Rms Mct Jan Exam

Front 1

In terms of p value, why is it not as effective as going by effect size?

Back 1

Because you can just copy and paste the results over and over thereby increasing sample size and the results become significant i.e. not down to chance but only when there are a LOT of participants can the effect be visible. Effect size estimates this without having to use a massive sample.

Front 2

What is another issue with only using p values?

Back 2

It is difficult to compare studies that only report p values because of different sample sizes

Front 3

What is a major flaw of using p values?

Back 3

They reject logic based on probability ie just because the experimental hypothesis is correct does not mean the null hypothesis is incorrect

Front 4

What is a major advantage of using effect sizes?

Back 4

They are standardized, therefore we can make cross study comparisons

Front 5

What are the three different ways of working out effect sizes?

Back 5

Partial eta squared, Cohens D, and Pearsons R

Front 6

How does partial eta appear?

Back 6

ηp2

Front 7

What is partial eta squared used in?

Back 7

ANOVA

Front 8

What are the small, medium and large values of partial eta squared?

Back 8

Small = 0.01+
Medium = 0.06+
Large = 0.14+

Front 9

What is Cohens D used for?

Back 9

Comparing the magnitude of the means of two tests, often used within t-tests

Front 10

Which figures are used to calculate Cohens d?

Back 10

Either the mean and SD of each group OR t value and degrees of freedom HOWEVER this is only for a independent group t-test

Front 11

What are the small, medium and large levels of Cohens d?

Back 11

Small = .2+
Medium= .5+
Large = .8+ (note: Cohen’s d can be over 1)

Front 12

When is Pearsons r used as a measure of effect size?

Back 12

Comparing the difference between two groups in addition to correlations

Front 13

How do we interpret pearsons r?

Back 13

Small =.1+
Medium = .3+
Large = .5+

Front 14

What does ANOVA stand for?

Back 14

Analysis of variance

Front 15

What is ANOVA a form of?

Back 15

Regression

Front 16

What is the aim of Anova?

Back 16

To reduce error rates associated with multiple pairwise comparisons

Front 17

Within ANOVA, if we need to compare 4 conditions, how many pairwise comparisons would we need to do?

Back 17

6 pairwise comparisons

Front 18

Before we do comparisons within ANOVA, what must we do first?

Back 18

Conduct one test to see if there is a difference overall ie a main effect; if the ANOVA is significant, we then have justification to conduct the comparisons

Front 19

After we run ANOVA, what do we do?

Back 19

Look at comparisons between groups using planned comparisons and post hoc tests

Front 20

What is the one way between subjects ANOVA used for?

Back 20

Where we need to explore differences between 3 or more independent conditions on a single variable

Front 21

Give an example of a one way between subjects ANOVA

Back 21

If we had a theory that hair colour is associated with working memory (one DV), we would compare differences between scores of a working memory test between people with blonde, brown, red and black hair (1 between subjects IV, which has four levels)

Front 22

What will a one way ANOVA do and what will it not do?

Back 22

It WILL tell us is there is a significant difference across conditions; it WON'T tell us which conditions significantly differ from each other

Front 23

What are the three assumptions of between subjects ANOVA?

Back 23

Homogenity of variance

Normal distribution of data

Interval/ratio data (although ordinal is often also used)

Front 24

What is the homogenity of varience?

Back 24

Within each level of the IV, variances should be very similar

Front 25

What do post hoc tests reveal?

Back 25

Where exactly significant differences lie

Front 26

What is Levenes?

Back 26

A test for homogenity of varience (ANOVA)

Front 27

Which tests are used to calculate F-values if homogenity of variance is violated within an ANOVA?

Back 27

Browns-Forsythe and Welch tests

Front 28

What is the Tamhane’s T2 test?

Back 28

A post hoc test used when Levene's reveals no homogenity of variance

Front 29

What is the aim of a Bonferroni correction?

Back 29

Corrects the p value for each comparison; ie whatever P value SPSS calculates, it corrects it by 3 to make it harder to be significant

Front 30

Which type of graph is usually used for a one way ANOVA?

Back 30

A simple bar graph

Front 31

What do error bars indicate?

Back 31

The amount of variance the data has around the mean and/or confidence intervals

Front 32

If the error bars on two separate graphs do NOT overlap, what is it indicative of?

Back 32

A statistically significant difference

Front 33

If the 95 % confidence interval bars on a graph do not overlap, what does this indicate?

Back 33

That they are different

Front 34

When do we use a one way within subjects ANOVA?

Back 34

When we have a single within subjects variable with 3 or more levels

Front 35

What are the assumptions of the within subjects ANOVA?

Back 35

Normally distributed data

Ratio, interval or ordinal data

Sphericity

Front 36

What is Sphericity?

Back 36

Sphericity = equality of variances in DIFFERENCES between levels

Front 37

A-B variance = 15.75
A-C variance = 16.42
B-C variance = 16.01

Is sphericity assumed here?

Back 37

Yes

Front 38

A-B variance = 150.75
A-C variance = 3.12
B-C variance = 147.01

I sphericity assumed here?

Back 38

No

Front 39

Why do we not need a homogenity of variance test for a within subjects ANOVA?

Back 39

Because there is no between subjects factor

Front 40

Within a one way ANOVA, we do not ask for post hoc tests; what do we do instead?

Back 40

We ask for estimated marginal means and compare main effects (it will calculate whether there a difference between levels of the ANOVA)

Front 41

Within an LSD adjustment, which correction is made?

Back 41

None

Front 42

What is a between subjects ANOVA? (aka complex ANOVA)

Back 42

Where we have more than one IV - sometimes many IVs

Front 43

If you expect a main effect AND an interaction, what do you need?

Back 43

Power ie sample size

Front 44

Why is sample size more important when using between subjects ANOVA?

Back 44

Because each person only gives one score

Front 45

What are the assumptions of between subjects ANOVA?

Back 45

Interval/ratio data but ordinal can be used

Homogenity of variance

Front 46

Why is there only one test statistic that looks at homogenity of variance with a between subjects ANOVA?

Back 46

Because it compares cells rather than IVs seperately

Front 47

Within subjects ANOVA requires...

Back 47

Sphericity

Front 48

Between subjects ANOVA requires....

Back 48

Homogenity of variance

Front 49

What does the Mauchly’s test test for?

Back 49

Sphericity

Front 50

What is a common method of doing planned comparisons on a within subjects ANOVA?

Back 50

T-tests

Front 51

When is a mixed ANOVA used?

Back 51

When we have a mix of within subject IVs and between subjects IVs

Front 52

In which case would we set up a mixed ANOVA?

Back 52

Where we think there will be the between subjects variable will have a differential effect on the different levels of the within subjects IV

Front 53

Give an example of a simple mixed ANOVA

Back 53

The effect of gender on different measures of IQ

Gender (male, female) between subjects
IQ measure (Verbal IQ, non verbal IQ) within subjects

Front 54

What is the Box's test for a mixed ANOVA?

Back 54

It compares variances within the two covariance matrices'

Front 55

What is the difference between Box's test and the other tests?

Back 55

We want Box’s test to be non-significant (like other ANOVA assumption tests) however, unlike the others it is only an issue if p<.001!

Front 56

What kind of graph is best for a mixed design?

Back 56

Clustered bar graph

Front 57

What is MANOVA?

Back 57

Multivariate analysis of variance

Front 58

When do we use MANOVA?

Back 58

When we have one IV and many DV's

Front 59

Give an example of the kind of experiment a MANOVA would be used for

Back 59

The effect of pet ownership (one IV) on mental health (many DV's ie anxiety, depression, eating disorders, OCD, etc etc)

Front 60

Why not just run many ANOVA's rather than one MANOVA?

Back 60

Because more DVs increases the risk of a type 1 error

Front 61

What does the MANOVA assess?

Back 61

Whether or not your IV has an OVERALL effect on your variables taken together, after which you can look closer and see WHICH variables it effects

Front 62

In a multivariate ANOVA will the IV will always be within or between subjects?

Back 62

Between subjects

Front 63

What kind of homogenity of variance test will be calculated?

Back 63

Levene’s will be calculated for each DV (you want them to be non-sig)

Front 64

MANOVA will produce two homogenity tests; which?

Back 64

Box's and Levene's

Front 65

If the Box's test produces a NON sig result, what does this mean?

Back 65

That the p value taken from the multivariate test box in the output is reliable

Front 66

If Levene's test produces a NON sig result, what does this mean?

Back 66

That each p value for each DV taken from the between subjects contrasts box is reliable

Front 67

In addition to a F value, degrees of freedom and a p value, which other four measures are given in an ANOVA?

Back 67

Pillai's Trace

Wilks' Lambda
Hotelling's Trace
Roy's Largest Root

Front 68

Why would a Tamhane’s T2” test be used within a MANOVA?

Back 68

Where one variable violates the assumption of equality of variances

Front 69

What does ANCOVA stand for?

Back 69

Analysis of Covariance

Front 70

What are covariates?

Back 70

Covariates are simply variables that are associated with your DV but are not part of your experimental manipulation

Front 71

Give an example of a covariate

Back 71

Effect of alcohol (placebo/low dose/high dose) on working memory or age

Front 72

What are two common mistakes people make with ANCOVA?

Back 72

- Throw in as many covariates as possible

- Put in cetergorical variables e.g. gender, racial group

Front 73

We use a covariate to reduce...

Back 73

...within group error variance

Front 74

What does reducing within group error variance mean?

Back 74

You are more likely to find a significant effect of your manipulation (proportion of between group variance to within group error variance will increase).

Front 75

What do covariates do to confounds?

Back 75

Reduces them (more control!)

Front 76

What is the independence of the covariate and the IV effect?

Back 76

The covariate should be not be different across the IV groups meaning if you did a ANOVA on the covariate (DV) and group (IV) there would be no significant effect

Front 77

What is meant by the homogenity of variance slope?

Back 77

Where the relationship between the covariate and the DV is consistent across groups, e.g

if age was negatively associated with working memory in the placebo it would have to be negatively associated with working memory in the low and high dose conditions

Front 78

We conduct a within subjects ANOVA and the variances in the differences between scores in each condition are similar. This means....

Back 78

We have sphericity

Front 79

The bigger an F statistic the...

Back 79

...smaller the df

Front 80

The multivariate test statistic in MANOVA tells us

Back 80

The overall effect of the IV

Front 81

A 3x3 ANOVA would produce how many post hoc comparisons for the interaction?

Back 81

9

Front 82

We run an ANCOVA when...

Back 82

When we think a variable will affect our DV

Front 83

Rank the following effect sizes from the smallest to the largest: Partial eta squared = .06; d = .2; r = .5

Back 83

.06, .5, .2

Front 84

Name an example of an invalid covariate

Back 84

Religion

Front 85

You run an experiment with gender, experimental condition (warm room, mild room and cold rom) as between subjects factors and working memory, IQ, and verbal fluency as dependent variables. What would be the appropriate analysis

Back 85

2x3x3 mixed ANOVA

Front 86

An effect size of r=.3 is...

Back 86

Medium/moderate effect

Front 87

What is wrong with the following F(36,2)=4.68, p=.002, ηp2 = .29

Back 87

The df is too large

Front 88

When graphing data what would be appropriate for error bars

Back 88

95% confidence intervals

Front 89

Box’s test is a concern when...

Back 89

p<.001 as it's highly sensitive

Front 90

You produce a Bonferroni corrected p value of .3 for a comparison following a univariate ANOVA with an IV with 4 levels, what would be the non corrected p value?

Back 90

.005

Front 91

When doing a Bonferroni, SPSS multiplies a calculated p by the number of what?

Back 91

Comparisons, which we
compare to the standard critical p of .05.

Front 92

We run a 3 x 4 x 2 ANOVA and use the least significant difference test to explore a three way interaction. What may this cause

Back 92

Type 2 errors

Front 93

What is a type 1 error?

Back 93

Detecting an effect that isn't there

Front 94

What is a type 2 error?

Back 94

Failing to detect an effect that is there

Front 95

Effect sizes are useful if we...

Back 95

Want to see what manipulation across different studies causes biggest effects

Front 96

If the p value for Box’s test statistic is .01 you have

Back 96

Have homogeneity of covariance matricies

Front 97

Larger F values are likely to be associated with..

Back 97

Smaller p values. The larger your F value, the easier it is for your results to be significant

Front 98

A covariate is useful if wish to...

Back 98

Reduce error variance