Causation and prediction of traffic fatalities /
Directed graph algorithms have been proposed as alternatives
| Main Author: | |
|---|---|
| Format: | Thesis Book |
| Language: | English |
| Published: |
[Place of publication not identified] :
[publisher not identified] ;
1997.
|
| Subjects: | |
| Online Access: | http://proxy.library.tamu.edu/login?url=http://proquest.umi.com/pqdweb?did=736562101&sid=1&Fmt=2&clientId=2945&RQT=309&VName=PQD |
| Summary: | Directed graph algorithms have been proposed as alternatives to regression based techniques for model specification on observational data. I apply directed graphs to an empirical analysis of traffic fatalities in the United States over the period 1947-1993. The algorithms revealed that traffic deaths are directly caused by income and speed. Speed is caused by vehicle mileage and safety device development. Vehicle mileage is affected by rural or urban driving ratio and income. Based on the information from these directed graph algorithms, models for total deaths, the total death rate, occupant deaths, and the occupant death rates are fit with yearly data over the period 1947-1989. Forecasts from these models are compared with forecasts from regression- based models over the period 1990-1993. The competing regression-based models of Peltzman, Crandall, and Graham were found from literature reviews, replicated over their initial data periods, and updated through 1989. Forecasts from the directed graph models outperform forecasts from the regression-based models in terms of mean squared forecasts error measures. These results suggests that the models based on directed graph algorithms are able to capture important empirical regularities, which earlier models were not able to capture. |
|---|---|
| Item Description: | Vita. "Major Subject: Economics". |
| Physical Description: | xiv, 195 leaves : illustrations ; 28 cm. Issued also on microfiche from University Microfilms Inc. |
| Bibliography: | Includes bibliographical references: pages 179-184. |