Police standards of severity in resident versus non-resident traffic enforcement: a Florida case study

¶ … police use different Standards of severity when dealing with resident vs. Non-Resident drivers. A Florida case study.

Regression analysis is a statistical and econometric tool used by the scholars to find interrelations between different phenomena and numerically estimate them. This methodology is further exploited to extrapolate possible future values of the dependant phenomenon based on its' further performance following the past performance of the dependant phenomenon. The statistic model is used employing economic theory which is the basis for formulation of the statistic interdependence model. Thus, the regression analysis estimates the significance of the variation of the dependant phenomenon with the independent, and the influence of the latter on the former.

In order for the regression analysis to be accurate and to have scientific value, the population for the research must be selected carefully and the population must have normal distribution to avoid flaws in the population study. The independent or explanatory variables must be selected based on economic or other theory and the variables which could be interrelated must also be singled out. Thus, the relationship between the dependant variable is modeled as follows: The model theoretically looks as follows: Y = a + ?1*X1 + ?2*X2 +...+ ?, where Y is the dependant variable for which the causal relationship is to be estimated, a is autonomous level of Y when all the explanatory variables are equal to zero, ?1 and ?2 are coefficients of explanatory variables and the last variable in the equation is the error, or the difference between predicted by the linear dependency model value of Y at specific observation time and its' actual observed value.

Coefficients reflect the nature and depth of the influence of the independent variables on the dependant phenomenon. Positive coefficients reflect positive relationship, while negative coefficient is the evidence of the inverse relationship. For example, coefficient of 0.6 implies that the increase of the dependant variable by 1 point will trigger the growth of the dependant phenomenon by only 0.6. The coefficient higher than one implies that the movements in the independent variables have accelerated than normal affect on the dependant variable.

Negative coefficient implies negative relationship between the variables.

The relationship between the variables modeling is facilitated through inclusion of "dummy variables" which can improve the model and include the influence of factors, considering the assumption that "the partial effect of each explanatory variable is the same regardless of the specific value at which the other explanatory variable is held constant." Inclusion of these factors into explanatory model eliminates possibility of not tracking the affect of specific factors on the dependant variable. At the same time, control variables are included in the model for the purpose "to single out" with the use of these variables their affect on the dependant variable and trace the affect of other variables within the model.

The case of the speeding tickets is aimed to trace numeric relationship between the severity of the violence of the law, or the speed at which the person was caught driving over the speed limit at that traffic point. Thus, the objective is to model and try to numerically estimate influence of other factors on severity of violation. It was decided not to include the car type to the regression model as the number of doors a car has would not have any psychological influence on the speeding issue. The survey could have included the car type from where it could be potentially hypothized that owners of more expensive cars were more tempted to exceed the speed limits. The color of the car was also considered as 1= brown, 2 = black, 3 = blue, 5 = red, 6 = yellow and 7 = white. Other variables included the time of the day and month at which the person was caught driving to control for the possible rush hour affect and month influence, the exact time of the stop was eliminated to avoid double counting in the model. The two dummy variables included the fact whether the person speeding was the resident or none resident. The none residents were assigned 1 to hypothize that none residents no matter of their gender and car type and time of the day, day of the month and month, are more tempted to speed over the limit. The second variable is the gender to assume that at all other things being equal, male are more prone to speed over the limit than men.

Thus, the hypothesis was that there is the following relationship to be estimated:

a + ?1*Car_Color1 + ?2*Time_Of_Day + ?3-Month+ ?2* Day + ?3-Month + + ?2*Resident_Dummy + ?3 *Gender_Dummy...+ ?

The hypothesis is tested with the confidence level of 95%, thus the allowed chance of rejecting no relationship between the variables when there is actually this relationship, is 5%. Decreasing the confidence level to 90% will give more errors in the model and the model did not result in better relationship. Having carried out this multifactor regression, the result revealed that there is no statistically significant relationship between the over speeding and the fact that the person is a resident or non-resident and the gender of the person. The first problem with the model could be the very data set as out of the 536 observations in the population, only 136 were the cases when people were none residents. Thus, the results could be distorted. The R2 in the model is extremely low and reveals that very little variation in the severity of this crime could be explain by the factors in the model. P-values are low only for the intercept and none-residence factor.

Regression Statistics

Multiple R

Square