Any rigorous discussion of bias in opinion surveys requires a scalar measure of survey accuracy. Martin, Traugott, and Kennedy (2005, Public Opinion Quarterly 69: 342-369) propose such a measure A for the two-party case, and Arzheimer and Evans (2014, Political Analysis 22: 31-44) demonstrate how measures A'i, B, and Bw for the more common multiparty case can be derived. We describe the commands surveybias, surveybiasi, and surveybiasseries, which enable the fast computation of these binomial and multinomial measures of bias in opinion surveys. While the examples are based on pre-election surveys, the methodology applies to any multinomial variable whose true distribution in the population is known (for example, through census data).
ArzheimerK., and EvansJ.2014. A new multinomial accuracy measure for polling bias. Political Analysis22: 31–44.
2.
CressieN., and ReadT. R. C.1989. Pearson's X2 and the loglikelihood ratio statistic G2: A comparative review. International Statistical Review57: 19–43.
3.
DurandC.2008. The polls of the 2007 French presidential campaign: Were lessons learned from the 2002 catastrophe?International Journal of Public Opinion Research20: 275–298.
MartinE. A., TraugottM. W., and KennedyC.2005. A review and proposal for a new measure of poll accuracy. Public Opinion Quarterly69: 342–369.
6.
MostellerF., HymanH., McCarthyP. J., MarksE. S., and TrumanD. B.1949. The Pre-Election Polls of 1948: Report to the Committee on Analysis of Pre-Election Polls and Forecasts.New York: Social Science Research Council.
7.
TellinghuisenJ.2001. Statistical error propagation. Journal of Physical Chemistry105: 3917–3921.
8.
WeisbergH. F.2005. The Total Survey Error Approach: A Guide to the New Science of Survey Research.Chicago: University of Chicago Press.
