ANOVA Assumptions Flashcards
What is the main assumptions of an independent ANOVA? - (3)
- normal distribution
- Independence of scores
- homogenity of variance?
What is main assumptions for repeated ANOVA? - (3)
- normal distribution
- repeated measure design (same participants)
- sphereicity (Maulchy’s test)
What is main assumptions of mixed ANOVA?- (3)
- normal distribution
- independent and repeated factors
- homogenity of variance for the independent factor + sphereicity for repeated factor
What test is used for homogenity of variance?
Levene’s test
Levens’ tests if the
variances in independent groups are similar
Spherecity is an assumption of both
repeated measures and mixed models
Spherecity is the assumption that
variances of differences between all combination of related groups (levels are equal)
The violation of spherecity is when the
variances of the differences between all combination of related grps (levels) are not equal
The violation of spherecity is serious for repeated measures ANOVA as
it causes test to be too liberal (i.e., increase in Type 1 error rate)
The spherecity assumption applies when there are more than
2 data points from eac pp (3 or more levels in IV)
With spherecity you need to check in SPSS using
Mauchly’s test
For assumption of spherecity to be satisfied we need the Maulchly’s test in SPSS to be
non-significant (p > 0.05)
If Mauchly’s Test of Spherecity is significant ( p < 0.05) then reject H0 that variances of differences are equal and
aceept H1 that the variances of the differences are not equal (i.e., spherecity assumption has been violated)
If assumption of spherecity is violated then correct for lack of spherecity using
Greenhouse Geisser, Huynh-Feldt or Lower-bound estimates
Greenhouse-Geisser, Huynh-Feldt or Lower-Bound estimates rely on
estimating spherecity
You need to use adjusted DF for
greenhouse-geisser, Huynh-Feldt or Lower-Bound estimates
If GG statistic less than 0.75 if spherecity is violated then
use GG
If GG statistic more than 0.75 if spherecity is violated then
Huynh-Felft (HF) statistic
Report the following F statistic for this output - (3)
- Spherecity is violated since p = 0.000
- GG is less than 0.75 so use GG
- F(1.24, 21.00)= 212.32, p < .001
The researcher hypothesized that there would be an interaction between dog breed (Collie or German Shepherd) and week of obedience school training (all dogs measured at 1 week and 5 weeks) as they relate to the number of times the dog growls per week. Specifically, it was hypothesized that Collies would show no difference in growls between 1 week and 5 weeks, but German Shepherds would growl less at 5 weeks than at 1 week
What is IV, more than one IV, levels of IV, DV, BS or WS , what type of design? - (6)
- Dog breed and measurement time
- Yes
- Collie-German Sehpard/Week 1 to Week 5
- Number of growls
- Dog breed, BS, measurement time - WS
- 2-way mixed ANOVA
The researcher hypothesized that there would be an interaction between dog breed (Collie or German Shepherd) and week of obedience school training (all dogs measured at 1 week and 5 weeks) as they relate to the number of times the dog growls per week. Specifically, it was hypothesized that Collies would show no difference in growls between 1 week and 5 weeks, but German Shepherds would growl less at 5 weeks than at 1 week
- IVs: Dog breed and measurement time
- more than one IV? : Yes
- Levels of IVs: Collie-German Sehpard/Week 1 to Week 5
- DV: Number of growls
- BS or WS? : Dog breed, BS, measurement time - WS
- Type of design2-way mixed ANOVA
What does SPSS output show? - (3)
- Is there an effect of time overall? = Yes as p = 0.00
- Is there an effect of breed ? Yes as p = 0.000
- Is there an interaction effect? Yes as p = 0.000
The F-ratio tells us only whether the model fifted to the data accounts for more variation than extraneous factors but does not tell us
where the differences lie
Two ways to know where differences lie between the groups in ANOVA is - (2)
- Post-hoc
- Planned contrasts
