2011 JSM Presentations
CALIBRATING NON-PROBABILITY INTERNET SAMPLES WITH PROBABILITY SAMPLES USING EARLY ADOPTER CHARACTERISTICS
Charles A. DiSogra, Knowledge Networks; Email
Curtiss Cobb, Knowledge Networks; Email
Elisa Chan, Knowledge Networks; Email
J. Michael Dennis; Email
A representative study sample drawn from a probability-based Web panel, after post-stratification weighting, will reliably generalize to the population of interest. Due to finite panel size, however, there are instances of too few panel members to meet sample size requirements. In such situations, a supplemental sample from a non-probability opt-in Internet panel may be added. When both samples are profiled with questions on early adopter (EA) behavior, opt-in samples tend to proportionally have more EA characteristics compared to probability samples. Taking advantage of these EA differences, this paper describes a statistical technique for calibrating opt-in cases blended with probability-based cases. Using data from attitudinal variables in a probability-based sample (n=611) and an opt-in sample (n=750), a reduction in the average mean squared error from 3.8 to 1.8 can be achieved with calibration. The average estimated bias is also reduced from 2.056 to 0.064. This approach is a viable methodology for combining probability and non-probability Internet panel samples. It is also a relatively efficient procedure that serves projects with rapid data turnaround requirements.
