Inconsistencies
A review of other demographics was conducted and there are some inconsistencies between the panel data and known data representing the Pensacola MSA and/or the counties. For example, on income and education, the lower income and education categories are underrepresented for the panel members. (For example, about 7% of the panel participants reported incomes of “under $15,000” while the Survey of Buying Power reports about 18% of the MSA with EBI (Effective Buying Income) in the “under $15,000 category . However, this is not unusual . While the sample plan ensured that all education and income category members had a chance of being included in the final sample, these populations traditionally are underrepresented in survey data because they have higher rates of declining to participate.

Consistent with Other Data Collected by Representative Surveys?
If a study’s results differ from another study that is representative it may be due to the present study not being representative, the original study not being representative, or it may be due to substantive changes in the population that occurred during the time lapse between the two studies. If we have reason to believe that the previous study is, in fact, representative, we would have greater confidence in the representativeness of the present study if data are consistent between the two studies on at least some of the questions surveyed. We could expect changes in some responses to have occurred due to changes in the population during the elapsed time between studies.

In 1993, the UWF Department of Marketing and Economics conducted a study, using a probability sampling plan required to generate a representative sample, of heads of household in Escambia County. Data showed no differences between certain demographic characteristics of the sample and those known to exist for the population through Census data (see above). Therefore, theoretically, the 1993 sample was representative and this was confirmed through analysis. Since several questions were asked of the present 2003 panel members which were identical to those asked in the 1993 study, we may use answers to those questions to make a second assessment of representativeness. First, the data (reported in Panel 0001) are strikingly similar on many questions. For example, there are questions asking respondents to rate several government entities on a scale from 1 to 10 . With few exceptions, the mean scores are very close to the mean scores in 1993. Furthermore, the same patterns exist. For example, the sheriff and police were rated the highest in 1993 and they were so again in 2003. In one case, the County Government in Escambia County , mean ratings were lower. We would expect this to occur since several Escambia County Commissioners were removed from office for unlawful acts. Therefore, the data appear to be internally consistent with findings we would expect in a representative study.

«« return