Recent polls of public approval of President George W. Bush have been at historically high levels. Typically these polls rely upon a dichotomous rating scale to measure approval. We sought to compare polychotomous approval scales with the typical dichotomous approval scale in their relative efficacy in tracking change in evaluations across time.
Method:
7 waves of online data collection began in September 2001. Each wave had approximately 2000 respondents who were randomly drawn from the Harris Poll Online panel. We used 4 response formats to measure approval (2,3,5,7 categories). In addition, job evaluation was measured across waves with a 4 category scale (Poor, Only fair, Pretty good, Excellent). This evaluation question was also asked in a parallel phone survey for 6 of the waves.
Results:
The 3, 5, and 7 category scales all picked up a significant decline in support, well before the 2 category approval rating began to drop. This decline in support was found to be replicated in the job evaluation measure administered both online and phone. We also found that the correlations between the approval scales with the job evaluation scale were significantly higher for the 5 and 7 category scales than for the 2 or 3 category scales.
Discussion:
It appears that a 5 or 7 category scale may be more sensitive to changes in strength of support, especially when compared to the dichotomous scale. In addition, the 5 and 7 category scales appear to have greater convergent validity based on the correspondence with a similar measure. We will also report on the correspondence of these various scales with party affiliation and political ideology.