Introduction

     I have served both as an independent scholar doing field research and as an evaluator doing work for an agency pursuant to a contract (sometimes called contract research in the university setting). In the course of my career, I have worked with data from the American Arbitration Association, the United States Postal Service, the U.S. Department of the Air Force, the National Institutes of Health, the U.S. Department of Agriculture, the U.S. Occupational Safety and Review Commission, the U.S. Department of Justice, and JAMS. The following scenarios reflect general lessons learned; the events described have been to some extent fictionalized to avoid attribution to any particular agency or provider. I have included a complete bibliography of my published work for those who would play sleuth to match scenarios with agencies.

     I believe that this kind of research is critically important. There is increasing privatization of justice through dispute resolution systems (Bingham 2004a; Lipsky et al. 2003). The reality of employer one-party control over their design raises public policy issues that are central to labor and employment law (Bingham 2002a, 2002c, 2004b). Only systematic, rigorous, multivariate research designs can inform the development of future policy (Bingham 2002b, 2002c; Lipsky and Avgar 2004).

Scenario One: Why Are You Out to Get Us?

        This scenario is about work in which you are examining outcomes in a sample comprised of individual dispute resolution cases. Typically, researchers look at outcomes in the aggregate, in relation to some variable of interest, like gender or race. However, that sample of cases comes from somewhere; it may come from an individual neutral, a group of neutrals, or a service provider. Thus, there is the risk that your research will be construed as reviewing the performance of an individual neutral in the sample, be it mediator or arbitrator.

     People in the field of dispute resolution are passionate about what they do. They believe it to be holy work. They see themselves as the good guys trying to help people work out problems and save emotional energy and money. They identify with being peacemakers. They tend to take evaluation very personally. This is not simply about whether a program or process is effective. A criticism of an outcome is a criticism of the neutral. It may go to that neutral's core sense of self. A third party service provider will present the same sort of problem, only scaled up to the organization's mission. The people in that organization may be passionately devoted to that dispute resolution mission and take personally any research finding as a reflection on their work. I have literally had people ask on a phone call, "Why are you out to get us?"

     The problem is that the average neutral thinks deeply about their practice at the level of the individual case but thinks less deeply about the dispute system design within which that case occurs. Often, neutrals in private practice have a series of cases from a variety of disputants and many different dispute system designs. For example, an individual arbitrator may do labor cases in the public and private sector, employment cases involving highly compensated executives pursuant to an individual employment contract, or adhesive mandatory employment arbitration pursuant to a personnel manual. That same arbitrator may do court-referred cases involving civil litigation. The Lone Ranger metaphor is appropriate: the neutral rides in, resolves the dispute, and rides off into the sunset.

     Thus, a key public relations task is to explain to the neutrals that your research or evaluation is not about their individual performance and that results will only be reported in the aggregate, without revealing the identity of any one neutral. Moreover, if you are looking at patterns in pooled outcomes that might be perceived to reflect adversely on the neutral or provider, some up-front disclosure about the nature of the research design, hypotheses, and range of possible findings may be in order. On the other hand, this may mean you never get access to the data.

Scenario Two: With Friends Like You . . .

     Let us assume you got access to data, did research, and published the results in reputable outlets after an appropriate period of review. Let us also assume that these results create a controversy in the press. As a researcher, you may be a staunch advocate of dispute resolution. You may be working to improve practice, improve justice, develop best practices in program design, and foster the development of the field. However, your research is hard to "sound bite" with accuracy; the press takes a complex phenomenon that is not well understood and oversimplifies it; or, they just get it wrong.

     In this scenario neutrals and providers who have taken the high road in support of research may find that it has come back to bite them. It is hard to set ground rules about this. For example, in labor negotiations there is a ground rule that there will be no press releases except by prior mutual agreement, or that all press releases shall be jointly issued. This is difficult for scholarship; the chief purpose of research is to build knowledge through publication and exchange of ideas. If you cannot publish research, what is the point? For this reason, many scientists refuse to do research for the Department of Defense because it may become classified.

        As a practical matter, it is always advisable to give any organization that has provided data a period of time to review the data analysis or report, respond, make suggestions, and offer alternative explanations or theories for your results. Another possible compromise would be for researchers who have published controversial results to offer to have the organization participate in any press interviews on the research.

        However, savvy organizations have a more effective response; simply explain that you have used the results of the research or evaluation to correct the problems or improve your program. You now have a different program from the one the researcher evaluated. This can quickly defuse criticism.

Scenario Three: Just Say No

     Neutrals and third-party providers sell a service. A key feature of that service is confidentiality. This, in turn, is a critical barrier to systematic, comprehensive, quantitative field research on dispute resolution. When you think about it, what is in it for them? Why on earth should neutrals and dispute resolution service providers release information that is legally proprietary and confidential? The best way to avoid having to explain that some research result does not really reflect badly on your practice or your roster is to never allow the research in the first place. As the debate over the privatization of justice and fairness in dispute resolution programs has heated up, there has also been a growing propensity among third-party providers to "just say no" to research.

     The problem with this approach is that the debate over privatization of justice is not simply hypothetical. People are being denied access to small claims courts because of adhesive arbitration clauses in consumer contracts (Bingham 2004b). There is a broad duty to the public to make the process of dispute resolution, the rules under which it occurs, and the programs through which it is offered as fair and just as humanly possible. It is for this reason that the California state legislature mandated that third-party providers disclose the outcomes of arbitration cases. In theory, this disclosure would make it possible for researchers to analyze patterns in the cases. In reality, however, third-party providers are not disclosing critical aspects of the cases, such as which party won (Bingham, Sternlight, and Healey 2005). At some point, there needs to be a compromise between confidentiality and disclosure in the public interest. Another way around this problem is to go after the data from the end user instead of the neutral or provider.

Scenario Four: A Change in Administration

     Let us assume you are no longer working with neutrals or providers but are instead embarking on a relationship with an agency or organization that is developing an in-house dispute resolution program for workplace conflict. You believe in collaborative evaluation design for a number of reasons. First, you feel you are more likely to get buy in if the people collecting the data have a say in what they collect. Second, for an evaluation to be useful to the agency, it should include indicators that measure progress toward the agency's goals, not simply a cookie-cutter set of variables or questions from other evaluations. Not all dispute resolution programs share the same goals. If the agency wants to reduce dispute processing costs, that suggests one set of measures; however, if it wants to improve the conflict management skills of employees and supervisors, that requires a different set of measures. So assume you allow the agency to convene a group of stakeholders to design the evaluation with you. You work for a year or more to reach agreement on a set of goals, variables, indicators, instruments, and protocols to collect data. Just as you are about to roll out the evaluation, there is a change in the leadership of the program. The new head honcho has played no role in designing the evaluation. He or she fails to see how it might be helpful.

     Moreover, let us assume that the evaluation attempts to use a more rigorous quasi-experimental design, one with experimental sites and control sites, which means that the hot new dispute resolution program will not be available to all employees in the agency at the same time. The new program manager sees this as delaying his or her ability to both demonstrate the effectiveness of dispute resolution and to make progress in conflict management at the agency. The program manager may also foresee political risks for him or herself if employees who do not have access to the program feel this is inequitable and blame program leadership. As a result, the program manager vetoes the evaluation and refuses to roll out data collection.

     You have no data, no prospects of data in the future, and have just wasted a year of your professional life. If you are a junior academic, this is a year during which you should have been publishing. While academic institutions recognize that it can require lead time to establish data collection for a new research project, the failure to publish during that period can be fatal to an academic career. And now there is no prospect for publishable research out of this collaboration. It is an ugly scene and illustrates the risks of embarking on evaluation as a means of data collection if you are a junior academic. Remember: this can happen to you, and it can happen at any time. It can happen to senior researchers after a decade-long collaboration, with little or no notice. It happened to me twice.

     One counter-argument that is sometimes persuasive is to emphasize the positive effect that the mere existence of an evaluation project has on a dispute resolution program. Employees are more likely to believe the program is neutral and credible if an outsider is watching. It also enhances their sense of voice to have the agency ask them for their assessment of their experience in the program. Federal Interagency ADR Working Group Best Practices emphasize the desirability of evaluation. The Government Performance and Results Act requires performance measurement. These arguments may persuade a new manager; or they may not.

Scenario Five: Sabotage

     Let us assume you have strong top-down support for data collection. You have designed the evaluation collaboratively with an agency stakeholder group. You roll out data collection and distribute all the instruments and wait for the data to start arriving in the mail. And you wait. And you wait. There is no data, or only a trickle of data. Your records indicate you have a very low response rate, perhaps 10 percent. This means your sample is not usable; it cannot tell you anything meaningful about the program. Any analysis you do of this small sample might produce results that are simply a reflection of selection bias; they may only be the views of a skewed handful of participants in the program. What has happened?

     It is sabotage. For reasons you failed to anticipate, there are people in the data collection chain who fear the results of the evaluation. For example, let us assume that you are evaluating mediation as an alternative to the traditional Equal Employment Opportunity (EEO) process for complaints of prohibited discrimination. You have data collection points both after the mediation process and after an EEO counselor attempts to conciliate the complaint. Essentially, this might permit a comparison of disputant satisfaction with mediators as compared with EEO counselors. Given what we know about the persistent high satisfaction with the mediation process and mediators, we can anticipate ex ante that EEO counselors will be the losers in this comparison. These EEO counselors engage in passive aggressive behavior; they simply do not collect the data. Depending on the size of the agency and the number of data collection points, it may require more enforcement than managers are capable of doing to ensure data collection.

     But, you say you had a representative stakeholder group, including EEO counselors, participate in the design! Why are they not cooperating now? There are a variety of reasons. Certainly, data collection adds to people's daily workload, and without understanding and support for that burden, this may present problems. For example, the people you expect to collect data are being evaluated not on whether they get it to you but whether they meet primary agency productivity goals. Another explanation is that, with strong top-down support for both the dispute resolution program and the evaluation, it was not politically correct or safe for them to express their fears about the results. It is simpler to keep heads down and then ignore this as another meaningless bureaucratic directive.

     In another scenario, you may have an agency with multiple units and ADR programs, each with a different dispute system design. You have designed a uniform system of data collection for all the programs so that you can systematically compare their effectiveness. However, that comparison might make some alternative dispute resolution (ADR) program managers look good at the expense of others. What is in it for them? Hence, you have another case of passive resistance to data collection.

     In yet another scenario, employees responsible for data collection are aware of an impending need to reduce force and reallocate resources or budget. If they engage in data collection, and their program does not fare well in the comparison, they may lose lines and resources, making it even harder for them to meet what they perceive to be their responsibilities. Particularly when complaints of discrimination are concerned, EEO program managers may have very strong views of justice and of the need for meritorious complaints to make their way through the adversarial system to provide a deterrent to discriminatory behavior. They may resent the new resources put into ADR programs, arguing that if they were given additional resources, they could produce better results.

     Sometimes, sabotage is not passive-aggressive; it is aggressive-aggressive. For example, sometimes the unions at a workplace may advocate boycotting data collection. A dispute resolution program that is part of an EEO complaint handling process may be exempt from collective bargaining under Alexander v. Gardner Denver and thus adopted unilaterally by the employer. The union may view it, at least initially, as undermining their position as the elected employee representative. Over time, in my experience unions come to see these programs as an asset, and indeed, as a second and sometimes more appropriate bite at the apple than a traditional grievance under the collective bargaining contract. Nevertheless, a union-ordered boycott of data collection can result in selection bias in your sample.

     It is important to build incentives and safeguards into the evaluation design. For example, the United States Postal Service (USPS) decided to use voluntary rates of participation in mediation as one criterion for evaluating managers and deciding on their bonuses. This created an incentive to support the program and data collection. The USPS also explained to EEO counselors that exit surveys would be reported in the aggregate, by zip code and not office, and that the survey would not be used in any way for performance evaluations. However, to create an incentive to cooperate, the USPS indicated that, depending on the evaluation, it was likely to change some EEO jobs to ADR counselor jobs, and when that happened, existing EEO counselors would get priority in bidding on those jobs. These jobs became desirable, as they had more positive connotations.

Scenario Six: Oh, Yeah, We Changed That . . .

     We have a program evaluation. It is up and running. We have data collection. We have a good response rate and we almost have a usable sample. Somewhere along the line, program administrators forget about the evaluation and keeping you in the loop. They decide to change dispute resolution service providers, and in the process, change mediation models. This means that you are no longer accumulating data on the first model; unbeknownst to you, the data you receive now addresses the second model. For example, the original program design may call for outside neutral mediators. Because this costs money, the agency, in an effort to cut the budget, decides to train in house employees to become mediators. The limited data suggests that employees will perceive the fairness of mediation by co-workers somewhat differently than they will view the fairness of an outside neutral. This change in program design may have a significant impact on the results of the evaluation. If you report results without taking this into account or controlling for it, it may damage your reputation.

     If your evaluation is longitudinal, and if you have consistent data collection, this can be an opportunity for a natural experiment instead of a cause for alarm. The key is getting a big enough sample size before the change is made.

Scenario Seven: What Is an "Exogenous Variable"?

        You are collecting data on complaint filing rates before and after the program was implemented, another kind of natural quasi-experiment. The agency takes some significant managerial action unrelated to the merits of the program, for example, a big reduction in force due to a Congressional budget freeze or continuing resolution. Typically, a reduction in force (RIF) is associated with a spike in complaint filings in the various available grievance procedures in a dispute system design. No one bothers to tell you about this RIF. Remember to check carefully on changing conditions on the ground in your agency before you report program results. If you report the complaint spike without controlling for the RIF, you will damage the credibility of the program.

     Again, in longitudinal studies this event may be an opportunity rather than a problem. For example, September 11, 2001, saw the demise of the Chambers Street Post Office, which was destroyed with the World Trace Center complex. Moreover, October 2001 was the period during which there was a terrorist incident involving anthrax through the mail. Ultimately, anthrax was found to have contaminated facilities throughout the New Jersey and District of Columbia regions, and in many instances, employees were asked to take the antibiotic Cipro as a protective measure after one postal worker sickened and died. These events now provide an opportunity for a time-series study of the impact of a terrorist act on an ADR program and its filing rates.

Scenario Eight: "We Can't Collect That; We'll Get Sued"

     You want to collect a dataset that you can use for rigorous, refereed, social science research. For you, important variables include the race and gender of disputants and mediators. However, the agency says, "We can't collect that; we'll get sued. Some lawyer might get discovery of the surveys and results in a class action. How would it look if white women like the program a lot more than black men?"

     There are other political arguments against collecting demographic data. In one case, a previous researcher poisoned the well. The researcher was studying the results of workplace drug testing. He or she received access to drug test results along with demographic data linked to those results. The way this story was told to me, the researcher published a study showing that one demographic group, African Americans, tests positive at a more frequent rate than another group, for example, Caucasian women. This gave rise to a major controversy in which the unions, representing aggrieved workers, filed various complaints against the employer. As a result, the agency categorically refused to collect demographic information as part of any evaluation. Alternative partial information on demographics may be available from the Census Bureau statistics on the demographic pattern where the workplace is located, or in the EEOC annual report that every employer submits. However, this information is not linked to program outcomes.

Scenario Nine: Low Response Rates

     Another political problem may arise when data collection has yielded low response rates, for example, 10 to 20 percent. You may do preliminary analyses on this dataset and find that the results are promising and positive. You share these results with management, at the same time cautioning that there is a low response rate so you cannot release the results publicly. The problem is the standards of some evaluators are regrettably not the same as those for peer review in academic journals. Many evaluators will release these findings, and the agency will happily take credit for a successful program. If you argue about the reliability of the sample and selection bias, they may respond that you are working for them, not for the greater academic community, and this is good enough for government work. The solution here requires up-front planning; the evaluation contract may need to contain a provision for a minimum response rate.

Scenario Ten: "But These Results Make Us Look Bad"

     Closely related is the scenario in which evaluation results are somehow negative or perceived as politically sensitive by the agency. Assume that you prepare a careful report and submit it to the agency. The response rate is good, and you have faith that the results fairly reflect the performance of the program. However, program administrators are convinced that if they release these results, they will jeopardize the program or its budget. They thank you for your report and promptly consign it to the circular file. They do not use your report to make program improvements or changes to address the issues you have identified. Distressed, you ask for permission to turn the report into an article for publication. However, you do not have an express right to publish the results in your contract, and the agency tells you not to.

     You have two choices. If the agency is part of the state or federal government, your report may be a public record under the state or federal Freedom of Information Act. You may either request its release yourself or ask a friend to obtain it. They may argue it is a draft and not a final, approved report, and you will also permanently destroy your relationship with this agency. Alternatively, you can simply view yourself as an evaluator for hire, and one who has fulfilled the terms of the contract; you can just walk away. Moreover, if the organization is a private sector one, there is no freedom of information right to access their records.

     To avoid this problem, it is best to negotiate an exclusive right to publish the results of your evaluation into the contract terms. It is good client relations to also negotiate a provision giving the agency a reasonable period to review the draft publication, for example, thirty to sixty days.

Scenario 11: "It's Too Good to Be True"

     The last political problem I have to share happens after the evaluation is complete. You publish and share the results, and someone says, "It's impossible; it's too good to be true." This happened to me in South Korea at the first ever national conference of the National Labor Relations Commission after I presented the USPS results. I am not sure what you can do other than offer others the opportunity to inspect the databases. The bottom line is that a mediation program, properly designed and implemented, can have a substantial positive effect on the workplace.

Conclusion

     Field research and evaluation of employment dispute resolution programs is a demanding and challenging enterprise. However, many of the scenarios I describe can be avoided by negotiating the terms of the project carefully at the outset. The great majority of problems I have encountered are not matters of ethics; they are simply pragmatic responses to the practical and political implications of data. By having a candid and thorough conversation up front, many of these problems can be avoided.

Acknowledgments

     Much of the research described here and listed in the Works Consulted was supported in whole or in part by a grant of general support to the Indiana Conflict Resolution Institute at Indiana University's School of Public and Environmental Affairs from the William and Flora Hewlett Foundation's Conflict Resolution Program. I wish to thank its former program director, Terry Amsler, as well as Cynthia J. Hallberlin, formerly of the USPS, for their advice and assistance on the institute's research program. Any errors are my own.

Works Consulted

Anderson, Jonathan F., and Lisa B. Bingham. 1997. "Upstream Effects from Mediation of Workplace Disputes: Some Preliminary Evidence from the USPS." Labor Law Journal, Vol. 48, pp. 601Ð615.

Bingham, Lisa B. 1995. "Is There a Bias in Arbitration of Nonunion Employment Disputes? An Analysis of Actual Cases and Outcomes." International Journal of Conflict Management, Vol. 6, no. 4, pp. 369Ð386.

Bingham, Lisa B. 1996. "Emerging Due Process Concerns in Employment Arbitration." Labor Law Journal, Vol. 47, no. 2, pp. 108Ð126.

Bingham, Lisa B. 1997a. "Employment Arbitration: The Repeat Player Effect." Employee Rights and Employment Policy Journal, Vol. 1, no. 1, pp. 189Ð220.

Bingham, Lisa B. 1997b. "Mediating Employment Disputes: Perceptions of REDRESS at the United States Postal Service." Review of Public Personnel Administration, Vol. 17, no. 2, pp.20Ð30.

Bingham, Lisa B. 1998a. "McGeorge Symposium on Arbitration: On Repeat Players, Adhesive Contracts, and the Use of Statistics in Judicial Review of Arbitration Awards." McGeorge Law Review, Vol. 29, no. 2, pp. 223Ð260.

Bingham, Lisa B. 1998b. "An Update on Employment Arbitration and the Courts." Perspectives On Work, Vol. 2., pp. 25Ð29.

Bingham, Lisa B. 1998c. "An Overview of Employment Arbitration in the United States: Law, Public Policy and Data." New Zealand Journal of Industrial Relations, Vol. 23, no. 2, pp. 5Ð19.

Bingham, Lisa B. 2001. "Addressing the ÔRedress': A Discussion of the Status of the United States Postal Service's Transformative Mediation Program." Cardozo On-Line Journal of Conflict Resolution, Vol. 2. Accessed from http://www.cardozo.yu.edu/cojcr/final_site/symposia/vol_2_symposia/postal_trans.htm.

Bingham, Lisa B. 2002a. "Self-Determination in Dispute System Design and Employment Arbitration." University of Miami Law Review, Vol. 56, no. 4, pp. 873Ð908.

Bingham, Lisa B. 2002b. "The Next Step: Research on How Dispute System Design Affects Function." Negotiation Journal, Vol. 18, no. 4, pp. 375Ð379

Bingham, Lisa B. 2002c. "Why Suppose? Let's Find Out: A Public Policy Research Program on Dispute Resolution." Journal of Dispute Resolution, Vol. 2002, no. 1, pp. 101Ð126.

Bingham, Lisa B. 2003. Mediation at Work: Transforming Workplace Conflict at the United States Postal Service. Arlington, VA: IBM Center for the Business of Government.

Bingham, Lisa B. 2004a. "Employment Dispute Resolution: The Case for Mediation." Conflict Resolution Quarterly, Vol. 22, nos. 1and 2, pp. 145Ð174.

Bingham, Lisa B. 2004b. "Control over Dispute System Design and Mandatory Commercial Arbitration." Law and Contemporary Problems, Vol. 67, nos. 1 and 2, pp. 221Ð251.

Bingham, L. B., and D. R. Chachere. 1999. "Dispute Resolution in Employment: The Need for Research." In A. E. Eaton and J. H. Keefe, eds., 1999 Industrial Relations Research Association Research Volume: Employment Dispute Resolution and Worker Rights in the Changing Workplace. Champaign, IL: Industrial Relations Research Association, pp. 95Ð135.

Bingham, L. B., G. Chesmore, Y. Moon, and L. M. Napoli. 2000. "Mediating Employment Disputes at the United States Postal Service: A Comparison of In-house and Outside Neutral Mediators." Review of Public Personnel Administration, Vol. 20, no. 1, pp. 5Ð19.

Bingham, Lisa B., Kiwhan Kim, and Susan Summers Raines. 2002. "Exploring the Role of Representation in Employment Mediation at the USPS." Ohio State Journal of Dispute Resolution, Vol. 17, no. 2, pp. 341Ð377.

Bingham, L. B., and D. Mesch. 2000. "Decision-Making in Employment and Labor Arbitration." Industrial Relations, Vol. 39, no. 4, pp. 671Ð694.

Bingham, L. B., T. Nabatchi, and M. S. Jackman. 2000. "Evaluation of the Occupational Safety and Health Review Commission Settlement Part Program." Available at www.spea.indiana.edu/icri/oshrc-eval.pdf.

Bingham, L. B., T. Nabatchi and M. S. Jackman. 2001. "Evaluation of the Occupational Safety and Health Review Commission E-Z Trial Program." Available at www.spea.indiana.edu/icri/oshrc-eztrial.pdf.

Bingham, L. B., and L. M. Napoli. 2001. "Employment Dispute Resolution and Workplace Culture: The REDRESSï Program at the United States Postal Service." In M. Breger and J. Schatz, eds., The Federal Alternative Dispute Resolution Deskbook. Washington, DC: American Bar Association, pp. 507Ð526.

Bingham, Lisa B., and M. Cristina Novac. 2001. "Mediation's Impact on Formal

Complaint Filing: Before and after the REDRESSŠ Program at the United States Postal

Service." Review of Public Personnel Administration, Vol. 21, no. 4, pp. 308Ð331.

Bingham, Lisa B., and David W. Pitts. 2002. "Research Report: Highlights of Mediation at Work: Studies of the National REDRESSñ Evaluation Project." Negotiation Journal, Vol. 18, no. 2, pp.135Ð146.

Bingham, Lisa B., and Shimon Sarraf. 2004. "Employment Arbitration Before and after the Due Process Protocol for Mediation and Arbitration of Statutory Disputes Arising out of Employment: Preliminary Evidence That Self-Regulation Makes a Difference." In S. Estreicher and D. Sherwyn, eds., Alternative Dispute Resolution in the Employment Arena: Proceedings of New York University 53rd Annual Conference on Labor. The Hague: Kluwer Law International, pp. 303Ð329.

Bingham, L. B., J. Sternlight, and J. Healey. 2005. "Arbitration Data Disclosure in California: What We Have and What We Need." Paper presented at the ABA Section of Dispute Resolution Conference, Los Angeles, CA, April 15.

Lipsky, David B., and Ariel C. Avgar. 2004. "Commentary: Research on Employment Dispute Resolution: Toward a New Paradigm." Conflict Resolution Quarterly, Vol. 22, nos. 1Ð2, pp. 175Ð190.

Lipsky, David B., Ronald L. Seeber, and Richard D. Fincher. 2003. Emerging Systems for Managing Workplace Conflict. San Francisco: Jossey-Bass.

Nabatchi, T., and L. B. Bingham. 2001. "Transformative Mediation in the United States Postal Service REDRESSï Program: Observations of ADR Specialists." Hofstra Labor and Employment Law Journal, Vol. 18, no. 2, pp.399Ð427.

Raines, S. S., T. Hedeen, L. M. Napoli, and L. B. Bingham. 2003. "A Tale of Three Cities: Communication and Conflict Before and after REDRESS." Report submitted to the United States Postal Service.