Used price: $35.86
Buy one from zShops for: $35.86
Used price: $10.00
Buy one from zShops for: $40.00
Used price: $30.00
Buy one from zShops for: $39.00
Used price: $15.00
Buy one from zShops for: $24.75
Used price: $2.98
Collectible price: $9.53
Buy one from zShops for: $16.00
Donohue draws three basic conclusions from his work. First, that the best elements of government to privatize are those in which the government can precisely specify in contractual terms the work to be performed. Second, He explains the lengths at which it is reasonable to go to make a privatization effort work. Third, to emphasize the need for competitiveness -- albeit from private companies or government itself.
The book is well written and researched, and is easy reading.
Used price: $34.95
Buy one from zShops for: $55.95
Used price: $47.26
Buy one from zShops for: $47.26
Many of the leaders of this group have participated in writing these studies including several who have moved on to careers elsewhere (e.g. Bob Bell, now at AT&T Labs - Research; John Rolph, now at USC; Jim Hodges, now at University of Minnesota; and Carl Morris, now at Harvard University).
Those currently at RAND who have contributed include Allan Abrahamse, John Adams, Phyllis Ellickson, Lionel Galway, Catherine Jackson, Dan McCaffrey, Sally Morton and Dan Relles. This is a mix of very seasoned RAND statisticians along with some junior members and colleagues. Several members of the group did not contribute to the case studies but well could have. I was particularly surprised at the absence of Naihua Duan who is an ASA fellow and has contributed to major health studies at RAND. Naihua has also been responsible for innovations while at RAND including contributions to sliced inverse regression and transformation "smearing" methods.
Nevertheless, the collection of studies are both interesting and important to the public and in several cases the findings go counter to the popular information in the media. Well-known Stanford Statistics Professor Brad Efron calls the media statements "misinformation" and "disinformation" and claims that RAND gets it right in his foreward to the book. Many Stanford students and colleagues of Efron had careers at RAND including Bob Bell, Naihua Duan, Carl Morris, Bill Rogers and Sally Morton.
There are a total of 10 case studies included. The first three are categorized as primarily addressing the collection of data (addressing issues in the design phase). The next three are considered to be primarily addressing the detection of effects (estimation or hypothesis testing aspects of statistical analysis) and the last four are considered to emphasize the understanding of relationships.
I have skimmed through all ten case studies and have read case numbers 3, 4, 5 and 7 in detail. The topics are as follows: 1. School-Based Drug Prevention by Bell and Ellickson. 2. The Health Insurance Experiment by Morris and Hill. 3. Counting the Homeless by Abrahamse. 4. Periodicity in the Global Mean Temperature Series? by Adams, Hammitt and Hodges. 5. Racial Bias in Death Sentencing by Morton and Rolph. 6. Malpractice and the Impaired Physician by McGuigan and Rolph. 7. Supply Delays for F-14 Jet Engine Repair Parts by Galway. 8. Hospital Mortality Rates by Thomas and Rolph. 9. Eye-Care Supply and Need by Relles, Jackson and Lee. 10. Modeling Block Grant Formulas for Substance Abuse Treatment by McCaffrey and Adams.
Analysis in #3 indicates that there are only about 400 homeless in Orange County as opposed to public estimates and claims of 4000 or more. Results in #4 indicate that the data are inconclusive regarding a global warming effect. In #5 both logistic regression and tree classification methods are used to show no clear bias in death sentencing based on the race of the victim. In #7 careful analysis of the data reveal that transporting supplies is the key factor in delays for getting repair parts for the engines and not the slow procurement process.
As an applied statistician who does a fair amount of consulting, I always find good case studies to be enlightening and helpful to me in my practice of statistics. These articles are very good and enlightening and they follow a common format. They start with an executive summary that provides an overview and the bottom line results. This is followed by an introductory section and then a section describing the study design, data collection, data sources and elements. The third section deals with datafile creation, descriptive statistics and exploratory data analysis. The fourth section covers statistical methods and models used. The fifth section gives results. Section 6 is a discussion section which may include summary, possible future extensions of the analysis, and conclusion and recommendations. The final section provides exercises. This last section is excellent for a course based on the case studies as it tests the student knowledge based on material learned in the case study. Sometimes self-contained problems are given but in other cases the reader is referred to the casebook web page at the RAND web site where data sources can be found to do the exercises.
In practical work I have always found that a clear understanding of the problem and good descriptive statistics and/or graphics are far more important than the particular method of analysis (which often times can be very elementary). These studies exhibit this principle well. In many cases good exploratory analysis, good design and clear understanding lead to the key results and the appropriate statistical methods. These methods are usually simple and elementary although some are fairly new tools (e.g. bootstrap, tree classification and empirical Bayes methods).
Used price: $26.47