What are the types of rationale?
- motivated by methodological problems
- considers different theories where there are gaps or problems
- links past to new research
What is a simple comparison design?
- 1 IV
- 2 conditions (either within or between)
What is a one way design?
- only 1 IV
- 3 or more conditions in one experiment
What is a factorial design?
- more than one IV
- equivalent to 2 experiments simultaneously
- measures interactions between factors
What is ANOVA?
Comparisons of within group error variance to between group variance
- if variance within each group is similar to variance between the groups
- -> same population
What is an F ratio?
- ratio of explained and unexplained variances
- F = between-groups variance/ within-groups variance
- tests likelihood of 2 variance estimates being from the same population
- F>1 = significant difference between groups
What are post-hoc comparisons?
- checks where the differences in a significant ANOVA are located
- problems with multiple comparisons = familywise error rate
What is a familywise error?
- usually p<0.05 means a significant results but with multiple comparisons the comparisons are multiplied together => greater significance = error
- to avoid this, set an acceptable overall error rate e.g. 0.05 and divide this by the number of comparisons
- -> new criterion for significance (Bonferonni correction)
What is the formula for calculating the number of interactions in ANOVA?
2(^k)-k-1
k = number of variables
What is an interaction?
An interaction consists of any variation in the scores which is not due to error or main effects
Can be represented graphically
Parallel lines = NO interaction
Not parallel = interaction
What is an independent t-test?
- parametric test (if not –> Mann Whitney)
- assess whether there are significant differences between 2 independent or separate groups (between ppts design)
Analyze –> compare means –> independent samples t-test
Reporting (Independent Sample Test table SPSS)
- if p
What is a paired samples t-test?
- parametric test (if not –> Wilcoxon)
- assess whether there are significant differences between 2 conditions completed by the same single group of people (within ppts design)
Analyze –> compare means –> paired samples t-test
Reporting (Paired Sample Test table SPSS)
- t(df) = t; p = p score
What is a Mann Whitney test?
- NON-PARAMETRIC equivalent of independent t-test
- used when there are different ppts in each condition
- looks at ranking position of values from each of the 2 conditions and provides a U-value
Analyze –> Nonparametric tests –> legacy dialogs –> 2 independent samples
Reporting (Test Statistics table SPSS)
- report medians not mean ranks
- U = u score; N = total; Z = z score; p = p score, 2 -tailed
What is a Wilcoxon test?
- NON-PARAMETRIC equivalent of paired samples t-test
- takes note of differences in scores for each data-pair, differences ranked lowest to highest
Analyze –> Nonparametric tests –> legacy dialogs –> 2 related samples
Reporting (Test Statistics table SPSS)
- report medians not mean ranks
- T = sum of ranks; N = total; Z = z score; p = p score, 2-tailed
What is a ONE-WAY BETWEEN ANOVA?
- 1 DV, 1 IV (grouping variable)
- univariate analysis of variance
- 2 SPSS columns (1 DV and 1 IV)
Report
F(df, error df) = F, p =, n^2p=
What is a Kruskal Wallis test?
- NON-PARAMETRIC equivalent of one way between ANOVA
- more than 2 independent groups
Analyse –> non-parametric –> legacy dialogs –> K independent samples
Report
X^2(df) = chi square, p = p score
POST HOC –> MANN WHITNEY (if significant)
What is a ONE-WAY WITHIN ANOVA?
- 1 DV and 1 factor (with more levels)
- each ppt takes part in all conditions
- 3 SPSS columns, one for each condition
Report
F(df, error df) = F, p =, n^2p=
What is a Friedman’s test?
- NON-PARAMETRIC equivalent of one way within ANOVA
Analyse –> non-parametric –> legacy dialogs –> K related samples
Report
X^2(df) = chi square, p = p score
POST HOC –> WILCOXON (if significant)
What is a 2x2 INDEPENDENT ANOVA?
- independent groups (between ppts)
- 2 factors = 2 grouping variables + 1 scale variable (DV) (3 SPSS columns)
Univariate analysis of variance
Test of between subject effects
- shows main effects and interactions
- significant main effect = find direction in descriptive statistics
Report
F(df, error df) = F, p =, n^2p=
SIMPLE EFFECT
- split file twice by each between ppts factor
- run independent samples t-test
What is a 2x2 REPEATED MEASURES ANOVA?
- repeated measures (within ppts)
- 4 conditions, ppts take part in all 4 (4 SPSS columns)
- 1 DV
Test of within subjects effects
- shows main effects and interaction
Report
F(df, error df) = F, p =, n^2p=
SIMPLE EFFECT
- run paired samples t-test
What is a 2x2 MIXED ANOVA?
- 2DVs and 2 IVs
- 1 grouping variable
- 3 columns (1 grouping variable and one for each within ppts condition)
Test of within subject effect
- main effect of repeated measures effect (e.g. sentence type)
- interaction between measures
Test of between subject effect
- main effect of independent measure (e.g. presentation order)
Report
F(df, error df) = F, p =, n^2p=
SIMPLE EFFECT
Looking at each IV at both levels of other IV
1. Looking at within factor at both levels of between
- independent samples T-test
- use both DVs for repeated measure
- Looking at between factor at both levels of within
- split file (organise output by groups)
- paired samples t-test
