Adjustment for Noncoverage of Non-landline Telephone Households in an RDD Survey

The National Immunization Survey (NIS) — a nationwide, list-assisted RDD survey conducted by the NORC for the Centers for Disease Control and Prevention — monitors vaccination rates of children between the ages of 19 and 35 months. As in any RDD survey, households without landline telephone (i.e., phoneless and cell-only households) are not sampled in the NIS. To compensate for this noncoverage, a special poststratification adjustment (Keeter 1995; Frankel et al. 2003) is applied, in which households with an interruption in landline service are used to represent nonlandline households. With the increasing proportion of cell-only households among the non-landline households, this adjustment may no longer be very effective unless the characteristics of cell-only households are similar to that of households with interruption in landline service. Also, the adjustment introduces a considerable variation in sampling weights. The paper evaluates the impact of this adjustment on bias and variance of the estimates. The characteristics of different types of households in terms of telephone availability are compared to assess the effectiveness of this adjustment and to explore the possibility of improving this adjustment by treating cell-only and phoneless households separately.