Managing Respondent Burden for a Panel Using Permanent Random Number Sampling
N. Ganesh
NORC at the University of Chicago
Vicki Pineau
NORC at the University of Chicago
J. Michael Dennis
NORC at the University of Chicago, Palo Alto, CA
Multi-client survey platforms such as probability or non-probability sample panels may inadvertently place a high respondent burden on a select number of panelists. When selecting samples from a panel for client studies, some panelists might be selected much more frequently; furthermore, some panelists may rarely get selected for client studies. The use of permanent random number sampling reduces the problem of unequal respondent burden by "equally distributing" client survey assignments across the entire panel. In this paper, we discuss how we implemented permanent random number sampling for selecting panelists for client studies conducted using NORC's AmeriSpeak® Panel. We also present results from a simulation study under the permanent random number sampling approach that compares two methods of within household sample selection; we show via simulation that for households with two or more recruited panelists that a probability proportional to size approach performs better in terms of managing respondent burden when compared to random selection of a panelist within a household.