Week 6-Simple Anova Flashcards
What is a Repeated Measures ANOVA?
Compares 3 or more means WITHIN the same individuals (we need normally distributed data for this)
What are some features of repeated measures ANOVA?
-Has one IV
-IV has 3 or more groups (levels) (can be longitudinal e.g., consecutive days, OR in one go)
-Independent participant groups=between-subjects
-Run multiple independent groups design OR One-way ANOVA for independent groups
-Related (Correlated) participant groups=within-subjects
-Run multiple correlated groups design OR One-way ANOVA for correlated groups
Define counterbalancing
Giving participants a different order of tasks e.g., rock, classical then pop OR pop, rock, classical OR pop, classical, rock etc.,
What are the benefits of a within-subjects design?
-We can control for individual variability as it’s the same participants in each condition = little to 0 error variance concerning individual differences between conditions (can find an effect easier)
-Smaller sample size can be used = a large number of observations will be collected from the same participants increasing the statistical power (meaning the ability to detect an effect)
What are the limitations of a between-subjects design?
-Differences among subjects are uncontrolled = Random error (can’t perfectly put people into conditions like matched pairs SO harder to find the effect)
-Larger sample size required (so you can detect an effect)
What are the benefits of Repeated Measures?
-Controlling for individual variability
-POWER determines the sample size needed (increased sensitivity, reduced noise so a smaller sample size needed as a result)
What are the problems with Repeated Measures design?
-Same participant in every condition = We violate the assumption of independence because behaviours within individuals are likely to correlate
-Missing data/statistical power
-If there is no data for one of the repeated measures, then data for that participant cannot be used (ANOVA is looking at variability within the individual looking for individual differences) SO can be a waste of time and resources
What is Sphericity?
The assumption that variability in scores across different groups is similar
-Likened to homogeneity of variances in One-way ANOVA
-Violating can lead to Type 2 error (don’t find an effect that is there)
What’s Maulchy’s test for sphericity?
-Tests whether variances across the contrasts are equal. If p > .05 we can say variances are not significantly different from one another (we want it to be non-significant)
-HOWEVER, some argue this test is useless as large and small samples can bias it (large = likely to violate sphericity/small = unlikely to violate)
-SPSS corrects the F-values for sphericity (correction is proportionate to violations) so might as well always correct
What are the assumptions of repeated measures ANOVA?
-DV should be at the scale level (Continuous interval and ratio data)
-Data should be normally distributed
-Sphericity (but assumption doesn’t necessarily need to be met as it can be corrected on SPSS which you can report JUST don’t add subjects into analysis)
-Data independence? (if same subjects in each group then assumption of independence is violated as not independent groups)
What function do you use on SPSS depending on the type of hypothesis?
-Directional=Contrasts button (press simple)
-Non-directional=EM Means
True or false: We want Maulchy’s test to be significant
False we do NOT
What should we do if we violate sphericity?
-Look at the Greenhouse-Geisser estimate of sphericity (E imagine it as a backwards 3 the e) (e=epsilon)
-When e > .75 then use the Huynh-Feldt correction
-When e < .75 then use the Greenhouse-Geisser correction (corrects it)
-Violation=report greenhouse-geisser
-Assumption met=report sphericity or greenhouse-geisser
Give an example of reporting repeated measures ANOVA (directional hypothesis)
‘A one-way repeated-measures ANOVA was conducted to compare the effect of [IV: in this case MUSIC type] on [DV: ANXIETY Symptoms]. The assumption of sphericity was met (X2(5) = 5.63, p = .344). There was a statistically significant effect of music type on subjective wellbeing (F(2.69, 72.52) = 4.06, p = .013, ηp2 = .113). Paired samples t-tests demonstrated that listening to Coldplay reduced anxiety compared to Eminem (t(27) = 3.22, p =.003) and Silence (t(27) =2.34, p = .027), but not Whale sounds (t(27) = 0.89, p = .380). Whale sounds were not significantly different from Silence (t(27) = 1.33, p = .194) but decreased anxiety compared to Eminem (t(27) = 2.67, p = .013). There was no difference between Silence and Eminem (t(27)= 0.72, p = .480).’ (IMPORTANT to put mean and standard deviation)
Give an example of reporting repeated measures ANOVA (non-directional hypothesis)
‘A one-way repeated-measures ANOVA was conducted to compare the effect of [IV: in this case MUSIC type] on [DV: ANXIETY Symptoms]. The assumption of sphericity was met (X2(5) = 5.63, p = .344). There was a statistically significant effect of music type on subjective wellbeing (F(2.69, 72.52) = 4.06, p = .013, ηp2 = .113). Bonferroni corrected post-hoc tests demonstrated that listening to Coldplay reduced anxiety compared to Eminem (p = .020), but not Silence (p = .162) or Whale sounds (p = 1.00). Whale sounds were not significantly different from Silence (p = 1.00) or Eminem (p = .077). There was no difference between Silence and Eminem (p = 1.00).’
