Weighting and missing values

In order to adequately implement a survey, it is necessary to consider how a suitable estimate for the target population can be made on the basis of the survey data. This also includes the calculation of design and adjustment weights.

In our consulting activities in the area of Survey Statistics , we focus on the problems associated with the estimation of population values. In addition to consulting on the weighting of survey data, we share our expertise on the imputation of missing values, non-response bias analysis, and variance estimation. We are happy to advise and support you on these topics with our experience.

Request consultation

Our consulting services for weighting and missing values

Weighting

A "weight" is a multiplicative factor that can be used for any analysis of a survey data set with the aim of obtaining a less biased estimator for the population. A basic distinction is made between design and calibration weights. While design weights are indispensable for surveys with different selection probabilities, calibration can reduce potential nonresponse bias and the variance of the estimator.

Imputation of missing values

Imputation describes the process of replacing missing values with existing or estimated values using auxiliary variables to obtain a complete data set. Imputation is usually done at the item level and allows working with a complete data set. GESIS has many years of experience in the use of imputation procedures and advises on the appropriate imputation strategy for the respective survey data set.

Nonresponse bias analysis

Nonresponse bias can arise when participants in a survey differ systematically from non-participants. If the reasons for non-participation are associated with target variables, it is not possible to reliably infer the population parameter from the responses. Nonresponse bias analysis allows us to assess these biases. We give recommendations for dealing with nonresponse bias and point out possible limitations of surveys, when these suffer from nonresponse bias.

Who is this service for?
We provide consultation to researchers within the context of research and infrastructure projects, whose outputs benefit the social sciences (e.g., through publications, publicly available data, or validated measurement instruments). Unfortunately, we are unable to offer advice on bachelor's or master's theses.

What are the costs?
Short consultations, offered as a form of guidance for self-help, are free of charge. We are also happy to support you with more in-depth, fee-based consulting services. Further details can be found in our pricing overview.

Consultation fees

Cognitive pretesting
Empirically assess the comprehensibility and validity of questions in-depth

Sampling
We draw telephone samples and complex samples for you

Weighting and analysis of complex data
We perform weighting, imputation, or nonresponse bias analyses for you

GESIS survey guides
Practical guidelines on survey methodology topics

SQP - Survey quality predictor
Estimate the measurement quality of questions using meta-analysis

Meet the Experts - Season 1: Survey Methodology
Recorded lectures from our "Meet the Experts" series

QuestionLink
Harmonization of questions for selected constructs with recoding scripts and an R package for your own projects

Questionnaire design

Survey Design

Sample Design