Mark Nord1, Heather Hopwood. 1. US Department of Agriculture, Economic Research Service, 1800 M Street NW, Washington, DC 20036-5831, USA.
Abstract
OBJECTIVE: To assess whether interview mode (telephone vs. in-person) affects the results of surveys that measure food security. DESIGN: Responses given by households interviewed by telephone and in-person in recent US Current Population Survey Food Security Supplements (CPS-FSS) were compared. Statistical methods based on the Rasch measurement model were used to assess whether response patterns differed between the two interview modes. Multivariate logistic regression analysis was then used to gauge the effect of interview mode on the measured household prevalence rates of food insecurity and very low food security while controlling for income, employment, household structure, and other household characteristics that affect food security. RESULTS: Response patterns to the indicators that comprise the food security scale did not differ substantially between interview modes. Prevalence rates of food insecurity and very low food security estimated from the two interview modes differed by only small proportions after accounting for differences in the socio-economic characteristics of households. CONCLUSIONS: Findings suggest that effects of interview mode on food security measurement in the CPS-FSS are small, or at most modest. Prevalence estimates may be biased upwards somewhat for households interviewed in-person compared with those interviewed by telephone. The extent to which these results can be generalised may depend, to some extent, on survey characteristics other than interview mode, such as surveyor name recognition and respondents' trust and confidence in the surveyor.
OBJECTIVE: To assess whether interview mode (telephone vs. in-person) affects the results of surveys that measure food security. DESIGN: Responses given by households interviewed by telephone and in-person in recent US Current Population Survey Food Security Supplements (CPS-FSS) were compared. Statistical methods based on the Rasch measurement model were used to assess whether response patterns differed between the two interview modes. Multivariate logistic regression analysis was then used to gauge the effect of interview mode on the measured household prevalence rates of food insecurity and very low food security while controlling for income, employment, household structure, and other household characteristics that affect food security. RESULTS: Response patterns to the indicators that comprise the food security scale did not differ substantially between interview modes. Prevalence rates of food insecurity and very low food security estimated from the two interview modes differed by only small proportions after accounting for differences in the socio-economic characteristics of households. CONCLUSIONS: Findings suggest that effects of interview mode on food security measurement in the CPS-FSS are small, or at most modest. Prevalence estimates may be biased upwards somewhat for households interviewed in-person compared with those interviewed by telephone. The extent to which these results can be generalised may depend, to some extent, on survey characteristics other than interview mode, such as surveyor name recognition and respondents' trust and confidence in the surveyor.