Statistics 1- 4 Flashcards
variance between IV levels is represented by
difference in means between IV levels
variance between IV levels is possibly caused by
Manipulation of IV (treatment effects)
Individual differences
Experimental error - Random or Constant error
variance within IV levels is possibly caused by
Indiv diff
Experimental error - Random error only
why is it so important that constant error is eliminated ( in the context of t-tests)
T tests use the variance Within IV levels to filter out the Variance Between IV levels.
Constant error can only create variance between IV levels and so can not be used by the variance within IV levels and filtered out.
Therefore, Its bias will continue through the T-test results
t value close to zero represents
small variance between IV levels relative to within IV levelst
value far from zero represents
large variance between IV levels relative to within IV levels
(manipulation of IV has had effect)
when is a DF more reliable
when it is higher
what does DF represent
degrees of freedom.
no. of pop parameters subtracted from sample size
Levenes test is a measure of
difference between variances
What result do we want from a Levenes test.
how is this shown
we want the null hypothesis.
because a null hypothesis represents no change in variance.
So don’t violate assumption of equal variance.
this is shown when p is greater than .05
if experiment is quasi, it may violate the assumption of independent observation.
How do we check for this and what does it mean
Group A cannot be used to predict Group B scores.
Check for a correlation between Groups. SPSS will show this.
Don’t want a correlation
if data seriously violates assumptions of t test , we should
use Mann whitney U test
what are the assumptions of a t test
normality: DV normally distributed under each level of IV
Homogeneity of variance: Varience in DV should be equivalent
Equivalent sample size: matters more for smaller samples
Independence of observations: Can group A predict group B .can assume is true if random allocation has occoured.
when can normality be assumed in a paired t-test
when n greater than 30
what assumptions do we make when conducting paired t test
- Normality (of difference scores of individuals between group (xiA - xi B)
- sample size (only threatened when get drop outs)
what parametric test do u use when assumptions are violated for paired t test
wilcoxon t test
can 95% confidence intervals be used to assume a difference in IV levels?
what do we take into account
(paired t test)
No. not alone.
Need to look into influence of IV in terms of size AND consistency of effect.
consistancy as in individual participant’s DV score changes at a similar rate between conditions
when is an anova test used.
when measuring diff in sample means of an IV that has more than 2 levels
how does the F value from ANOVA test relate to t value
F = t ^2
if ANOVA tests the same thing as t tests, why not just run three seperate t tests, one between each level?
because in each t test, there is a 5% chance of a type 1 error.
If we run multiple tests (from same population), this chance will increase.
e.g. more likely that will find a difference from chance alone
Familywise error rate
the probability that at least one of a family of comparisons, run on the same data, will result in a type 1 error
It provides a corrected significance level (a), expressing the probability of making a Type 1 error.
1 - (1-a)^c
what are omnibus tests
tests that control the familywise error rate e.g. ANOVA
assumptions for ANOVA test
- Normality
- Homogeneity of variance
- Equivalent sample size
- Independent observations
s
s
what is a small, medium, large value of eta ^2
small > .01, med > .06, large > .14
what does partial eta ^2 test
how much variance in the DV is explained by the manipulation of the IV overall
assumptions of RM ANOVA test
- normality of distribution of difference scores under each IV level pair
- Sphericity (homogeneity of Covariance). variance in difference scores under each IV level pair should be reasonably equivalent
what stats test do we use to test Sphericity in ANOVA RM
what does H0 state
Mauchly’s test
H0: no diff between covariances under each IV level pair (what we want)
what is sensitisation - order effects
p.p works out aim of study
what are carry over effects - order effects
effects of exposure to first IV will impact next IV
alternatives when counterbalamcing isn’t viable
practice - extensive pre study practice
fatigue - short experiment
sensitisatipn - longer gap between
carry over - include control