The name is absent



Stata Technical Bulletin

25


label variable hybcor "hybrid 90% c.i. correct?"
label values hybcor correct
*

generate stulo=orb-t95*bootSE

generate stuhi=orb-t05*bootSE

generate Stuwide=Stuhi-Stulo

label variable Stuwide "width percentile-t interval"
generate Stucor=O

replace Stucor=I if stulo<3 & 3<stuhi

label variable stucor "student 90% c.i. correct?"
label values stucor correct

*

label data "n=80 bootstrap∕Monte Carlo"

label variable orb "Y=3X + Xe,e~chi2-1,X~chi2"

save boot4.dta, replace
end

example4.ado creates a dataset, boot4.dta, containing information on the width and inclusion rates of four types of “90%
confidence” intervals: standard t-table, bootstrap percentile, hybrid bootstrap percentile (equation [3]), and bootstrap percentile-t
intervals (equation [5]). Here are results based on 400 Monte Carlo samples:

Variable

Obs

Mean

Std. Dev.

Min

Max

iterate I

400

200. ε

115.6143

1

400

orb I

400

2.999104

.5820095

2.119488

5.551427

orSE I

400

.17784

.0804133

.0574202

.4905333

bootb I

400

3.003376

.5399735

2.153254

5.26923

bootSE I

400

.4232904

.2956025

.0889773

1.708986

poε I

400

2.402226

.2288134

2.019951

3.410539

p9ε I

400

3.748616

1.005218

2.370434

7.698514

toε I

400

-6.176697

4.773216

-33.64367

-1.241619

t9ε I

400

3.466492

1.069656

1.383403

7.744884

Stanlo I

400

2.7035

.4828733

1.942544

4.943048

stanhi I

400

3.294708

.6929325

2.242965

6.159807

Stanwide

400

.5912081

.2677761

.1912093

1.633476

stancor I

400

.3475

.4767725

0

1

perwide

400

1.346389

.9071145

.2817495

5.096023

percor I

400

.7625

.4260841

0

1

hyblo I

400

2.249593

.4123866

.7103348

3.764181

hybhi I

400

3.595982

1.019064

2.19773

8.523996

hybwide

400

1.346389

.9071145

.2817495

5.096023

hybcor I

400

.615

.4872047

0

1

stulo I

400

1.359049

1.242871

-5.784867

3.046353

stuhi I

400

6.822549

7.106885

2.264905

53.62768

Stuwide I

400

5.4635

8.198419

.3112769

57.97469

stucor I

400

.9025

.2970089

0

1

Means of stancor, percor, hybcor, and stucor indicate the proportion of “90% confidence” intervals that actually
contained
β = 3. Of course the standard t-table interval fails completely: only about 35% of these “90%” intervals contain the
true parameter. The narrow intervals dictated by this method drastically understate actual sampling variation (Figure 5). Neither
bootstrap percentile approach succeeds either, obtaining about 76% and 61% coverage. (Theoretically the hybrid-percentile
method should work better than percentiles, but in experiments it often seems not to.) But the studentized or percentile-t method
seemingly works: 90% of its “90% confidence” intervals contain 3.

The percentile-t method succeeds by constructing much wider confidence intervals, which more accurately reflect true
sampling variation. The median width of percentile-t intervals is 2.66, compared with only .59 for standard t-table intervals. The
mean percentile-t interval width (5.46) reflects the pull of occasional extremely wide intervals, as seen in Figure 6.

In Hamilton (1992), I report on another OLS experiment using a somewhat less pathological regression model. There too,
bootstrap percentile-t methods achieved nominal coverage rates (over 1,000 Monte Carlo samples) when other methods did not.
That discussion includes a closer look at how studentization behaves in the presence of outliers. Bootstrap confidence intervals
based on robust estimators and standard errors (for example, see Hamilton 1991b) might achieve equally good coverage with
narrower intervals—one of many bootstrap/Monte Carlo experiments worth trying.

Warning: even with just 100 Monte Carlo samples and 2,000 bootstrap resamplings of each, example4.ado requires hours
of computing time and over three megabytes of disk space. Scaled-down experiments can convey a feel for such work, and
explore promising new ideas. For example, change the two “while 7,_mcit<101 {” statements to “while 7,_mcit<51 {”
and change “while 7,_bs ample <2001 {” to “while 7,_bsample<101 {”. Full-scale efforts might easily require four million
iterations per model/sample size (2,000 bootstrap resamplings for each of 2,000 Monte Carlo samples), tying up a desktop



More intriguing information

1. On s-additive robust representation of convex risk measures for unbounded financial positions in the presence of uncertainty about the market model
2. IMPROVING THE UNIVERSITY'S PERFORMANCE IN PUBLIC POLICY EDUCATION
3. Discourse Patterns in First Language Use at Hcme and Second Language Learning at School: an Ethnographic Approach
4. The name is absent
5. EMU's Decentralized System of Fiscal Policy
6. Automatic Dream Sentiment Analysis
7. The Value of Cultural Heritage Sites in Armenia: Evidence From a Travel Cost Method Study
8. Economie de l’entrepreneur faits et théories (The economics of entrepreneur facts and theories)
9. Learning-by-Exporting? Firm-Level Evidence for UK Manufacturing and Services Sectors
10. Dendritic Inhibition Enhances Neural Coding Properties
11. The magnitude and Cyclical Behavior of Financial Market Frictions
12. Bidding for Envy-Freeness: A Procedural Approach to n-Player Fair Division Problems
13. Gerontocracy in Motion? – European Cross-Country Evidence on the Labor Market Consequences of Population Ageing
14. The changing face of Chicago: demographic trends in the 1990s
15. The name is absent
16. The name is absent
17. THE ANDEAN PRICE BAND SYSTEM: EFFECTS ON PRICES, PROTECTION AND PRODUCER WELFARE
18. Philosophical Perspectives on Trustworthiness and Open-mindedness as Professional Virtues for the Practice of Nursing: Implications for he Moral Education of Nurses
19. The Dictator and the Parties A Study on Policy Co-operation in Mineral Economies
20. Equity Markets and Economic Development: What Do We Know