What drives road rage?
In 2007, for the second year in a row, Auto Vantage commissioned Prince Market Research (PMR), an independent market research company, to conduct a national opinion poll on the topic of road rage. Consequently, commuters in the 25 largest US metropolitan statistical areas (MSA) were interviewed on their driving habits and attitudes. In order to qualify for the survey, respondents had to be 21 years of age or older and drive at least ten minutes during rush hour every day from Monday through Friday. A minimum of 100 interviews were completed in each of the 25 designated metropolitan areas amounting to a total of 2,521 interviews. The margin of error was [+ or -] 2.0% for the full sample. Interviews lasted six to eight minutes on average and offered no tangible incentives.
According to the survey, the five worst cities in terms of people's reported driving behavior were (in descending order) Miami, New York, Boston, Los Angeles, and Washington. Analogously, the five most courteous cities in terms of people's reported driving behavior were (in descending order) Portland, Pittsburgh, Seattle/ Tacoma, St. Louis, and Dallas-Fort Worth. The survey reports only 6% of the regular commuters interviewed in Miami felt drivers in their city were more courteous than commuters in other major American cities. In contrast, almost half of the interviewed commuters in Portland reported that drivers were more courteous in their city. Based on these responses, PMR computes the net courtesy score (used for city ranking) that ranges from a low of -54 for Miami to a high of 39 for Portland.
These disparities in the driving etiquette or courtesy score warrant an empirical investigation. For this purpose, we estimate the relationship between the net courtesy score and various demographic, geographic, and economic variables collected from the U.S. Census Bureau for the 25 largest metropolitan areas.
Since the net courtesy score is a continuous (interval) variable and the city ranking on which it is based is an ordinal variable, the conventional ordinary least squares (OLS) and ordered logit or ordered probit models can be used, respectively, to estimate the relationships between the two different types of dependent variables and corresponding MSA characteristics. However, a high degree of multicollinearity between many independent variables and a small number of observations (25) necessitate the use of pairwise correlation or regression analysis.
A large number of pairwise correlations were examined, but only a few variables turned out to be statistically significant. These statistically significant variables are the percentage of Hispanics in the population, percentage of high school graduates, percentage of foreigners, percentage of homeowners, average travel time to work, and population density in each MSA. In contrast to the pairwise correlations, the pairwise ordered logit and pairwise ordered probit regressions are conducted using city courtesy rankings as the dependent variable. Their results are similar to those obtained previously except for one variable (average annual temperature) in the probit regression that now appears to be statistically significant (p-value of 0.06) and negatively signed. The variables that do not show up as statistically significant in any of the pairwise or correlation regression analyses include the percentage of males, percentage of married males, percentage of African Americans, percentage of individuals below poverty level, unemployment rate, city age, median age, median income, and per capita income.
These empirical findings indicate that cities with higher home ownership rates and more educated populations are likely to have better driving etiquette or lower levels of road rage. On the other hand, densely populated cities and cities with longer commutes to work appear to suffer from higher levels of road rage. Interestingly, only Hispanics and foreigners, but not African Americans, are negatively and significantly correlated with the road rage measures. Perhaps, Hispanics and foreigners are highly correlated with some other variables that also affect road rage. In our data, Hispanics and foreigners have a statistically significant negative correlation with homeownership rates and a statistically significant positive correlation with commute time and population density. Thus, our findings suggest that education, population density, commute time, and home ownership rates have a strong connection with road rage in US cities. This study offers new and rare empirical evidence on the topic of road rage and its determinants.
Published online: 18 April 2008
Pavel Yakovlev * Arzu Sen
P. Yakovlev ([mail])
Duquesne University, Pittsburgh, PA, USA
West Virginia University, Morgantown, WV, USA
|Printer friendly Cite/link Email Feedback|
|Author:||Yakovlev, Pavel; Sen, Arzu|
|Publication:||Atlantic Economic Journal|
|Date:||Sep 1, 2008|
|Previous Article:||Did stealing signs help the 1951 NY Giants?|
|Next Article:||Defined contribution beta when combined with a defined benefit plan.|