Week 2: One-way ANOVA, SS, Df, MSE, Fstat, Effect sizes Flashcards

1
Q

what is the underlying ANOVA equation?

A

Score = Grand mean + treatment effect + residual error (measurement and individual differences)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

Essentially, how does ANOVA break down variance? and what does the comparison of this breakdown produce?

A

Into 2 parts

  • Variance due to treatment
  • Error variance

Comparison of these two variances produces the F STATISTIC

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

What are the assumptions of a between ANOVA?

A

Homogeneity of variance: SD for all groups is about the same
Normality: error is normally distributed
Independence of observation: truly between design

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

What are degrees of freedom?

A

They represent the number of independent pieces of data, or values that are free to vary after a mean has been estimated

Can have total, treatment and error df

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

What is the calculation for total degrees of freedom?

A

Number of observation - 1

N-1

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

What is the calculation for treatment degrees of freedom?

A

Number of groups - 1

k-1

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

What is the calculation for error degrees of freedom?

A

df(total) X df(treat)

k(N-1)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

What is mean square?

A

Variance. Obtained by dividing the sum of squares for each effect by the corresponding degrees of freedom
So for total =SS(total)/Df(total)
Treatment = SS(treat)/Df(treat)
Error = MSE/df(error)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

F statistic calculation?

A

MS(treatment)/MS(error)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

What can the F statistic tell us about the null hypothesis?

A

Because of the way is is calculated, if the null is true these values should be similar and therefore F= close to 1

The further away from 1 F is, the less likely it is that the null is true

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

How do you set up data in jamovi for a one-way ANOVA?

A

One column for each IV and the DV

One row for each participant

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

How do you do a one-way ANOVA in Jamovi?

A

ANOVA -> ANOVA
Drag DV into the dependent variable box
IV into fixed factors
Descriptives (explore)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

What do we do then if the ANOVA output is significant (one-way)?

A

Click on the post hoc section under analysis
Put IV into the box on the right and click on post hoc desired - typically tukey

Can go into estimated marginal means - get a plot

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

Reporting a one-way ANOVA?

A

Means, SD, sample sizes

F(dfb,dfw) = F, p = p

Significant:
Tukey post hoc tests at an alpha level of 0.05…

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

What is the difference between statistical significance and psychological significance?

A

Statistical: tells us whether it would occur by chance or not - relates to alpha

Psychological: Asks whether a finding is meaningful - relates to effect size

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

What are d family effect sizes and what do they do?

A

Cohens d, hedges g, cohens f

They quantify the size of the difference between means in terms of SD units

17
Q

What are R family effect sizes and what do they do?

A

Eta squared, omega squared, adjusted Rsquared

Look at how strongly associated the IV and DV are

18
Q

Explain cohens D

A

part of d family so looks at the difference in means in terms of their SD
The further away this is from 0, the larger the effect

19
Q

What is considered a small, moderate and large cohens d?

A
  1. 2 small
  2. 5 moderate
  3. 8 large
20
Q

Is cohens d biased at all? how do you correct this?

A

It can be biased when you have a smaller sample size (overly generous for treatment effect)
- apply a correction factor. in this case you can use hedges g. It is equivalent to cohens d but accounts for small sample size bias

21
Q

What is cohens f?

A

Used when there are 3+ means

1/2 D so a small is 0.1, moderate 0.25, large 0.4

22
Q

What is eta squared and r squared?

A

They are squared correlations between the IV and DV

=SS(treatment)/SS(total)
Gives a score between 1 and 0
close to 1 as possible is best but not common in psychology

23
Q

What does it mean if we have a smaller eta squared?

A

There is more variation in DV measurement from other things, not IV

24
Q

How would we interpret, say an eta squared value of 0.477?

A

Means that 47.7% of the variance in the DV can be explained by the IV

25
Q

Eta squared and Rsquared can be biased… what can we use instead?

Effect stengths?

A

Omega squared and adjusted R squared
(eta squared can be biased because the means are sample means derived from a larger population)

Small effect - 0.02, medium - 0.15, large - 0.35

26
Q

What are confidence intervals?

A

Interval estimate that attempts to set limits which have a high level of encompassing the true population mean

27
Q

Explain a 95% confidence interval

A

If we repeated 100 times, 95% of these will include the true population mean

28
Q

In relation to confidence intervals, how do we know if the difference in means is statistically significant?

A

If 0 is outside range (doesn’t contain null hypothesis value)