2001 Progress Report: Tracking Deterrent Messages in Environmental EnforcementEPA Grant Number: R828827
Title: Tracking Deterrent Messages in Environmental Enforcement
Investigators: Kagan, Robert A. , Gunningham, Neil , Thornton, Dorothy
Institution: University of California - Berkeley
Current Institution: University of California - Berkeley , Australian National University
EPA Project Officer: Lee, Sonja
Project Period: June 6, 2001 through May 31, 2003
Project Period Covered by this Report: June 6, 2001 through May 31, 2002
Project Amount: $349,981
RFA: Corporate Environmental Performance and the Effectiveness of Government Interventions (2000) RFA Text | Recipients Lists
Research Category: Economics and Decision Sciences
The overall objective of this research project is to increase knowledge about the role of general deterrence in the implementation of U.S. environmental law. The specific objectives are to: (1) track the dissemination (in various communications media) of information about enforcement measures and legal penalties against violators; (2) survey similarly situated enterprises about their knowledge of, and response to, those deterrent messages; and (3) evaluate the salience of general deterrence messages as compared to other factors, as stimuli for compliance efforts by regulated business.
The research approach involves: (1) identifying and characterizing deterrent messages, based on legal sanctions announced by the U.S. Environmental Protection Agency (EPA); (2) mapping dissemination of the deterrent messages through content analysis of various media; (3) mapping reception of deterrent messages, via telephone survey; (4) assessing the behavioral impact of deterrent messages, via telephone survey; (5) assessing the salience of deterrent messages, via indepth interviews; and (6) analyzing the relative effectiveness of general deterrence enforcement strategies.
The expected results will provide insights into, and provisional answers for, a range of questions concerning: (1) the relative importance of general deterrence and the circumstances in which it is most likely to be achieved; (2) the extent to which, and the circumstances in which, deterrence messages "get through" to target populations; and (3) how salient deterrent messages are, as compared to other factors. It is hoped that the research results will facilitate better regulatory design, as well as facilitate effective integration of deterrence and non-deterrence oriented regulatory strategies.
Project Year 1 has focused on data collection. Data collection for Phase 1, identification and characterization of deterrent messages based on legal sanctions announced by the EPA, and mapping dissemination of the deterrent messages for Phase 2, have been completed. The data have been analyzed, and a working paper describing the results has been written.
Data collection for Phases 3 and 4 (mapping deterrent messages and determining their behavioral impact) has been completed for six industries: (1) sanitary treatment plants in California; (2) sanitary treatment plants in Florida; (3) electroplaters in Colorado; (4) aluminum manufacturers in three southern states; (5) steel fabricators in Illinois; and (6) chemical manufacturing or blending facilities in Kentucky.
The raw data have been coded into a database, and a preliminary analysis of the data has been conducted. A working paper, "General Deterrence and Corporate Environmental Behavior" currently is being written and will be presented at the Law and Society Association Meeting in Pittsburgh in May 2003.
Data collection for Phase 5 largely has been completed, with 28 completed in--depth interviews and 2 additional interviews scheduled. A preliminary analysis of the data collected in the first 28 interviews has been completed, and the first draft of a working paper has been written.
Difficulties and Changes. Our progress in the Phase 2 study of dissemination of deterrent messages was frustrated and primarily delayed by a change in the nature of the articles print publishers send to large electronic databases such as Lexis-Nexis. The change stemmed from a U.S. Supreme Court decision in June 2001, protecting the copyrights of freelance writers. Major newspapers no longer deposit in Lexis-Nexis articles republished from wire services or other newspapers. This, in turn, affected the reliability of our searches for the total number of articles relating to particular EPA press releases. Finally, we found the journals of smaller trade associations generally are not available in libraries or online.
We dealt with the reliability difficulty first by having a single researcher do a second round of media coverage searches, using a smaller sample (n = 40) of press releases, and a different set of databases (general internet searches, listserv searches, and the RDS Industry and Business database were not conducted; Lexis-Nexis regional news and wire database searches were added). Because of the U.S. Supreme Court decision and industry coverage issues, we did not believe that our media coverage metric was precise enough to assign each press a precise quantitative media coverage score. However, we did believe that it was precise enough to be able to identify some press releases as definitely receiving relatively high media coverage, and others as receiving relatively low media coverage. Consequently, the 40 press releases were ranked independently by both sets of search results.
First, the start of Phase 3 of the project (telephone survey of firms similar to the sanctioned firm) was delayed by the problems encountered in Phase 2. Second, sampling frames have been far more difficult to construct than originally anticipated. Databases with SIC code information and yellow pages databases often contain erroneous and outdated information, and we have found no single source that lists all facilities within an industry (EPA, http://yellowpages.com Exit , http://yahoo.com Exit , and RDS Business and Industry databases all contain overlapping lists of companies, but one list is never entirely nested in another list). Sample frame construction has involved tedious assemblage from multiple sources of the most complete possible sampling frame. Third, although the telephone interviews are only 15 minutes long, each interview generally takes 7 or 8 contacts to set up, and on occasion, can take more than 20 contacts.
First, to continue our efforts to maintain high response rates and yet still complete the project on schedule, we have decided to conduct surveys for a longer period than originally estimated. Second, we will decrease the number of different industries studied from 15 to 9, thereby decreasing the total number of interviews conducted from between 450 and 600 to between 270 to 360 (assuming between 30 and 40 interviews per industry).
We also have decided to abandon Phase 4B, which was designed to explore whether deterrent messages are associated with improvements in reported firm-level measures of environmental compliance and performance (in the same industry as the sanctioned firm). We originally intended to compare compliance rates with permits and inspections (as noted in EPA databases) for the 2 years before the announced legal sanction and for the period following it up to the present, as well as interrupted time series for air and water discharge data, to see if changes in trends occur at the time of media coverage. However, our media coverage research has revealed that this exercise would not lead to useful conclusions. Interrupted time-series methods assume an intervention (the deterrence signal) that is discrete in time. We found, however, that deterrence signals are not discrete in time. Media coverage of EPA deterrence signals includes coverage of the initial event that led to action, speculation about possible charges, coverage of the initial charge, speculation about possible penalties, coverage of the case or settlement, speculation about possible penalties, and coverage of the penalty. In addition, there often is no single deterrence message separate from others. In most industries, a number of similar cases occur in the year before and after the signal case. Thus, any "before and after" statistics regarding one deterrence signal are hopelessly confounded by other deterrence signals. Abandoning this phase, therefore, seems sensible, while also freeing up researchers' time to continue to conduct telephone surveys that, in Phase 4A, address the same research questions as Phase 4B.
With respect to Phase 5, entailing more in-depth interviews of firms in the same industry as those sanctioned by the EPA, our delays in obtaining cooperation for the 15-minute telephone interviews of Phase 3 have led us to anticipate some difficulty in obtaining good response rates to a survey that is designed to involve a 1 to 1.5-hour interview, particularly from people who work in small- or medium-sized enterprises. We addressed this issue by offering interviewees monetary compensation for their time, and offering to conduct interviews outside of business hours. We do not believe that offering compensation will bias the results of our survey because the research has no agenda interviewees might feel obliged to endorse because they have accepted money from the research team. We submitted this change in the proposal to the University's Human Subjects Committee and obtained permission to proceed with our study.
Conclusion. Although the aims of our project remain unchanged, the means of attaining those aims have been modified. The proposed changes to the original scope of the project, and the reasons for the proposed changes, are summarized in Table 1.
|Anticipated % Complete||Actual % Complete of Original Scope||Actual % Complete of New Scope||Proposed Change in Scope||Reason|
|Phase 2||100%||100%||100%||Decrease in sample size from 60 to 40.||Inter-rater reliability.
Supreme court decision.
|Phase 3 and 4A||100%||56%||90%||Decrease in sample size from 450 to 270 surveys.||Delayed start due to Phase 2 overun.
Sampling frame delays.
Response rate delays.
|Phase 4B||75%||0%||NA||Propose abandoning scope.||Phase 2 revealed methodology inappropriate.|
|Phase 5||95%||95%||95%||Propose paying people to participate ($100/person).||Response rate delays.|
|Phase 6||10%||50%||50%||None.||Have completed preliminary analysis of all 5 phases.|
We are extremely pleased by the quality of data we have gathered, and the illumination it has provided. In a recent assessment of our project's results and status, we determined that so many aspects of our findings are both new and interesting that we probably will have to write two or three more articles than originally planned to explore the richness of the data and the issues they address. Thus, despite some delays and changes, we remain excited and optimistic about the project as a whole.
Future activities are to: (1) complete telephone surveys for another two (or three, if possible) industries; (2) complete the last few in-depth semi--structured Phase 5 interviews; and (3) complete the paper and report writing for all five phases of the project.