How the AJC/Channel 2 Action News poll was conducted


General information:

The poll was conducted for The Atlanta Journal-Constitution and Channel 2 Action News by the School of Public and International Affairs Survey Research Center at the University of Georgia from Sept. 30 to Oct .9, 2018, and included 1,232 likely general election voters in Georgia. Likely voters were respondents who had voted in the 2018 May primaries and/or the 2016 general election, who indicated they were currently registered to vote in Georgia, and were definitely or probably going to vote in the 2018 November election. Interviews were conducted in English. A dual-frame statewide random sample consisting of approximately 65 percent cellphone numbers and 35 percent landline numbers was obtained through Self Made Insights Co. (SMI is a sampling vendor that maintains a database constructed from state voter registration lists. Through commercial sources, phone numbers have been appended to the individual records (registrants) that make up these lists.) The survey results were weighted using iterative proportional raking in order to ensure the sample was representative of the 2018 midterm electorate in terms of race, sex and age. The calculated margin of error for the total sample is +/-2.8 percentage points at the 95 percent confidence level. This would mean that if 50 percent of respondents indicate a topline view on an issue, we can be 95 percent confident that the population’s view on that issue is somewhere between 52.8 percent and 47.2 percent. In 2014, turnout in the general election in Georgia equated to 42.9 percent of the total electorate (measured as the number of registered voters).

Verification:

Starting this fall, the SPIA Survey Research Center has implemented a new verification policy. Before results for any head-to-head election matchup are released to the AJC, they are corroborated by an outside entity. More specifically, the outside entity replicates our reported results for any head-to-head races and confirms they are the same. For this survey, the results we report for the gubernatorial election were verified by the Wason Center for Public Policy at Christopher Newport University in Newport News, Va.

Statistical calculations:

Governor: The difference between Kemp’s estimated vote share and Abrams’ estimated vote share, at 1.4 percent, is not statistically significant. In plain English, the gubernatorial race is currently a statistical tie.

Lt. governor: The difference between Duncan’s estimated vote share and Amico’s estimated vote share, at 6.1 percent, is statistically significant. In this race, one can be confident that Duncan is leading Amico.

Secretary of state: The difference between Raffensperger’s estimated vote share and Barrow’s estimated vote share, at 4.6 percent, is not statistically significant. In this contest, Raffensperger and Barrow are statistically tied.

M.V. “Trey” Hood III is director of the Survey Research Center at the University of Georgia School of Public and International Affairs.


Reader Comments

Next Up in Homepage