IV. PROGRAM EVALUATION
The evaluation involved a number of components. Various types of data were collected before and after each holiday campaign period using statewide telephone surveys, direct observations of drinking and driving at sobriety checkpoints before and after each holiday enhanced enforcement effort, DWI arrest data, and alcohol-related fatality data. Table 1 exemplifies each of the holiday campaign periods and the accompanying evaluation activities that spanned the entire campaign.
Table 1. Typical Holiday Campaign and Sustained Activities and Evaluation Efforts
|
Pre-Holiday Period |
Week 1 |
Week 2 |
Week 3 |
Post-Holiday Period |
|---|---|---|---|---|---|
Program Activities |
|
|
|
|
|
Enforcement |
|
|
|
|
|
Paid Media |
|
|
|
|
|
Earned Media |
|
|
|
|
|
Evaluation Activities |
|
|
|
|
|
Roadside Survey BAC Data |
|
|
|
|
|
Statewide Phone Survey |
|
|
|
|
|
Alcohol-Related Fatality Data |
|
|
|
|
|
DWI Arrest Data |
|
|
|
|
|
A. Connecticut Statewide Telephone Survey
TMR, Inc. conducted four statewide telephone surveys, one before and one directly after each holiday period for Connecticut. The survey telephone instrument (Appendix B) contained 30 questions and took about 20 minutes to complete. It was designed to measure drivers’ perceptions and behaviors related to drinking and driving and also included some questions regarding safety belt use and awareness of safety belt enforcement efforts. TMR, Inc., used the instrument to interview approximately 600 Connecticut drivers both before and after each holiday campaign period began for a total of about 1,200 interviews. A telephone survey conducted in March 2003 gathered baseline information and a survey conducted in July 2003 gathered information immediately after the July 4th enhanced three-week holiday enforcement period ended. For the winter campaign, the baseline telephone survey was conducted in October 2003 and the post-campaign survey was conducted in March 2004, about one year following the administration of the first survey.
The sampling plan was designed to ensure a representative sample of Connecticut drivers and used a random digit dialing approach. Random digit dialing ensures that households with unlisted numbers will be represented in the sample. The survey protocol provided that numerous attempts were made to contact selected households so that the survey could be completed. Individuals within the household were randomly selected to interview by choosing the person with the next or most recent birthday in the home. All Connecticut households were randomly sampled. Telephone survey responses were weighted by age group using the U.S. Census Bureau population estimates for Connecticut for each group.