Public Perception of Safety Messages and Public Service Announcements on Dynamic Message Signs in Rural Areas

Chapter 4. Analysis Methodology

The statistical analysis of the information collected from travelers as part of this test was performed using SAS™ statistical analysis software. For each question in the questionnaire, means and standard deviations (for continuous responses) or contingency tables (for categorical responses) were prepared. In addition, graphical summaries (histograms, mean and confidence interval plots, etc.) were prepared for select questions. Statistical procedures were used to create confidence intervals and to compare responses for a particular question. Wherever possible, appropriate population weighting was applied to group statistics.

Comparisons between groups of respondents, such as between the four study corridors, were completed using Analysis of Variance (ANOVA) techniques that included as explanatory factors traveler demographics, traffic conditions, time of day/day of week, etc. The inclusion of these other factors and the use of a general linear modeling approach facilitated the ability to detect significant differences after accounting for different sources of variation when compared to simple t-tests or categorical frequency table analysis. Survey weighting procedures were utilized in the models whenever possible.