Those conducting peace polls should always use generally accepted scientific methods. However, there are a number of good practices that apply specifically to peace polls to ensure impartiality.
Management and control of the project
The key ingredient for the successful management of a peace poll is for those responsible for the peace poll to follow the ethical and best practice guidelines provided here. A peace poll can be carried out by a variety of individuals and organizations, for example: a university lecturer and his or her students, a polling company, newspaper, NGO, state government or inter-governmental body such as the United Nations Development Program (UNDP). In all cases those responsible for the maintenance of standards in those organizations (e.g. university ethics committee, NGO board of trustees) should make sure these ethical and best practice guidelines for peace polls are followed at all times. Beyond this fundamental principle of management the widest possible consent of all the parties to the conflict, with regard to all management decisions that must be made, is to be encouraged. In addition to agreement of the questions to be asked issues relating to sampling, funding, timing, writing up the results and dissemination should be made collectively whenever possible. In practice an initial peace poll undertaken by individuals may function quite informally on all these points but if the research develops into an important part of a negotiations/peace process then it may be necessary to establish a steering committee to oversee these activities in collaboration with those responsible for the negotiations. The inclusion of the most eminent and respected local and international scholars and practitioners in such a committee is to be encouraged, as they will help to give authority to the findings of the research and any difficult and politically sensitive decisions that may have to be made.
Peace polls often require data collection in difficult environments where extensive local knowledge and experience is essential to the success of the project. But as local polling organizations frequently only have such expertise in their own community it is often necessary to put together a consortium of organizations that can pool their intellectual and field resources across a variety of populations. If such cooperation proves difficult to establish, then it may be helpful to introduce a neutral third party into the research team with peace poll management experience.
As with the research team costs can be shared by those undertaking the research in the various communities effected by the conflict and/or media interests that represent the different communities. Alternatively a neutral third party can cover such costs providing they are acceptable to all of the parties to the conflict.
Impartiality in question design can be achieved by involving all the relevant parties to the conflict in the design and agreement of the questions asked. Focus groups are often used for this purpose; but where highly sensitive and politically charged issues are discussed, it is often better to conduct such interviews in a 'one-on-one' format with such interviews undertaken by persons from the informants community and/or a neutral third party. Additionally as the research is being undertaken to assist the parties to bring an end to their conflict, it is most helpful to include them in the questionnaire design process and, where circumstances allow, give them 'ownership' of that process. Finally as the objective of a peace poll is to help the parties to a conflict achieve peace, it is necessary to move beyond simple abstract judgments about the conflict and parties to the conflict to deal with the problems that caused the conflict and appropriate solutions to those problems that can be translated into policies and agreements. Involving political elites and their staff in such activities can be very productive although such involvement may have to be anonymous. Confidentiality will give political elites opportunities to test controversial policies against public opinion without running the risk of criticism from 'spoilers'.
Partisan polls frequently 'cherry pick' the results by selectively releasing results in order to present a biased picture of public opinion that favours the position of those who commissioned the poll. Such biased framing of the issues at stake can also occur at the questionnaire design phase of the research by selectively asking different questions to different communities. This practice should be avoided by making every effort to ask the same questions of all the communities surveyed with an explanation provided when this cannot be done. For the same reasons, all the communities and peoples effected by a conflict should be included in the research process.
Defining relevant populations
Conflicts (particularly in the popular press) are frequently defined in simple polar 'them and us' terms. Inevitably such simplifications tend to ignore the views of the 'silent majorities' in many populations and/or significant minorities whose views are often shaped by their need to live with the various parties in conflict. Conversely identifying radical groups and the extent to which their views are marginal and not mainstream is equally important. For all these reasons the sample and its demographics should be as representative of all relevant groups as resources will allow.
General samples of populations cannot always generate statistically significant samples of minorities whose views may be essential to the understanding of a conflict (e.g. terrorist groups). In these circumstances over samples should be collected of such groups along with a clear explanation of how the sample was generated.
Ethnic conflicts frequently involve populations and subpopulations that speak different languages. Every effort should be made to use the double translation method to provide copies of the questionnaire in all the relevant languages. When this is not possible, an explanation should be given about how the fieldwork was conducted so as to provide the respondents with the best possible opportunity to participate in the research in the language of their choice.