Have political attacks against pollsters in the Republic of Georgia worked: Evidence from a quasi-experiment and lab experiment replication

The Georgian Dream Coalition (GD) has criticized the National Democratic Institute-Caucasus Research Resource Centers (NDI-CRRC) public opinion surveys since first entering the electoral field in 2012. Criticisms have generally combined vague questioning of the survey methodology or brief expressions of misinformation which associated the polling organization with the formerly governing and now opposition United National Movement (UNM). But have the attacks worked, and what can the attacks tell us about political attacks and negative campaigns more generally? Are there spillover effects to attitudes towards other institutions? This paper will present evidence to begin to answer these questions based on an October, 2015 quasi-experiment and an April, 2016 lab experiment. Negative campaigns have been the subject of intensive study in political communications research over the last 30 some years, yet research has been inconclusive and focused largely on the United States. Moreover, negative campaign research has focused largely on electoral campaign advertising to the neglect of other negative political campaigns. This paper will present research on the effects of a negative campaign outside of the United States which is not specifically aimed at a political organization. The quasi-experiment took place in October, 2015. On October 17th, NDI-CRRC released its regular public opinion survey, and members of the currently ruling Georgian Dream Coalition including then Prime Minister Irakli Gharibashvili attacked the results of the polls. Several days prior, Caucasus Barometer 2015 fieldwork, an annual poll carried out in Armenia and Georgia, had started. The survey contained a number of questions about attitudes towards public opinion polls and their role in society. This confluence of events allows for the examination of the Caucasus Barometer data using a quasi-experimental regression discontinuity design. The paper will present evidence related to the following hypotheses: H1: Trust in public opinion polls declined following the political attacks; H2: Response rate declined, due to increased distrust of the polls following the political attacks; H3: The political attacks led to spillover effects for trust in institutions. In order to attempt to replicate findings, a lab experiment will be carried out in the Republic of Georgia in April of 2016. Respondents will be divided into three treatment groups and one control group. Respondents in the treatment groups will be asked to listen to a short radio broadcast which describes a politician’s attacks on pollsters. In one treatment group, the politician’s party will not be listed, in the second treatment group the politician will be identified as part of the ruling party, and in the third treatment group, the politician will be identified as a member of the opposition party. This design will enable the testing of the above hypothesis and thus provide evidence either in support of or against the results of the quasi experiment as well as allow for the testing of an additional hypotheses: H4: Trust in negative political campaigns is mediated through partisanship. This paper will examine the role of negative campaigns using a quasi-experiment and lab experiment replication.