Calibration Estimators in Survey Sampling

Heterogeneous response to HAART across a diverse population of people living with HIV: results from the ANRS-EN12-VESPA Study

“…Major socio-demographic and health-related characteristics were collected both among respondents and non-respondents, allowing for a weighting procedure to take into account this participation bias. This weighting procedure included two steps: first, considering the unequal probability of enrolment related to the heterogeneous frequency of hospital visits, a weight was attributed to each individual corresponding to the inverse number of hospital visits he(she) had reported for the preceding year; second, to account for non-response, an additional weight was computed, using a method of calibration adjustment [17], in such a way that the weighted distribution of the participants regarding transmission group, employment status and immunological status was comparable to that of the entire eligible population.…”

Post-Stratification and Calibration-A Synthesis

“…Jagers (1986). A few exceptions apart (Fuller, 1966), calibration estimation (Deville and Särndal, 1992;Deville, Särndal, and Sautory, 1993) provides an alternative general methodology.…”

Estimation when the Covariance Structure of the Variable of Interest is Positive Definite

“…When the variance matrix S is diagonal, it is known that the calibration estimator introduced by Deville and Särndal (1992), or the equivalent generalized regression estimator, asymptotically attains the Godambe and Joshi lower bound (see, for example Särndal et al 1992). In the next section, the calibration estimator will be generalized to the case of a positive definite variance matrix S. It will then be shown that this generalized calibration estimator asymptotically attains the lower bound given in (11).…”