Hypothesis Tests and Confidence Intervals in Multiple Regressors
description
Transcript of Hypothesis Tests and Confidence Intervals in Multiple Regressors
![Page 1: Hypothesis Tests and Confidence Intervals in Multiple Regressors](https://reader030.fdocuments.us/reader030/viewer/2022020721/56813e10550346895da7f01b/html5/thumbnails/1.jpg)
Hypothesis Tests and Confidence Intervals in
Multiple Regressors
1
![Page 2: Hypothesis Tests and Confidence Intervals in Multiple Regressors](https://reader030.fdocuments.us/reader030/viewer/2022020721/56813e10550346895da7f01b/html5/thumbnails/2.jpg)
Outline
Hypothesis Testing Joint Hypotheses Single Restriction Test Test Score Data
2
![Page 3: Hypothesis Tests and Confidence Intervals in Multiple Regressors](https://reader030.fdocuments.us/reader030/viewer/2022020721/56813e10550346895da7f01b/html5/thumbnails/3.jpg)
Hypothesis Tests and Confidence Intervals for a Single Coefficient
is approximately distributed N(0, 1). (CLT)
Thus hypotheses on can be tested using the usual t-statistic, and confidence intervals are constructed as
So too for and are generally not independently
distributed - so neither are their t-statistics (more on this
later).
3
![Page 4: Hypothesis Tests and Confidence Intervals in Multiple Regressors](https://reader030.fdocuments.us/reader030/viewer/2022020721/56813e10550346895da7f01b/html5/thumbnails/4.jpg)
Example: The California class size data
The coefficient on STR in (2) is the effect on Test Score of a unit change in STR, holding constant the percentage of English Learners in the district.
Coefficient on STR falls by one-half. 95% confidence interval for coefficient on STR in (2) is
{−1.10 ± 1.96 0.43} = (− 1.95, − 0.26) The t-statistic for STR = − 1.10/0.43 = − 2.54, so we
reject the hypothesis at the 5 % significance level.4
![Page 5: Hypothesis Tests and Confidence Intervals in Multiple Regressors](https://reader030.fdocuments.us/reader030/viewer/2022020721/56813e10550346895da7f01b/html5/thumbnails/5.jpg)
Tests of Joint Hypotheses
Let Expn = expenditures per pupil and consider the population regression model
The null hypothesis that “school resources don’t matter,” and the alternative that they do, corresponds to
5
![Page 6: Hypothesis Tests and Confidence Intervals in Multiple Regressors](https://reader030.fdocuments.us/reader030/viewer/2022020721/56813e10550346895da7f01b/html5/thumbnails/6.jpg)
A joint hypothesis specifies a value for two or more coefficients, that is, it imposes a restriction on two or more coefficients.
A “common sense” test is to reject if either of the individual t-statistics exceeds 1.96 in absolute value.
But this “common sense” approach doesn’t work. The resulting test doesn’t have the right significance level.
6
![Page 7: Hypothesis Tests and Confidence Intervals in Multiple Regressors](https://reader030.fdocuments.us/reader030/viewer/2022020721/56813e10550346895da7f01b/html5/thumbnails/7.jpg)
Here’s why: Calculate the probability of incorrectly rejecting the null using the “common sense” test based on the two individual t-statistics.
To simplify the calculation, suppose that and are independently distributed. Let t1 and t2 be the t-statistics.
The “common sense” test is reject if |t1|>1.96 and/or |t2| > 1.96. What is the probability that this “common sense” test rejects H0 when H0 is actually true? (It should be 5%.)
7
![Page 8: Hypothesis Tests and Confidence Intervals in Multiple Regressors](https://reader030.fdocuments.us/reader030/viewer/2022020721/56813e10550346895da7f01b/html5/thumbnails/8.jpg)
Probability of incorrectly rejecting the null
8
![Page 9: Hypothesis Tests and Confidence Intervals in Multiple Regressors](https://reader030.fdocuments.us/reader030/viewer/2022020721/56813e10550346895da7f01b/html5/thumbnails/9.jpg)
which is not the desired 5%.
9
![Page 10: Hypothesis Tests and Confidence Intervals in Multiple Regressors](https://reader030.fdocuments.us/reader030/viewer/2022020721/56813e10550346895da7f01b/html5/thumbnails/10.jpg)
The size of a test is the actual rejection rate under the null hypothesis.
The size of the “common sense” test isn’t 5%. Its size actually depends on the correlation between
t1 and t2 (and thus on the correlation between and ).
Two Solutions. Use a different critical value in this procedure - not
1.96 (this is the “Bonferroni method” – see App. 7.1). This is rarely used in practice.
Use a different test statistic that test both and at once— the F-statistic.
10
![Page 11: Hypothesis Tests and Confidence Intervals in Multiple Regressors](https://reader030.fdocuments.us/reader030/viewer/2022020721/56813e10550346895da7f01b/html5/thumbnails/11.jpg)
The F-statistic
The F-statistic tests all parts of a joint hypothesis at once.
Formula for the special case of the joint hypothesis and
in a regression with two regressors.
where estimates the correlation between t1 and t2.
Reject when F is “large.”11
![Page 12: Hypothesis Tests and Confidence Intervals in Multiple Regressors](https://reader030.fdocuments.us/reader030/viewer/2022020721/56813e10550346895da7f01b/html5/thumbnails/12.jpg)
The F-statistic testing and
The F-statistic is large when t1 and/or t2 is large. The F-statistic corrects (in just the right way) for
the correlation between t1 and t2.
12
![Page 13: Hypothesis Tests and Confidence Intervals in Multiple Regressors](https://reader030.fdocuments.us/reader030/viewer/2022020721/56813e10550346895da7f01b/html5/thumbnails/13.jpg)
Large-sample distribution of the F-statistic
Consider a special case that t1 and t2 are independent, so . In large samples the formula becomes
Under the null, t1 and t2 have standard normal distributions that, in this special case, are independent.
The large-sample distribution of the F-statistic is the distribution of the average of two independently distributed squared standard normal random variables.
13
![Page 14: Hypothesis Tests and Confidence Intervals in Multiple Regressors](https://reader030.fdocuments.us/reader030/viewer/2022020721/56813e10550346895da7f01b/html5/thumbnails/14.jpg)
The chi-squared distribution with q degrees of freedom ( ) is defined to be the distribution of the sum of q independent squared standard normal random variables. In large samples, F-statistic is distributed as .
Selected large-sample critical values of
14
![Page 15: Hypothesis Tests and Confidence Intervals in Multiple Regressors](https://reader030.fdocuments.us/reader030/viewer/2022020721/56813e10550346895da7f01b/html5/thumbnails/15.jpg)
Compute p-value using the F-statistic:p-value = tail probability of the distribution beyond the F-statistic actually computed.
Implementation in STATAUse the “test” command after the regression.Example: Test the joint hypothesis that the population coefficients on STR and expenditures per pupil (expn_stu) are both zero, against the alternative that at least one of the population coefficients is nonzero.
15
![Page 16: Hypothesis Tests and Confidence Intervals in Multiple Regressors](https://reader030.fdocuments.us/reader030/viewer/2022020721/56813e10550346895da7f01b/html5/thumbnails/16.jpg)
16
![Page 17: Hypothesis Tests and Confidence Intervals in Multiple Regressors](https://reader030.fdocuments.us/reader030/viewer/2022020721/56813e10550346895da7f01b/html5/thumbnails/17.jpg)
The homoskedasticity-only F-statistic
To compute the homoskedasticity-only F-statistic Use the previous formulas, but using
homoskedasticity-only standard errors. Or Run two regressions, one under the null
hypothesis (the “restricted” regression) and one under the alternative hypothesis (the “unrestricted” regression).
The second method gives a simple formula.
17
![Page 18: Hypothesis Tests and Confidence Intervals in Multiple Regressors](https://reader030.fdocuments.us/reader030/viewer/2022020721/56813e10550346895da7f01b/html5/thumbnails/18.jpg)
The “restricted” and “unrestricted” regressions
Example: are the coefficients on STR and Expn zero?Restricted population regression (that is, under H0):
Unrestricted population regression (under H1):
The number of restrictions under H0 = q = 2. The fit will be better (R2 will be higher) in the
unrestricted regression (why?)
18
![Page 19: Hypothesis Tests and Confidence Intervals in Multiple Regressors](https://reader030.fdocuments.us/reader030/viewer/2022020721/56813e10550346895da7f01b/html5/thumbnails/19.jpg)
By how much must the R2 increase for the coefficients on Expn and Pct EL to be judged statistically significant?
Simple formula for the homoskedasticity-only F-statistic
where = the R2 for the restricted regression = the R2 for the unrestricted regressionq = the number of restrictions under the null
= the number of regressors in the unrestricted regression.
19
![Page 20: Hypothesis Tests and Confidence Intervals in Multiple Regressors](https://reader030.fdocuments.us/reader030/viewer/2022020721/56813e10550346895da7f01b/html5/thumbnails/20.jpg)
Example:
Restricted regression:
Unrestricted regression:
20
![Page 21: Hypothesis Tests and Confidence Intervals in Multiple Regressors](https://reader030.fdocuments.us/reader030/viewer/2022020721/56813e10550346895da7f01b/html5/thumbnails/21.jpg)
21
![Page 22: Hypothesis Tests and Confidence Intervals in Multiple Regressors](https://reader030.fdocuments.us/reader030/viewer/2022020721/56813e10550346895da7f01b/html5/thumbnails/22.jpg)
The homoskedasticity-only F-statistic-summary
The homoskedasticity-only F-statistic rejects when adding the two variables increased the R2 by “enough” - that is, when adding the two variables improves the fit of the regression by “enough.”
If the errors are homoskedastic, then the homoskedasticity-only F-statistic has a large-sample distribution that is .
But if the errors are heteroskedastic, the large-sample distribution is a mess and is not .
22
![Page 23: Hypothesis Tests and Confidence Intervals in Multiple Regressors](https://reader030.fdocuments.us/reader030/viewer/2022020721/56813e10550346895da7f01b/html5/thumbnails/23.jpg)
23
Joint confidence sets ctd.
Let F(1,0,2,0) be the (heteroskedasticity-robust) F-statistic testing the hypothesis that 1 = 1,0 and 2 = 2,0: 95% confidence set = {1,0, 2,0: F(1,0, 2,0) < 3.00}
3.00 is the 5% critical value of the F2, distribution
This set has coverage rate 95% because the test on which it is based (the test it “inverts”) has size of 5%
5% of the time, the test incorrectly rejects the null when the null is true, so 95% of the time it does not; therefore the confidence set constructed as the nonrejected values contains the true value 95% of the time (in 95% of all samples).
![Page 24: Hypothesis Tests and Confidence Intervals in Multiple Regressors](https://reader030.fdocuments.us/reader030/viewer/2022020721/56813e10550346895da7f01b/html5/thumbnails/24.jpg)
24
Confidence set based on inverting the F-statistic
![Page 25: Hypothesis Tests and Confidence Intervals in Multiple Regressors](https://reader030.fdocuments.us/reader030/viewer/2022020721/56813e10550346895da7f01b/html5/thumbnails/25.jpg)
Summary: testing joint hypotheses
The “common-sense” approach of rejecting if either of the t-statistics exceeds 1.96 rejects more than 5% of the time under the null (the size exceeds the desired significance level).
The heteroskedasticity-robust F-statistic is built in to STATA (“test” command). This tests all q restrictions at once.
For large n, F is distributed as . The homoskedasticity-only F-statistic is
important historically (and thus in practice), and is intuitively appealing, but invalid when there is heteroskedasticity.
25
![Page 26: Hypothesis Tests and Confidence Intervals in Multiple Regressors](https://reader030.fdocuments.us/reader030/viewer/2022020721/56813e10550346895da7f01b/html5/thumbnails/26.jpg)
Testing Single Restrictions on Multiple Coefficients
Consider the null and alternative hypothesis,
This null imposes a single restriction (q = 1) on multiple coefficients - it is not a joint hypothesis with multiple restrictions (compare with = 0 and = 0).
26
![Page 27: Hypothesis Tests and Confidence Intervals in Multiple Regressors](https://reader030.fdocuments.us/reader030/viewer/2022020721/56813e10550346895da7f01b/html5/thumbnails/27.jpg)
Two methods for testing single restrictions on multiple coefficients:
Rearrange (“transform”) the regression.Rearrange the regressors so that the restriction becomes a restriction on a single coefficient in an equivalent regression.
Perform the test directly.Some software, including STATA, lets you test restrictions using multiple coefficients directly.
27
![Page 28: Hypothesis Tests and Confidence Intervals in Multiple Regressors](https://reader030.fdocuments.us/reader030/viewer/2022020721/56813e10550346895da7f01b/html5/thumbnails/28.jpg)
Method 1: Rearrange (”transform”) the regression.
Add and subtract
where
28
![Page 29: Hypothesis Tests and Confidence Intervals in Multiple Regressors](https://reader030.fdocuments.us/reader030/viewer/2022020721/56813e10550346895da7f01b/html5/thumbnails/29.jpg)
(a) Original system:
(b) Rearranged (”transformed”) system:
so
The testing problem is now a simple one: test whether γ1 = 0 in specification (b).
29
![Page 30: Hypothesis Tests and Confidence Intervals in Multiple Regressors](https://reader030.fdocuments.us/reader030/viewer/2022020721/56813e10550346895da7f01b/html5/thumbnails/30.jpg)
Method 2: Perform the test directly
Example:
To test, using STATA, whether = : regress testscore str expn pctel, r test str=expn
30
![Page 31: Hypothesis Tests and Confidence Intervals in Multiple Regressors](https://reader030.fdocuments.us/reader030/viewer/2022020721/56813e10550346895da7f01b/html5/thumbnails/31.jpg)
Analysis of the Test Score Data
A general approach to variable selection and “model specification”
Specify a “base” or “benchmark” model. Specify a range of plausible alternative models,
which include additional candidate variables. Does a candidate variable change the coefficient
of interest ( )?
31
![Page 32: Hypothesis Tests and Confidence Intervals in Multiple Regressors](https://reader030.fdocuments.us/reader030/viewer/2022020721/56813e10550346895da7f01b/html5/thumbnails/32.jpg)
Is a candidate variable statistically significant? Use judgment, not a mechanical recipe. And don’t just maximize R2.
32
![Page 33: Hypothesis Tests and Confidence Intervals in Multiple Regressors](https://reader030.fdocuments.us/reader030/viewer/2022020721/56813e10550346895da7f01b/html5/thumbnails/33.jpg)
33
Digression about measures of fit… It is easy to fall into the trap of maximizing the R2 and 2R –
but this loses sight of our real objective, an unbiased estimator of the class size effect. A high R2 (or 2R ) means that the regressors explain the
variation in Y. A high R2 (or 2R ) does not mean that you have eliminated
omitted variable bias. A high R2 (or 2R ) does not mean that you have an unbiased
estimator of a causal effect (1). A high R2 (or 2R ) does not mean that the included variables
are statistically significant – this must be determined using hypotheses tests.
![Page 34: Hypothesis Tests and Confidence Intervals in Multiple Regressors](https://reader030.fdocuments.us/reader030/viewer/2022020721/56813e10550346895da7f01b/html5/thumbnails/34.jpg)
Variables we would like to see in the California data set
School characteristics student-teacher ratio teacher quality computers (non-teaching resources) per student measures of curriculum designStudent characteristics English proficiency availability of extracurricular enrichment home learning environment parent’s education level
34
![Page 35: Hypothesis Tests and Confidence Intervals in Multiple Regressors](https://reader030.fdocuments.us/reader030/viewer/2022020721/56813e10550346895da7f01b/html5/thumbnails/35.jpg)
Variables actually in the California class sizedata set
student-teacher ratio (STR) percent English learners in the district (PctEL) percent eligible for subsidized/free lunch percent on public income assistance average district income
35
![Page 36: Hypothesis Tests and Confidence Intervals in Multiple Regressors](https://reader030.fdocuments.us/reader030/viewer/2022020721/56813e10550346895da7f01b/html5/thumbnails/36.jpg)
A look at more of the California data
36
![Page 37: Hypothesis Tests and Confidence Intervals in Multiple Regressors](https://reader030.fdocuments.us/reader030/viewer/2022020721/56813e10550346895da7f01b/html5/thumbnails/37.jpg)
Digression: presentation of regression results in a table Listing regressions in “equation” form can be cumbersome
with many regressors and many regressions. Tables of regression results can present the key
information compactly. Information to include:
variables in the regression (dependent and independent).
estimated coefficients. standard errors. results of F-tests of joint hypotheses. some measure of fit (adjusted R2). number of observations.
37
![Page 38: Hypothesis Tests and Confidence Intervals in Multiple Regressors](https://reader030.fdocuments.us/reader030/viewer/2022020721/56813e10550346895da7f01b/html5/thumbnails/38.jpg)
38
![Page 39: Hypothesis Tests and Confidence Intervals in Multiple Regressors](https://reader030.fdocuments.us/reader030/viewer/2022020721/56813e10550346895da7f01b/html5/thumbnails/39.jpg)
Summary: Multiple Regression Multiple regression allows you to estimate the
effect on Y of a change in X1, holding X2 constant. If you can measure a variable, you can avoid
omitted variable bias from that variable by including it.
There is no simple recipe for deciding which variables belong in a regression - you must exercise judgment.
One approach is to specify a base model - relying on a-priori reasoning - then explore the sensitivity of the key estimate(s) in alternative specifications.
39