20090414 (1)
TRANSCRIPT
-
7/31/2019 20090414 (1)
1/15
-
7/31/2019 20090414 (1)
2/15
Lab Stuff
Questions about Chi-Square?
Intro to Analysis of Variance (ANOVA)
2
-
7/31/2019 20090414 (1)
3/15
Final lab will be distributed onThursday Very similar to lab 3, but with different
dataYou will be expected to find appropriate
variables for three major tests
(correlation, t-test, chi-square test ofindependence)
You will be expected to interpret thefindings from each test (one short
paragraph per test). 3
-
7/31/2019 20090414 (1)
4/15
4
Student Not Student Total
Males 46(40.97)
71(76.02)
117
Females 37(42.03) 83 (77.97) 120
Total 83 154 237 Observed
Expected
-
7/31/2019 20090414 (1)
5/15
5
-
7/31/2019 20090414 (1)
6/15
In its simplest form, it is used tocompare means for three or morecategories. Example:
Income (metric) and Marital Status (many
categories)
Relies on the F-distributionJust like the t-distribution and chi-square
distribution, there are several sampling 6
-
7/31/2019 20090414 (1)
7/15
If we have a categorical variable with 3+categories and a metric/scale variable, wecould just run 3 t-tests.
One problem is that the 3 tests would not beindependent of each other (i.e., all of theinformation is known).
As number of comparisons grow, likelihood ofsome differences are expected but do not
necessarily indicate an overall difference.
A better approach: compare the variabilitybetween groups (treatment variance + error)
to the variability within the groups (error) 7
-
7/31/2019 20090414 (1)
8/15
MS = mean square bg = between groups wg = within groups
The numerator anddenominator have their own
degrees of freedom df = of cate ories 1 k-1
wg
bg
MS
MSF =
8
-
7/31/2019 20090414 (1)
9/15
Generally, an f-ratio is a measure of
how different the means are relativeto the variability within each sample
Larger values greater likelihoodthat the difference between meansare not just due to chance alone
9
-
7/31/2019 20090414 (1)
10/15
If there is no difference between themeans, then the between-group sum of
squares should = the within-group sum ofsquares.
wg
bg
MS
MSF =
10
-
7/31/2019 20090414 (1)
11/15
A right-skewed distribution
It is a ratio of two chi-squaredistributions
11
-
7/31/2019 20090414 (1)
12/15
F-test for ANOVA is a one-tailed test.
12
-
7/31/2019 20090414 (1)
13/15
http://tinyurl.com/271ANOVA
13
http://tinyurl.com/271ANOVAhttp://tinyurl.com/271ANOVA -
7/31/2019 20090414 (1)
14/15
How do we know where the differences exist oncewe know that we have an overall differencebetween groups?
t-tests become important after an ANOVA so thatwe can find out which pairs are significantlydifferent (post-hoc tests).
Certain corrections can be applied to such post-
hoc t-tests so that we account for multiplecomparisons (e.g., Bonferroni correction, whichdivides p-value by the number of comparisonsbeing made)
There are many means comparisons test available(Tukey, Sidak, Bonferroni, etc). All are basically 14
-
7/31/2019 20090414 (1)
15/15
Conceptual Intro to ANOVA
Class Example: anova.do
GSS96_small.dta
15
http://faculty.vassar.edu/lowry/ch13pt1.htmlhttp://faculty.vassar.edu/lowry/ch13pt1.html