Exam 3

When you don't know the population's variance, why do you have to compute your sample's variance differently? (What are you doing? What are you fixing?)
Click the card to flip 👆
1 / 23
Terms in this set (23)
When you don't know the population's variance, why do you have to compute your sample's variance differently? (What are you doing? What are you fixing?)
It systematically underestimates the variance , so you use degrees of freedom to correct, which is the number of scores in a sample that are free to vary
Know when and how the T-distribution differs from the Z distribution
(see chart)
Know whether you use a standard error or a standard deviation to compute Cohen's d
You use standard deviation (Spooled)
Know why a within subjects design is superior to an equivalent between subjects design
It's powerful because you're using the same people so there's less room for variance ( less variability = less overlap = less standard deviation = more power)
Know your sources of non-independence of observations
...
Correct way to report a statistic: what your distribution is (t test), what your degrees of freedom are (15), what the test statistic is (5.62) and then what your alpha level is (.05)
t(15) = 5.62, p<.05
If you reject the null and(in?) an independent means t test, what does that mean in real life?
I think it means the two samples are drawn from different populations with unequal means
Know why we compute a pooled variance instead of just averaging our two variances when we have an independent means t test
...
Know when you would use a Satterthwaite approximation
when variances are unequal (heterogeneity of variance) -- because the homogeneity of variance assumption is violated
Know how to compute degrees of freedom for an independent means t test
dftotal = df1 + df2
Know what things will increase your level of power (from exam 2)Use a more homogeneous population, Increase sample size, Use a more lenient p-value (e.g., α =.10), Use a one-tailed testKnow what kinds variables you would use in a chi squared testcategoricalKnow what the basic idea of a chi squared test ischi squared tests test how well patterns of observed frequencies fit an expected pattern of frequenciesKnow how to compute degrees of freedom for a chi squared test for independencedf = (R-1)(C-1)Know what expected means when we're talking about chi square Why do we expect it When do we expect it Which hypothesis are we referring to when are talking about expected frequenciesExpected frequencies are the frequency value that is predicted from the null hypothesis; E or fe Expected = predicted Defined as the ideal hypothetical distribution that is in perfect agreement with null hypothesisKnow what your advantage is/ are of using chi square rather than parametric testCan examine data from nominal and ordinal scales (whereas parametric tests require data from interval or ratio scale)Know what your effect size cut offs are for phiSmall .01 Medium .03 Large .05Know when phi and Cramer's phi will be the same and when they will be differentCramer's phi and phi are the same in a 2x2 matrix or when K-1=1 aka dfsmaller-1=1Be able to tell me whether within subjects or between designs are more powerful and whyWithin subjects are more powerful because error is low and each subject is their own control; in within-subjects, randomly chosen members of the population are exposed to all levels of a casual variable (repeated measures and cross over) For between subjects, members of a population are randomly and independently assigned to specific levels of a casual variable; independently designed to one conditionBe able to fill in a table that describes what data you have and be able to tell me what your comparison distribution is and what shape that distribution is(see table from lab exercise)Be able to draw a chi square distribution depending on whether it has very few degrees of freedom or more degrees of freedom and how's that going to change in terms of its looks. Be able to draw a rough sketch of that and show me where the region of rejection should fallKnow what your 3 assumptions are in the independence means t test and also be able to identify what your two sources of error are for the independent means t test- observations must be independent; each participant is independently assigned to one condition/"cell" - two populations from which the samples are selected must be normal - two populations from which samples are selected must have equal variances (homogeneity of variance)Be able to list and define your reasons for non-independence of observationsCompositional effects: individuals are non-randomly sorted into groups ex.) members of different sororities Common fate: individuals share a context or environment ex.) group members who live together Mutual influence: individuals interact with one another ex.) individuals in a common fate situation who become friends Same individual: repeated measures or within-subject studies