total variability
Deviation of each score from the grand mean
SSw (Sum of squares within)
Within participant variation
How SS vary about their own mean
SSr / Error
Variability of ind scores about their treatment mean
Q. True or false? In a repeated measures ANOVA we can ignore the between subject variance.
True
Q. True or false? A repeated measures ANOVA allows us to account for individual differences.
True
Q. In an independent ANOVA if the SSbetween value is 225.47 and the dfbetween value is 2. What is the MSbetween value?
112.74
Q. When running an ANOVA what do we use to identify which means differ from which?
Post hoc tests
Q. You have an SPSS output that reveals a value for Sphericity of 0.089. Should we conclude that the variances of the differences are roughly equal?
Yes
Q. In an independent ANOVA you have 3 groups each with 10 participants. If your SSm value is 22.48, what is your SSbetween value?
224.80
SSm x no. of people in each group
Q. In an independent ANOVA if the SSwithin value is 347.21 and the dfwithin value is 15. What is the MSwithin value?
23.15
SS within / df within
Q. If you want to compare 3 groups of different participants which test should you use?
One way Independent ANOVA
Q. If your alpha is 0.05 and you run 12 tests, what would your familywise error rate be?
46%
1-(1-0.05)^12
Q. In a repeated measures ANOVA, if your SSr value is 124.56 and your dfr value is 23, what is your error variance (MSr)?
5.42
Q. You have an SPSS output that reveals a value for Sphericity of 0.01. Should we conclude that the variances of the differences are roughly equal?
No
Q. True or false? Conservative post hoc tests reduce your probability of type 1 and type 2 error?
False
Q. Your output from a one way independent ANOVA revels a p value of 0.021. Is there a significant difference between the groups?
Yes
What is the test called that tests for sphericity?
Mauchlys test of sphericity
What does it mean if the mauchlys test of sphericity test statistic is greater than 0.05?
Conclude the variances of the differences are roughly equal.
= Assume sphericity
What are the types of corrections to violating the sphericity?
Greenhouse-Geisser (Recommended!)
Huynh-Feldt (sometimes too liberal)
Lower-bound (sometimes too conservative)
What must you do if the Mauchlys test is significant?
Read the Greenhouse Geisser row in SPSS output + report:
- DFs
- F value
- P value
Overall steps to calculate the F ratio
- Calculate total variance
- Calculate w/in participant variation
- Calculate effect (model) variance
- Calculate error (residual) variance
- Partition the degrees of freedom.
STEPS TO CALCULATING THE F RATIO
- Calculating total variance
- Find grand mean (mean of entire data set)
- Calculate grand variance
- Grand variance^2 x (n-1)
STEPS TO CALCULATING THE F RATIO
- Calculating total variance
How to calculate grand variance
Each value in data set - grand mean
Square each one
Add together
Divide by degrees of freedom
STEPS TO CALCULATING THE F RATIO
- Calculating w/in participant variation
Need variance column and number of groups.
Variance x (n-1) for each row then add all together.
STEPS TO CALCULATING THE F RATIO
- Calculating effect (model) variance
No. of participants x (condition mean - grand mean) ^2
STEPS TO CALCULATING THE F RATIO
- Calculating error (residual) variance
SSw - SSm
End equation to calculate F ratio
MSm / MSr
Q. If you have 10 participants who are each completing 5 different conditions which test should you use?
Repeated measures ANOVA
Q. In an independent ANOVA if the SSbetween value is 254.98 and the SStotal value is 499.74.What is the effect size (ŋp2)?
0.51
Q. In an independent ANOVA if the MSbetween value is 131.28 and the MSwithin value is 30.24. What is the F value?
4.34
Q. In an independent ANOVA you have 3 groups with the following means:
Group 1: 5, 16, 3, 21, 18, 7
Group 2: 4, 5, 12, 12, 18, 3
Group 3: 15, 22, 8, 10, 6, 6
What is the SS value for the means (SSm)?
4.03
Q. What term is used to describe your ability to detect an effect of a certain size?
Experimental power
Q. Your output from a one way independent ANOVA revels a p value of 0.021. Is there a significant difference between the groups?
yes
Q. In an independent ANOVA you have 3 groups each with 10 participants. If your SSm value is 22.48, what is your SSbetween value?
224.80
Q. In an independent ANOVA if the SSwithin value is 347.21 and the dfwithin value is 15. What is the MSwithin value?
23.15
Calculate the sample variance of the following numbers:
58, 26, 66, 74, 44
360.80
Q. In an ANOVA the F value represents the ratio of effect/model variance compared to which type of variance?
Error/residual
Q. True or false? A repeated measures ANOVA allows us to account for individual differences.
True
Q. What value do we use to tell us how large the effect of our experimental manipulation was?
Effect size
Q. You have an SPSS output that reveals a value for Sphericity of 0.568. Are the variances of the differences roughly equal?
Yes
Q. In an independent ANOVA if the SSbetween value is 206.54 and the SSwithin value is 316.87.What is the effect size (ŋp2)?
0.39
Q. In a repeated measures ANOVA, if your SSw value is 254.86 and your SSm value is 76.58, what is your error/residual (SSr) variance?
178.28
Q. If your alpha is 0.05 and you run 4 tests, what would your familywise error rate be?
19%
Q. In a repeated measures ANOVA, if your SSm value is 68.94 and your dfm value is 4, what is your effect variance (MSm)?
17.23
Q. Calculate the population variance of the following numbers:
12, 16, 15, 17, 8, 25, 21, 33, 29
59.14
