1,721,090 research outputs found
ADRC-E: making a methodological impact
No full textAs part of its remit, the Administrative Data Research Centre for England is undertaking a programme of methodological research. In this paper I will describe a few examples from this programme including work on: guidance about the information that needs to be made available about the data linkage process by data providers, data linkers, analysts and researchers; a new approach to linkage based upon weights derived using a scaling algorithm; and the representativeness of surveys using a unique data set linking call record paradata from three UK social surveys to census auxiliary attribute information on sample households
Knock-to-nudge methods to improve survey participation in the UK
No full textRelevance & research question: data collection organisations are shifting toward new approaches, with social surveys undergoing significant design and implementation changes. Since the COVID-19 pandemic, agencies have increasingly moved to online data collection due to dwindling response rates and rising fieldwork costs. A key challenge for self-completion general population surveys is the absence of field interviewers to facilitate recruitment and participant retention. This research examines the UK survey landscape, aiming to identify recruitment methods for self-administered surveys, that can produce more representative samples of the general population.Methods & data: we present findings from an information request sent to the UK’s nine most important survey agencies. We collected information on surveys without field interviewers conducted between 2018 and early 2024, including publicly available technical and methodological reports and other survey materials, along with internal reports provided by the agencies. We processed and codified this information, building a spreadsheet containing 144 instances of 59 longitudinal and cross-sectional surveys, along with 227 communication materials. Results: the responses for the surveys in our dataset use 57% online, 38% paper, and 5% telephone modes. Most surveys (84%) offer incentives to participants, with 92% being monetary and only 33% given unconditionally. Response rates vary widely – household-based cross-sectional surveys tend to have lower response rates (81% at 30% response or lower) than individual-based ones (47% at 30% or lower). Longitudinal surveys generally have the highest response rates. While only 35% of reports assess sample representativeness, the general trend confirms that mixed-mode surveys yield more representative samples than single-mode surveys.Added value: to our knowledge, this review is the first coordinated effort to collate and summarise recruitment strategies for surveys without field interviewers in the UK. It covers sampling design, communication strategies and materials, incentivisation, fieldwork procedures, response rates, and report quality assessments. Our dataset provides insights into the current state of survey practice and helps identifying practices that might contribute towards higher response rates and sample representativeness
When to stop calling? Using call record data to assess nonresponse bias in a longitudinal study
No full textModel-based inference for categorical survey data subject to non-ignorable non-response (with discussion)
No full textWe consider non-response models for a single categorical response with categorical covariates whose values are always observed. We present Bayesian methods for ignorable models and a particular non-ignorable model, and we argue that standard methods of model comparison are inappropriate for comparing ignorable and non-ignorable models. Uncertainty about ignorability of non-response is incorporated by introducing parameters describing the extent of non-ignorability into a pattern mixture specification and integrating over the prior uncertainty associated with these parameters. Our approach is illustrated using polling data from the 1992 British general election panel survey. We suggest sample size adjustments for surveys when non-ignorable non-response is expected
Assessment of multiple membership multilevel models: an application to interviewer effects on nonresponse
No full text
Using call record data to investigate household response outcome:: Evidence from understanding society data
No full textSurvey practitioners are increasingly interested in how best to use paradata including call record data to investigate nonresponse and to improve data collection processes. One particular question is whether it is possible to identify early on during the fieldwork sample cases that may require longer time until interviewing is completed and therefore require a lot of financial and staff resources. Another question is how useful paradata, including call record data, from a cross-sectional study and from the current and previous waves of a longitudinal study are in predicting response outcomes in both contexts.This paper uses data including call record data from the Understanding Society Waves 1-3 to investigate and predict household response outcomes. We first analyse call sequences within households using sequence analysis. The main aim of using sequence analysis is to understand better the survey processes. We then employ call record data in cross-sectional and longitudinal contexts to predict final call outcomes of households. The results indicate that outcomes of previous calls, in particular from the most recent call, are highly predictive in both contexts. The timing of calls in cross-sectional context as well as previous wave paradata in longitudinal context, although significant in the models, only slightly improve the predictive power of the models. <br/
- …
