Assessing Data Quality of Cell Phone Random Digit Dial Surveys

Objective: To compare the quality of data collected in health surveys conducted via cell phone with the data quality of landline surveys.

Methods: Data from the New Jersey Adult Tobacco Survey (NJATS) conducted in April through June, 2008 were used to compare the quality of cell phone and landline survey data. We measured the quality of the survey data collected in three ways:

• Selected respondent attrition

• Item non-response

• Richness of responses to open ended questions

Results: Attrition from cell phone surveys was reliably greater than attrition from landline surveys. Controlling for demographic differences in the two samples, there was a small difference in the quality of the observed data as measured by item non-response, but there was no difference in the quality of responses given to open ends.

Conclusions: Conducting cell phone interviews is increasingly important to reaching critical population segments including young people and minorities, but, because of the mode of interview, the quality of data collected via cell phone may be lower. Strategies for mitigating this consequence of cell phone interviewing are discussed.