U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590
202-366-4000


Skip to content
Facebook iconYouTube iconTwitter iconFlickr iconLinkedInInstagram

Safety

FHWA Home / Safety / Pedestrian & Bicycle / Las Vegas Pedestrian Safety Project: Phase 2 Final Technical Report

Las Vegas Pedestrian Safety Project: Phase 2 Final Technical Report

< Previous Table of Content Next >

CHAPTER 4 DATA COLLECTION AT SITES BEFORE AND AFTER COUNTERMEASURE INSTALLATION

Data on the number of pedestrians crossing the street were collected at each of the 19 locations. Data were collected for five hours on each day. Of the five hours, two hours were during the morning and three hours of evening peak hours for vehicle traffic. At some locations, where the pedestrian sample is low due to the non-similarity of the vehicle and pedestrian peak volume times, data was collected for eight to twelve hours. The data collection days were primarily weekdays and data collection over the weekends was minimal. Weekend data collection was mainly intended for locations where pedestrian activities proximate to recreational and shopping areas are expected to be greater during the weekends. At other locations, such as the residential and small commercial locations, more pedestrian activities are expected during weekdays.

The pedestrians’ crossing behaviors were observed at a crosswalk and approximately within 200 feet from a crosswalk at all approaches of an intersection. All pedestrians were observed at mid-block locations, where distance from a crosswalk was not a deciding factor. The yielding behavior and whether a pedestrian was trapped or not trapped in the middle of the street while crossing were recorded. All observed pedestrian data were analyzed based on crossing locations. Both of the crossing behaviors consist of two options for each observation. The yielding behavior consists of two options, either “yielding” or “not yielding.” Likewise, the observation on pedestrians trapped in the middle of the street has two options either “trapped” or “not trapped” while crossing.

After the collection of various elements of the data, data was analyzed to determine the effectiveness of the countermeasure deployed. There are multiple countermeasures deployed at various sites to address multiple problems. Analysis of each site includes site description, aerial photo of the site showing injury and fatal crash locations, problems identified at that particular location, countermeasure proposed to improve the pedestrian safety at that location, countermeasures implementation details, data collection dates and analysis of the data collection at the respective locations. Data was also analyzed based on the type of countermeasure installed. Most of the proposed countermeasures were installed at more than one location. Therefore for each countermeasure, data from different sites was collected and analyzed to determine the overall effectiveness of the countermeasure. Analysis of individual sites and individual countermeasures follows.

Evaluation of Countermeasures

Several statistical tools are used to evaluate the effectiveness of the deployed countermeasures in enhancing pedestrian safety. The types of statistical tools are based on the considered measures of effectiveness (MOEs) for evaluation. The evaluation strategy and the statistical tools used for some of the countermeasures are discussed next:

A before and after study strategy was conducted to evaluate the effectiveness of most of the countermeasures. Data were collected in the morning and afternoon peak periods. This was done both prior and after the deployment of the above mentioned countermeasures (“before and after” condition). Data are stratified and analyzed for morning and evening peak hours based on total observations. The percentage of motorists yielding is obtained for both before and after study evaluation periods.

Z-Test

The z-test for two proportions, a statistical tool, is used to determine if the proportions obtained during the two study periods are significantly different.

Let PB = proportion of vehicles yielding during the “before” period

PA = proportion of vehicles yielding during the “after” period

The null hypothesis (H0) is that the percentage of motorists yielding during “before” period (PB) and “after” period (PA) is the same. The alternative hypothesis (Ha) is the percentage of motorists yielding during “after” (PA) period is greater than the percentage of motorists yielding during “before” period (PB). They are expressed as follows:

      H0: PB = PA

      Ha: PB < PA

The one-tail test for proportions is used to test these hypotheses at a 95 percent confidence level.

Let XB = number of vehicles yielding in the “before” period, out of a total of nB vehicles

      XA = number of vehicles yielding in the “after” period, out of a total of nA vehicles

The population proportions Accent Capital Letter P with Sub Capital Letter A and Accent Capital Letter P with Sub Capital Letter B are estimated by the sample proportions:

Equation  and Equation

For large sample sizes, the two sample proportions are approximately and normally distributed, and the z-test for testing the equality of the two proportions vs. the 1-sided alternative can be used. The test statistic used is Z0, and is defined as follows:

      Equation

Where,    Math Equation

Z0 is distributed approximately N (0, 1) when H0 is true.

The significant probability or P-value for equality of proportions vs. the 1-sided alternative is calculated by:

P-value = P(Z < Z0 )

The null hypothesis is rejected if the P-value < 0.05 (for 95% confidence level).

T-Test

A paired t-test and Welch-Satterthwaite t-test are used to compare if speeds are statistically different at two evaluation periods at the 95 percent confidence level. The Welch-Satterthwaite t-test is used when the assumption that the two populations have equal variances seems unreasonable. It is used to identify the difference between means of independent samples.

Let μB = population mean during before evaluation period,

      nB = number of observations during before evaluation period,

      Equation = sample mean of nB observations,

     Equation = sample variance of observations during before study.

Similarly, μA, nA,, and  are the population mean, number of observations, sample mean, and sample variance of after evaluation period, respectively.

The null hypothesis of equal means for “before” and “after” periods vs. the 1-sided alternative is expressed as:

      H0: μBμA = 0

      Ha: μBμA > 0

The test statistic computed from the sample is:

     Equation

The distribution of the test statistic when H0 is true is a t-distribution with approximate degree of freedom given by:

          Equation

 

The significance probability or P-value for equality of means vs. the 1-sided alternative is calculated by:

      P-value = P(tdf > t0)

If the obtained P-value is greater than the critical α-value, i.e., 0.05 at the 95 percent confidence level, then H0 is accepted. Similarly, if the P-value is less than the α-value, then H0 is rejected at the 95 percent confidence level.

 

< Previous Table of Content Next >
Page last modified on February 1, 2013
Safe Roads for a Safer Future - Investment in roadway safety saves lives
Federal Highway Administration | 1200 New Jersey Avenue, SE | Washington, DC 20590 | 202-366-4000