JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

a government victory, a rebel victory, or a negotiated settlement? To explore this question we present an expected utility model of the choice rebels and governments face between accepting a settlement or continuing to fight. The model implies that a settlement becomes more likely as (1) estimates of the probability of victory decline; (2) costs of conflict increase; (3) estimates of the time required to win increase; or (4) the utility from a settlement increases relative to that of victory.Factors that (1) increase one party's probability of victory; (2) increase its payoffs from victory; (3) reduce the costs of conflict; and/or (4) reduce the time required to achieve victory increase that party's probability of winning and decrease the likelihood of a settlement. We test these propositions with a multinomial choice model that correctly predicts the outcome in 86 percent of the cases. NOTE: An earlier version of this article was presented at the 1996 Annual Meeting of the American Political Science Association. The authors are indebted to the editors and several anonymous reviewers for the suggestions on how to revise the manuscript. Any remaining errors remain the responsibility of the authors.

The concept that most transitional cell neoplasms of the urinary bladder exist as either nonaggressive lesions of low cytologic grade or aggressive anaplastic cancers is gradually gaining acceptance. The extent to which the biological potential of these neoplasms is revealed in their pathologic features is the subject of this article. Using guidelines developed in experimental models, a series of 400 transitional cell neoplasms selected for long-term follow-up were classified into the WHO system. The results indicate that (1) almost all transitional cell tumors can be grouped into low and high grades at initial presentation; (2) the low grade lesions (usually designated transitional cell carcinoma, Grade I) are benign and should be called papillomas rather than carcinomas; (3) the risk of progression is not a function of the number of recurrences for these noninvasive, low-grade, papillary tumors; (4) the high-grade neoplasms are aggressive whether papillary or nodular and account for greater than 93% of tumor-related deaths; (5) patients with high-grade lesions have a reduced life expectancy even if progression does not occur; (6) depth of invasion and growth pattern are limited as predictive factors compared with histologic grade; (7) histologic grading of the initial tumor tissue can be highly predictive of outcome.

Urban-rural quality of care differences may be a function not just of geography but also of the presence of a large nonwhite population.

This article evaluates the relative effectiveness of quality improvement interventions on increasing the time to antibiotic administration after a diagnosis of pneumonia. Clinical data were abstracted from the medical records of 17,040 Medicare beneficiaries discharged from one of 215 acute-care hospitals across 15 states. Thirteen Quality Improvement Organizations collected data on hospital quality improvement interventions from each hospital in this study. Medicare discharges between January 1997 and January 2002 define the study period. Most hospitals implemented multiple interventions to improve pneumonia care. Only 3 individual interventions were found to be effective in increasing time to antibiotic administration. Data feedback and benchmarking and medical records checklists had a positive effect on time to antibiotic administration. Administrative support by itself had a negative effect on the quality indicator. Although several combinations of interventions were also found effective, generalizations about the use of multiple interventions in quality improvement are difficult to make from retrospective studies.