ESRA logo
Tuesday 14th July      Wednesday 15th July      Thursday 16th July      Friday 17th July     

Thursday 16th July, 11:00 - 12:30 Room: HT-102

Unit and item nonresponse 1

Convenor Mr Peter Linde (Statistics Denmark )

Session Details

Greater response for less money
Quality improvements frequently result in increased costs, e.g. in order to reduce nonresponse. However, nonresponse does not only depend on the number of attempts at making contact and mode, but also on the actual and experienced burden. Consequently, increased focus on reducing the experienced burden can be instrumental in increasing the achievement. The theme is discussed by examples from Statistics Denmark, where new digital solutions, better letters, prizes, reminders as to agreements, follow-up and interview training resulted in a higher degree of achievement as well as lower total costs. Concrete 6% less nonresponse and 10% less interview cost.

Paper Details

1. Unit non-response in household wealth surveys: experience from the Eurosystem’s Household Finance and Consumption Survey
Mr Guillaume Osier (European Central Bank)

This contribution presents descriptive statistics, mainly response rates and response representativity indicators (R-indicators) in relation to unit non-response in the Household Finance and Consumption Survey (HFCS), which is a recent initiative from a network of the ECB, National Central Banks and National Statistical Institutes to collect comparable micro-data on household wealth and indebtedness among the euro area countries. This contribution also compares several reweighting strategies for coping with unit non-response, in particular simple and generalised calibration methods. The methods are assessed with respect to their impact on the main HFCS-based estimates.

2. Unit-Nonresponse in the IAB-Establishment Panel: The influence of the interaction between interviewer and respondent
Dr Susanne Kohaut (Institute for Labour Market Research)
Mr Peter Ellguth (Institute for Labour Market Research)

The IAB-Establishment Panel is a longitudinal survey in which the same establishments are contacted every year. In this paper reasons why some establishments do not respond to the questionnaire any longer are analyzed. To model the nonresponse process the conceptual framework for the decision process in establishments by Willimack/ Snijkers (2013) is used. We are especially interested in the interaction between interviewer and respondent. Earlier research (Janik/Kohaut 2012) showed that a change of the interviewer decreased the probability of further participation significantly. Meanwhile we are able to include information on the respondent as well.

3. Individual, Family, Interviewer and Survey Effects on Nonresponse and Bio-specimen Quality in Five Representative US Studies
Dr Colter Mitchell (University of Michigan)
Dr Jessica Faul (University of Michigan)

As part of a broader push to elucidate biological mechanisms linking social life and health, many population-based studies are incorporating biological specimen collections into the survey process. Relatively unknown is the degree to which those who provide biological data are different from those who do not, or if there are differences in subsequent data quality by important groups. In this paper we utilize five large representative studies to examine predictors of biological specimen nonresponse and specimen quality. We expand on previous literature examining individual factors and investigate family, interviewer, and study design predictors of nonresponse and specimen quality.

4. MI Double Feature: Multiple Imputation to Address Nonresponse and Rounding Errors in Income Questions
Dr Jörg Drechsler (Institute for Employment Research)
Professor Hans Kiesl (Ostbayerische Technische Hochschule Regensburg)
Mr Matthias Speidel (Institute for Employment Research)

Obtaining reliable income information in surveys is difficult for two reasons. On the one hand, many survey respondents consider income to be sensitive information and thus are reluctant to answer questions regarding their income. On the other hand, respondents tend to round their income. Especially this second source of error is usually ignored when analyzing the income information.

We present how inferences based on the collected information can be biased if the rounding is ignored and how multiple imputation can be used to account for the rounding in reported income as well as to address the nonresponse problem.