Survey Sampling Inference under Mode Effects and Proxy Surveys, with Non-ignorable Nonresponse* 

Danny Pfeffermann
Hebrew University of Jerusalem, Israel, and University of
of Southampton, UK .

In this presentation, I propose a new methodology to address simultaneously three major problems in survey sampling: The first is known as mode effects, under which responses of sampled units are related or depend on the mode of response, whether by internet, telephone, personal interview, etc. The second is of proxy surveys, whereby sampled units respond not only about themselves but also about other sampled units. For example, in many household surveys, one member of the household provides information for all other members, possibly with measurement errors. The third problem is not missing at random (NMAR) nonresponse, under which the nonresponse depends on the target outcome variable, even after accounting for known covariates.

Ignoring even one of these three problems may result in bias of point estimators and subsequent inference.  As far as I can tell, none of them is considered in standard (basic) academic sample survey courses and I personally only became aware of their effect during my long service at the Central Bureau of Statistics in Israel. I shall illustrate the proposed methodology by use of simulation experiments and real sample data, with known true population values.

*Talk based on collaborative research with Arie Preminger, Central Bureau of Statistics, Jerusalem, Israel

PLEASE JOIN US FOR A RECEPTION AFTERWARDS.

Venue:  AJC  (A. James Clark Hall) Forum 1101 

Please RSVP below if you would like to attend remotely or in-person.

REGISTRATION IS REQUIRED (FREE EVENT)

WATCH VIDEO