214 test 1 Flashcards
(118 cards)
1
Q
what is the standard deviation (SD)?
A
The average amount by which scores differ from the mean, it captures the amount of individual difference in scores and describes how the data is spread around the mean.
2
Q
What is the Sum of Squares (SS)?
A
The sum of all squared deviations over the number of data points.
3
Q
What is variance?
A
Average deviation around the mean of a distribution (average of SS).
4
Q
How do you work out the variance?
A
Mean-X then square, add up squared numbers and divide by n-1.
5
Q
Why is ANOVA an example of a General Linear Model?
A
Because the SS (and hence variance) can be expressed as an equation.
6
Q
What is the individual difference of a score?
A
The difference of a score from the mean (-1 / +2)
7
Q
Scores in relation to the mean: What is the linear equation for score.
A
Score = Population mean + Individual Difference
8
Q
Why would the means and SDs from the SAME sample not be equal?
A
Random Sample Variability (RSV).
9
Q
What is the grand mean?
A
The mean of the two samples.
10
Q
Partitioning Scores: What is the population mean?
A
The grand mean, and is shared by the whole population.
11
Q
Partitioning scores: What is the experimental effect?
A
This is the difference between the mean of the samples and the mean of the population.
12
Q
How can two samples from DIFFERENT populations be expressed using a general form?
A
Score = Overall Population effect + Experimental Effect + Individual Difference.
13
Q
How can two samples from the SAME population be expressed using a general form?
A
Same as Different (Score = Overall Population effect + Experimental Effect + Individual Difference)
But- when samples come from a single population there is no experimental manipulation so experimental effect is always 0.
14
Q
What is assumed if the null hypothesis is true?
A
- Samples come from the same population
- Exp effect = 0
- All differences are due to RSV
15
Q
What is assumed if H1 is true?
A
- Samples come from different populations
- Exp effect does not = 0
- Differences are due to RSV and Exp Effect.
16
Q
What is the linear equation for variability in the single between factor ANOVA?
A
SS(score)= SS(Between groups) + SS(Within groups)
17
Q
Why is it impossible to know the grand mean?
A
Because we never know the population values, we have to estimate from the data extracted from our samples.
18
Q
How do you work out the estimated exp effect?
A
You take the grand mean (est. pop mean) away from the group mean for each group.
19
Q
What are the parametric assumptions?
A
- Interval or ratio scale of measurement
- Normal distribution in the population
- Population variances are equal –> homogeneity of variance and/or sphericity
20
Q
What is the Levene’s test used for?
A
To test if the samples have equal variance
21
Q
What do you want when running a Levene’s test?
A
A p that is NOT significant. You want a P of more than 0.05
22
Q
What do you need to consider to work out if there is a difference between groups?
A
The average difference within the groups and the average difference between the groups and then compare them.
23
Q
When is a One Between-Subject Factor ANOVA used?
A
When you have:
- More than 2 conditions
- 1 IV
- Between-subject factors
24
Q
What is a One Between-Subject factor ANOVA known as in SPSS?
A
One-Way ANOVA
25
When would you generally use a One-Way ANOVA?
When you want to compare the means of three or more samples (That differ along one dimension e.g. training scheme, attractiveness, personality).
26
What is the Levene's test?
It is used to test if samples have equal variances
27
What is homogeneity of variance?
When variance across samples is equal
28
What do you want when running a Levene's test?
You do not want a sig P (p value smaller than 0.05) because you want variance to be equal
29
when would you use a One Between-Subject factor ANOVA?
When you have:
- More than 2 conditions
- 1 IV
- Between subjects factors
30
What is a one between-subject factor ANOVA known as in SPSS?
One-way ANOVA
31
What does a One-way ANOVA not tell us?
Which group is significant, it just tells you that there is a significant difference
32
How do you calculate the F value in a One-way ANOVA?
Variance (expt effect) / Variance (error)
33
What happens when H0 is true in a One-Way ANOVA??
- Variability of result is only due to RSV
- F (Obs) compares components that measure the same thing
- F = 1
34
What happens when H0 is false in a one-way ANOVA ?
- Variability is a result of an experimental effect and RSV
- F(obs) compares components which measure different things
- F >>1
35
What do you need in order to find the critical value of F in a table?
df (expt. effect)
| df (error)
36
What are degrees freedom?
A measure of the number of ways a set of data is free to vary. This is related to the number of scores in the data set
37
What is p(obs)?
The probability of obtaining F(obs) when H0 is true
38
What happens if p(obs) is small?
The likelihood of obtaining F(obs) is unlikely when H0 is true.
39
What happens if p(obs) is large?
The likelihood ofobtaining F9obs) is likely when H0 is true
40
How do you input data in SPSS for Between-subject Factors?
One column for each factor, one column for scores.
41
One way ANOVA SPSS
1. Analyse
2. Compare means
3. One-Way ANOVA
4. Select DV and Factor
5. Post Hoc
6. Tukey
42
In a One-Way ANOVA summary table what do the following refer to:
a) between groups
b) Within groups
c) Total
A) Experimental effect
B) Residual component often called the error
C) Total Variability in the experiment
43
How is the Mean Square Calculated?
SS/ df
44
What is the Mean Square?
Between - The variance associated with the experimental component
Within- variance associated with the error component (RSV).
45
F(obs) = Variance (expt effect)/ Variance (error) OR
MS(Expt. effect) / MS (error)
46
How do you calculate the number of pairwise comparisons?
n(n-1)/ 2
47
Why do you need to work out how many pairwise comparisons there are?
To see how many post hoc tests you have to run
48
What is the experiment error rate?
The likelihood that you can commit a Type 1 error (rejecting the null when, you should accept the null).
49
How do you calculate the experimentwise error rate?
No. of comparisons x Error rate per comparison(0.05)
50
Why don't we run multiple t-tests?
Risk of type 1 errors.
51
What do post-hoc tests do?
They shift the criterion for rejecting H0 to lower the risk of type 1 errors for each test.
52
How to report the ANOVA results?
1. Describe the raw data (conditions etc)
2. Describe how the data was measured
3. Describe the type of analysis and the design
4. Describe the results of the ANOVA ( F (..) =)
5. Describe any additional analyses required (post hoc)
53
When do you run post-hoc tests?
When you have more than 2 conditions and if you have a significant result.
54
What variance test do you use with within design?
Mauchly's test of sphericity
A sphericity test
55
What variance test do you use for between?
Levene
| A Homogeneity test
56
When is sphericity assumed?
When there are only two levels fo a factor.
57
What do you want in a Mauchly's Sphericity test?
A p bigger than 0.05
58
What happens when you have p<0.05 in variance tests?
You have to correct, so you look at the greenhouse-geisser
59
What is a within-factor ANOVA known as in SPSS?
A repeated-measures ANOVA
60
What is a matched design?
When data is collected from two different samples that are matched based on relevant variables.
61
What are the benefits of using a within factor design?
SENSITIVITY
- get rid of a lot of error (nonsystematic variance)
- more sensitive to experimental effects
ECONOMY
- less pps are needed (but be careful of fatigue).
62
When do you use a within factor design?
When you are interested in performance of the same variable over time
OR
If the same subject is measured multiple times in different conditions.
OR
The same subjects provide measures/ratings on different characteristics. (Opinions, within normally better)
63
What is the issue with within-factor design?
People who perform better in one condition are more likely to perform better in the other condition. (e.g. being better at memory).
64
Within- Subjects representation of variability
SS(total)= SS(subjetcs) + SS(expt effect) + SS(RSV)
There is also variability within the subjects
65
How do you work out the variability for each subject?
Compare their scores from each condition to the mean for each condition.
66
What variability does SPSS not show in the Within factor ANOVA table?
SS Subjects
67
Within Factor ANOVA data entry in SPSS
1. Each subjects data on a separate row
| 2. One column for each level of the factor
68
Within Factor ANOVA in SPSS
1. Analyse
2. General Linear model
3. Repeated measures
4. Define factors and click add
5. Define
6. Move labels into variables
7. Options / EM means
8. Move factors to display means for box
9. Click compare main effects
10. Click Bonferroni
11. Click descriptive statistics, estimates of effect size and homogeneity tests
69
Statistic associated with the sphericity test?
Mauchly's W (NOT SIG VALUE)
70
What is Eta Squared?
The effect size.
71
BLT?
Between
Levene's
Tukey
72
BMW?
Bonferroni
Mauchly's test
Within
73
What are confidence intervals?
They represent the ranges over which you are 95% confident that the actual population means are somewhere in between.
74
Why are confidence intervals important?
They allow you to compare groups and estimate the likelihood of them being the same/different
75
What are interactions?
Findings that cannot be explained by a single factor, when one factor affects the other factor
The combined effect of two or more factors
76
What are the three experimental components in a two-factor ANOVA?
Two Main Effects
One Interaction
These are independent of each other
77
What does it mean that the experimental components are independent?
You can observe any combination of results. Whether one factor is significant or not has no bearing on the other factors
78
What are the three types of 2 factor ANOVA's?
2 Between-Subject Factor Design
2 Within-Subject Factor Design
Mixed Design
79
How many error terms in a 2 between factor design
1
80
What is the variance broken down into for a 2 Between Factor Design?
- Variance of one IV
- Variance of other IV
- Variance of the interaction of IVs
- Error
81
How do you report the F value for a two between-subject factor ?
F (Between df, within df)= value, p <> 0.05
82
How many error terms does a two within subject factor ANOVA have?
3
83
How would you report the F for a two-within subject factor?
F (Exp df, error df-of that exp condition-)= value, p<>0.05
84
In a mixed design what is the interaction considered a part of?
The within group
85
What are the adavantages of running a two way ANOVA?
Interactions- can look at how factors affect each other
Reduced type 2 errors
86
What are the three main questions each related to each experimental component, that we might ask in a two-way ANOVA?
1. Does the response variable depend on Factor A?
(Main effect 1)
2. Does the response variable depend on Factor B?
(Main effect 2)
3. Does the response variable depend on Factor A differently for different values of factor B and/or vice versa?
(Interaction)
87
What can we decompose interactions into?
Simple main effects
88
When do you look at simple main effects?
When you have a significant interaction
89
What should you see from the SMEs if the factors interact.
They should differ, if there is no interaction they will be the same
90
What are SMEs?
The decomposed components of one factor at a single level of the other factor
91
What is the structure for SMEs?
Factor A @ Level of Factor B
or
IV1 @ specific condition in IV 2
92
Why decompose into SMEs?
To determine whether:
1. Factor A affects how pps respond to Factor B
2. Factor B affects how pps respond to Factor A
3. Both 1 and 2
93
How do you write patterns for SMEs?
Level a > level b
94
What will you find with 2 factor interactions with SMEs?
1. Both sets of SMEs differ: Two causes- Both need to be investigated
2. One set of SMEs differ - one cause
3. Neither set of SMEs differ- no cause.
95
How do you determine the number of SMEs?
Add up the number of levels across all factors
96
What do sets of SMEs relate to?
IVs
| If there are 2 IVs then there are 2 sets of SMEs
97
How do you determine cause in SMEs?
By looking at the pattern, if they differ then that is one cause.
98
How do you determine SME patterns?
- 95% confidence intervals
or
- ANOVA, including posthoc tests
99
How should you enter the data for 2 within subject factor in SPSS?
Scores entered in columns:
| - One column for each level of the factor(s)
100
How do you run a 2 within ANOVA in SPSS?
1. Analyze
2. General Linear Model
3. Repeated measures
4. specify factors including number of levels
5. Move factors into within subject variables
6. EM Means
7. Display means (all factors)
8. Descriptive stats, estimates of effect size, homogeneity, compare main effects
9. Bonferroni
101
How do you work out SMEs in SPSS?
You have to run a series of within factor ANOVA (single)
You would have to have to run three (e.g.) for
A@ B1
A@ B2
A@B3
Then
B@A1
B@A2
B@A3
102
How do you calculate SMEs in SPSS?
1. Only put in the factor you are interested in (e.g. complexity)
2. Choose the variables with the level of factor B you are interested in (e.g. small)
3. Run the second set of SMEs
(if it has 3 or more levels then run bonferroni)
103
How to report the results of a 2-Factor ANOVA
1. Report the main effects
F=..., compare means of levels of factors, if posthoc analysis done then report p
2. Report the Interaction
f=..., and talk about analysis of SMEs
3. Report SMEs (all sets) - include posthoc of SMEs if done
104
How do you run a mixed ANOVA in SPSS?
1. Analyze
2. General Linear Model
3. Repeated Measures
4. Define within subject factor, levels, click add then define
5. Put variables into within factor variables and between factor into between box below
6. Options/EM means - descriptive stats, estimates of effect size, homogeneity
7. Factors and interaction into the display means for box
8. click compare main effects- choose Bonferroni in confidence interval adjustment- continue
9. Plots- within (horizontal)- between (separate lines)- add- continue- ok
105
How do you enter mixed data in SPSS?
| Age, regular, irregular e.g.
One column for between factor (group 1.00 /2.00 etc)
One column for each level of the factor
Age - Regular- Irregular
1. 00 - 12.00- 10.00
2. 00- 8.00- 11.00
106
How do you analyse SMEs for mixed?
Within Factor Component...
- Data window
- split data
- select organize output by groups
- enter the between factor to split data by
- conduct single factor ANOVA to examine SME components (one for each within factor or each SME)
Between Factor Component...
- remove between factor split in data
- One-way ANOVA
- Put within in Dependent and Between in factor (conduct one-way ANOVA)
- Use ANOVA table to see SME sig.
107
What is ANCOVA?
Analysis of Covariance- it allows you to consider the effects of the IVs, while filtering out or removing the effect of any other uncontrolled/confounding variables, the covariates.
108
When do you use ANCOVA?
When there are other variables that may affect your results (age, memory etc)
or
Pretest-posttest designs- take a measure before exp to filter out pre-existing differences
109
What does an ANCOVA do?
Also Compares variances (SS) between and within the groups/ conditions but in addition, it removes some of the variance/noise from any nuisance covariate(s). As a consequence, *it reduces some of the within-group variance.* (key)
110
What does an ANOVA do?
Compares the variance (ss) between and within groups/ conditions.
111
What is the consequence of an ANCOVA
Greater F values
| Adjust the means of the DV to an estimate of what they would have been had they not been affected by the covariates
112
When do you use ANOCVA?
Between subjects- when you cannot put the pps into separate groups.
Within subjects- when people have different baseline abilities (e.g. memory)
Pretest-posttest design, test given before pps are allocated to a condition and then again after.
113
What do you need to meet to run an ANCOVA?
1. A covariate should be chosen on the basis of existing theory and research
2. A covariate should ideally be measured using a scale at ratio, interval or ordinal.
3. Covariate should be measured before experimental manipulation
4. Relationship between covariate and DV must be linear (i.e. the relationship should be the same throughout the data).
5. If there is more than one covariate . they should not correlate with each other
6. There should be homogeneity of regression, i.e. the relationship between the DV and the covariate should be similar for all experimental groups
7. The assumptions of ANOVA should also be met
114
How do you run ANCOVA in SPSS?
1. general linear model
2. univariate
3. put your dv, iv, and covariate measures in.
4. click model, click custom, put all exp components in model box
5. Check to see if your homogeneity of regression is sig, in NOT SIG, continue
6. Go back to model and select full factorial (not custom as before)
7. descriptive stats, estimates of effecr size, homogeneity tests, display means for all, compare main effects, Bonferroni
115
How do you enter ANOCVA data in SPSS?
One row per subject
In pretest-posttest design for between-subject IVs
- one column for the IV/ Exp condition (1.00/2.00)
- one column for pretest
- one column for posttest score
116
What do you want for homogeneity of regression test?
A non significant P - because then your covariates have the same relationship for each condition/group.
117
What is normally different in the results of an ANOVA and an ANCOVA?
The significance for condition and the df or error
118
How do you report results for ANCOVA?
Include levels, say what was covaried out, the main effect on posttest ... was sig/not sig, f=...
Also report changes to the mean once the covariate was removed
