Evaluation of ARRA Comparative Effectiveness Research Dissemination Contract Efforts

The consumer survey had a total of 948 respondents. The methodology for generating the survey sample is described in the report body. We followed the weighting procedure described below to make the survey results representative of the US population above the age of eighteen who visited a doctor or other health care professional in the past 12 months.

In generating the weights, the IMPAQ team has assumed that there is no survey non-response bias. Since no demographic information on potential respondents, other than their names, addresses and phone numbers, are available in the SSI database of 11,500 that we purchased for sampling, we are unable to compare the demographic characteristics of the sampling universe and the final survey sample. Therefore, we assume that the survey respondents and non-respondents are similar in terms of their demographic characteristics and potential responses to survey questions. We acknowledge that this is an important assumption which we are unable to test due to lack of necessary data.

As mentioned in the report body, the SSI sample is that of individuals listed in landline telephone directories. If the population of landline-owning individuals is different from the "cell phone only" population—a group that is absent from our survey sample—that is another source of bias too. We assume that this non-coverage bias may not be too large in the case of the research questions addressed by the consumer survey.²⁵

Table H-1: Consumer Survey Weights

Table H-1 provides the final survey weights for various age groups (column 6) and the data used for generating the weights. The IMPAQ team generated the survey weights by dividing the estimated US population of various age groups who visited a doctor or other health care professional in the past 12 months (column 4) by the number of survey respondents from the corresponding age group (column 5). The US population of various age groups that visited a doctor or other health care professional in the past 12 months is estimated by multiplying the US population of various age groups (column 2) with the percentage of population belonging to each age group who visited a doctor or other health care professional in the past 12 months (column 3). The US population belonging to various age groups used in generating the weights is based on the 2010 US Census.²⁶ The percentages of people who visited a doctor or other health care professional in the past 12 months were generated from the National Health Interview Survey 2011 estimates.²⁷

²⁵ Based on a comparison of estimates for total and landline owning adults using the 2007 National Health Interview Survey data, Blumberg and Luke (2009) state that telephone surveys limited to landline households may still be appropriate for health surveys of adults except in the case of some behavioral risk factors and health care service use indicators. American Journal of Public Health. 2009; Vol. 99, No. 10: 1806–1810.

²⁶ Table 1. Howden ML and Meyer AJ. Age and Sex Composition: 2010, 2010 Census Briefs. May 2011. Available at http://www.census.gov/prod/cen2010/briefs/c2010br-03.pdf (2 MB). Accessed on 02/12/2013.

²⁷ Table 35. Schiller JS, Lucas JW, Ward BW, Peregoy JA. Summary Health Statistics for U.S. adults: National Health Interview Survey, 2010. National Center for Health Statistics. Vital Health Stat 10 (252). 2012.

Return to Contents

Proceed to Next Section