Parametric Tests Of Difference Flashcards
Directional hypothesis are more
Powerful
Power refers to
My chances of detecting a sig result
Why don’t we use directional hypotheses all the time?
Don’t have enough previous evidence
Effect size does not reveal
Significance level. Or size of effect
Effect sizes helps you compare data across…
Studies. Ie meta analysis
Effect size is the
Magnitude of difference btwn two populations
When we get more than two means/groups we use an
ANOVA
Z scores are calculated so that you can see
Where a score lies within a sample
The smaller the z score then the
More likely that that sample is part of the population
If the z score is higher than 1.96 we conclude that the
Sample has not come from the population
What is independent design?
Expose diff ppl to diff experimental manipulations
What is repeated measures or within-subjects design?
Single group of ppl exposed to different experimental manipulations at different times
Dichotomous variables what type of level measurement?
Nominal
Dichotomous variables have only ……..types of categories or levels
2 (ie male/female)
What is a paired-samples t-test?
Two experimental conditions and the sane participants take part in both conditions
If samples come from the same pop then the means will be
Similar
If the standard error is small what does that mean the sample means will look like?
Small SE = samples have similar means
Large SE = sample means big diff
What does signal to noise ratio mean??
Variance explained by the model / by the variance the model can’t explain OR effect / error
What are some of the assumptions we look for in our data before running parametric statistics
Normal distribution
Homogeniety of variance
Level of measurements (interval, ratio)
Outliers
T-test is the simplest form of
ANOVA
Effect size is looking at the
Relationship btwn the IV & DV.
How much variation in our DV can be attributed to our IV…is calculated by the
Effect size
Pearsons r is used to calculate the
Effect size
When the sample size is less than 30 the shape of the t-distribution is
Flatter than normal distribution. Greater area under tails
What is the t-test calculation?
t= observed diff btwn sample means - expected diff btwn pop means divided by est of SE of the diff of the 2 sample means
You can look at the upper and lower bounds of the 95% CI to see if your result will be sig. As it will include…….in the range
0
Independent test 3 assumptions
Normal distribution
Independence of scores in sample
Homogeneity of variances (levenes must have a non-sig value p=>.05
