Empirical Methods in Law

Empirical Methods in Law

Homework – Chapter 10

Robert M. Lawless

Jennifer K. Robbennolt

Fall Semester 2016

1. In NAACP v. City of Mansfield (page 224), the question was whether the court should issue a preliminary injunction based partly on the statistical evidence of the discriminatory hiring practices of the city's police and fire departments. The court characterized the chi-square statistic on the pass rate for the police department exam to be only marginally statistically significant. For the fire department, the data showed that of 10 black applicants only 1 passed the exam but that 38 of 86 white applicants had passed. Without looking up the case, can you calculate the chi-square statistic for these data? Should the court have used the chi-square statistic? If not, what statistic should it have used?

1. In a recent study, one of our colleagues conducted a study of the “integrity” rationale for the exclusionary rule. The integrity rationale “counsels exclusion when the illegal search violates a cherished principle, such that introducing the evidence would ‘dirty the hands’ of the court that used it.” Our colleague asked law students and lawyers to take part in “a judicial role-playing exercise where they were induced at trial to either use dirty evidence (‘dirty hands’) or not use it (‘clean hands’). The main dependent measure was whether those in the ‘dirty hands’ condition were more likely to take, as a thank-you gift, a small bottle of alcohol based hand disinfectant (Purell) instead of a highlighter pen with Post-It flags in the barrel (Pen).” The output below is from a statistical program known as SPSS. How would you interpret the output?

Crosstabs

a. 0 cells (0.0%) have expected count less than 5. The minimum expected count is 7.51.

b. Computed only for a 2x2 table

1. Your authors were having a lunch-time discussion about the virtues of living in a small college town. One of these virtues is a short daily commute. One of us asserted that we were all doomed to long commutes because we had college degrees, and persons with college degrees tend to have longer commutes than persons who do not. That seemed really silly to the other two of us. The dispute could be settled only one way—with data. The U.S. Census conducts the American Community Survey (ACS) every year to collect population and housing information from the U.S. public. Among the data collected are data on educational attainment and data on commute time to the respondent's place of employment. The output below is from SPSS, using data from the 2007 version of the ACS. Looking at this output, what do you think? Do you feel comfortable concluding that persons with college degrees have longer commutes than persons who do not? Is the difference a meaningful difference?

Group Statistics

Independent Samples Test

1. The American Community Survey (ACS) discussed above also reports on the total income earned by its respondents. Imagine that we wanted to know whether income differs across U.S. Census region. Like most income data, the ACS data are severely positively skewed. Hence, we applied a logarithmic transformation to these data before we analyzed them using ANOVA and an F-statistic. The output below is from a statistical program called Stata and is again from the 2007 version of the ACS. Looking at this output, what do you conclude? Is total income the same across all four U.S. Census regions? Is there anything you would want to note in coming to a conclusion about that question? As you review the output remember that the data were log-transformed before being analyzed.

Bartlett's test for equal variances: chi2(3) = 759.8090 Prob>chi2= 0.000

1. Explain whether a chi-square test, a t-test, or ANOVA would be more appropriate for examining the following questions. Explain how to perform the correct analysis.
2. Whether government litigants win more often in court than do private litigants.
3. Whether a jury pool represents the racial composition of its community.
4. Whether creditor recoveries are different in corporate bankruptcies and consumer bankruptcies.
5. Whether any of the 12 federal judicial circuits awards higher damages in admiralty cases than the others.
6. Whether an apology results in higher or lower damage awards in lawsuits arising out of automobile accidents.
7. Whether state or federal courts convict more criminal defendants.
8. Whether judges are more likely to overturn a statute when they belong to the same political party that controls the legislative branch.
9. Whether Organization for Economic Co-operation and Development (OECD) countries have the same GDP per capita.
10. Whether the length of a criminal defendant's sentence is related to whether the defendant has a previous conviction and whether the defendant has expressed remorse.
11. Whether states with more lawyers per capita tend to have high bar passage rates.