Targeted response inducement strategies for self-completion surveys with address-based sample frames

---

Abstract

Targeted response inducement strategies have become increasingly popular in recent years. A number of studies have shown that such strategies can be a cost-effective way to bring about a useful improvement in response rate and/or sample balance. However, the design and implementation of targeted strategies depends on prior information about sample members, for example from an informative sampling frame, to identify groups to target. Furthermore, many of the successful implementations of targeted methods have been on interviewer-administered surveys, where the interviewers can be the conduit of the targeted procedures. The design of targeted procedures is therefore particularly challenging in the case of self-completion surveys with address-based sample frames, as there is neither an informative sampling frame nor are there interviewers who can behave in targeted ways. In this article, we attempt to identify and describe promising approaches to targeted design in this situation. We also identify a research agenda which could lead to better-informed survey design and hence more effective survey data collection by self-completion in future.

Keywords

data linkage, incentives, Non-response, prenotification, Representativeness, web survey

Acknowledgement

This research was supported by UKRI-ESRC strategic research grant ES/X014150/1 for “Survey data collection methods collaboration: securing the future of social surveys”, known as Survey Futures. Survey Futures is directed by Professor Peter Lynn, University of Essex, and is a collaboration of twelve organisations, benefitting from additional support from the Office for National Statistics and the ESRC National Centre for Research Methods. Further information can be found at www.surveyfutures.net.

1. Introduction

Survey research commonly relies on standardisation, where all sample members are subject to the same set of recruitment and data collection procedures. In recent years, targeted designs have gained popularity as an alternative to standardised designs (Lynn, 2017a). Published literature provides evidence of the effectiveness of targeted response inducement strategies and a number of surveys now utilise targeted features in their designs (Sladka & Lynn, 2025).

A targeted survey design is characterized by the variation of one or more design features between subgroups of sample members, where the variation(s) are identified and planned prior to the start of data collection (Lynn, 2017). Targeted designs are a particular type of adaptive designs (Schouten et al, 2018). Adaptive designs also include both responsive designs and tailored designs. Responsive designs involve monitoring indicators throughout the fieldwork period and making adaptations after data collection has already begun in response to preliminary outcomes (Groves & Heeringa, 2006), whereas targeted designs involve pre-planned interventions, which do not necessitate making any changes during the course of fieldwork. Tailored designs treat individual sample members differently based on their individual behaviour and characteristics (Couper & Wagner, 2011), whereas targeted designs apply interventions to broad subgroups.

The pre-planned nature of targeted designs means that information about sample members needs to be available before data collection begins in order to identify the subgroups to be targeted and the appropriate variation(s) in treatment. In some cases, the required information can be obtained from the sampling frame, however, in many countries only address-based samples – which typically lack any useful information about the addresses – are possible for general population surveys. For example, 14 of the 31 countries participating in round 11 of the European Social Survey relied on an address-based frame (European Social Survey, 2023). This makes the implementation of targeted procedures challenging for many surveys, especially cross-sectional ones.

The changing role of interviewers in the post-pandemic world where a number of surveys are shifting to self-completion (Charman et al, 2025; Gummer et al., 2025) poses another challenge, as much of the existing research on targeted response inducement strategies was carried out on interviewer-administered surveys and relied on interviewers to implement the strategies (Sladka & Lynn, 2025). Consequently, some of the targeted strategies that were found to be effective in earlier studies (Luiten & Schouten, 2013; Gummer & Blumenstiel, 2018) cannot be applied to self-completion surveys.

This article will attempt to tackle the challenge of designing targeted procedures for self-completion surveys with address-based sample frames. It will do this by identifying the kinds of targeted response inducement strategies that can be used on self-completion surveys (Section 2), discussing how subgroups suitable for targeting can be identified in address-based samples (Section 3) and reviewing existing evidence as to which targeted procedures are effective (Section 4). Following this, we will discuss the implications and will identify what else we need to learn in order to better inform future targeting strategies for self-completion surveys with address-based sample frames.

2. Targeted procedures for self-completion surveys

A targeted design feature can be categorised across 3 dimensions: the agent of change, the mechanism through which the change is achieved and the outcome that will be affected (Lynn, 2017a). Many of the existing studies of targeted design features rely on the interviewer as the agent of change, making them mostly irrelevant to the self-completion context. The one exception of an interviewer-administered feature that could be relevant to self-completion surveys is the knock-to-nudge approach, where respondents receive an in-person visit in order to nudge them to complete the survey (Domarchi, Maslovskaya and Smith, 2025). Knock-to-nudge visits can be introduced, or their number can be varied as a way of targeting particular subgroups. This is done on the UK Labour Force Survey (Fraser, 2024).

Targeted features aimed directly at the respondent, on the other hand, all have relevancy to self-completion and studies of these features conducted in a face-to-face context can be drawn upon when designing targeted strategies for self-completion. The content or design of respondent communications such as invitation letters, reminders, between-wave mailings or landing pages of web surveys can be modified to appeal to targeted subgroups. The mode, number and timing of respondent communications such as reminders can be varied between different subgroups. Additional mailings such as between wave mailings or a short report of the findings can be introduced for select subgroups. Respondent incentives can be introduced or the nature, value or form of incentives can be modified as a way of targeting particular subgroups. Finally, an alternative data collection mode, such as a paper questionnaire as an alternative to a web questionnaire, can be offered to certain groups of respondents. Detailed descriptions and examples of use of these features can be found in Section 4. Table 1 summarises design features that could be applied in a targeted way in self-completion surveys.

Of the features listed here, the following have all been shown to be effective by at least one experimental study: targeted invitation letters (Lynn, 2016; Atchison et al., 2025), additional mailings (Fumagalli et al., 2013; Cleary & Balmer, 2015), knock-to-nudge visits (Fraser, 2024), mode assignment (Luiten & Schouten, 2013), incentive value (Jackson et al., 2020; Beste et al., 2023), extra incentives (Zukerberg et al., 2007; Atchison, 2025) and incentive timing (McGonagle et al., 2023)

Furthermore, some other targeted interventions have been used or are currently being used on surveys in the UK and are believed to be effective, such as contact information request (Lynn, 2012), varying number of reminders (DCMS, 2022; DCMS, 2023; HMRC, 2023; Kantar Public, 2023), and provision of alternative mode (DESNZ, 2023; DCMS, 2022; DCMS, 2023; HMRC, 2023; Kantar Public, 2023).

 

Table 1: Design features that could be applied in a targeted way in self-completion surveys

Design feature	Mechanism & outcome	Intervention
Communication content/design	Should motivate the sample member to take part in the survey	Invitation letter
		Reminder letter
	Should motivate sample members to begin the questionnaire	Landing page
		Questionnaire design
	If promised (e.g. feedback), should motivate the sample member to take part; if between waves of a longitudinal survey, should motivate participation in the next wave	Additional mailings
	Should motivate the respondent to complete the questionnaire	Questionnaire design and length
Communication protocol	Should improve chances that sample members are located at next wave of a longitudinal survey	Contact information request
	Should improve the chances that sample members are contacted	Timing/number of reminders
		Use of registered/special mail
		Knock-to-nudge visit
		Contact mode
		Reminder mode
Data collection mode	Should make accessing the questionnaire easier for the sample member	Mode assignment
		Provision of alternative mode
Incentives	Should motivate the sample member to take part in the survey	Incentive value or form
		Extra incentive
		Incentive timing

The remaining interventions listed are ones that – to our knowledge – have neither been tested experimentally nor used in practice but have potential to be applied in a targeted way. It is also possible that other features not listed here could be used for targeting.

3. Targeted procedures when using address-based sample frames

A persisting challenge in address-based samples is how to obtain the information necessary to identify the subgroups to be targeted. Typically, address lists used as sampling frames contain no information about individual addresses other than geographical location. A solution is to link data from external sources to the sampling frame using geographical identifiers – an approach already utilised by some survey agencies (DESNZ, 2023; DCMS, 2023; DCMS, 2022; HMRC, 2023; Kantar Public, 2023), as well as published studies (Jackson et al., 2020; Fraser, 2024). The availability of such data differs by country and the sampling frame used.

In the UK there are many geographically referenced datasets available for linkage to the postcode address file (PAF), the address list commonly used for social survey sampling (https://www.poweredbypaf.com/postcode-address-file/). Linkage can be done either via unit level geographical identifiers, such as postcodes or grid co-ordinates, to any other data sources that use the same geographical identifiers, or via standardised areal units that can be delineated in terms of those identifiers, such as output areas, lower layer super output areas (LSOAs) or electoral wards.

Many small area statistics that are available for some of those standardised areal units come from official sources such as government departments or the Office for National Statistics – including data from the Census of Population (https://www.ons.gov.uk/census), Index of Multiple Deprivation (https://data.geods.ac.uk/dataset/index-of-multiple-deprivation-imd), Labour Market Statistics (https://www.nomisweb.co.uk/home/profiles.asp), Crime statistics or Rural-Urban classification (https://www.ons.gov.uk/methodology/geography/geographicalproducts/ruralurbanclassifications). While some of these are public access, others require a special license or other form of access agreement. Apart from official sources, geodemographic data can also be obtained from commercial sources.

As an example of this approach, several studies conducted by Verian use the Index of Multiple Deprivation combined with predictive household structure data from CACI Ltd to inform their targeted designs (DCMS, 2023; DCMS, 2022; HMRC, 2023; Kantar Public, 2023). With the help of this data, they successfully identified small geographical areas containing high proportions of households in relative deprivation and/or high proportions of older people who would benefit from targeted provision of an alternative data collection mode. By including a paper version of the questionnaire in one of the reminder mailings, response rate was increased in these target groups in a cost-effective manner. Similar approaches are likely to be possible for other surveys and in other countries.

A screening stage prior to the start of the main stage of a survey could also be used to obtain information about sample members. At first glance, this might seem at odds with the pre-planned nature of targeting, however, a screening stage is conducted independently of the survey itself, allowing for subgroups to be identified and variations in treatment to be implemented before actual data collection begins. This method has – to our knowledge – never been tested as the sole source of information for a targeted design, however, one experimental study did utilise data from a screener stage as an additional source informing their targeted design (Zheng et al., 2024).

Finally, it is worth noting that longitudinal studies are a special case among address-based samples, where obtaining information about sample members is only an issue at the first wave. In subsequent waves, data collected at previous wave(s) is commonly used to inform targeting. The issues discussed in this section apply primarily to cross-sectional surveys and to the first wave of longitudinal surveys.

4. Which targeted procedures are effective for self-completion surveys with address-based sample frames?

Table 2 summarises studies of effectiveness of targeted features that are relevant to the self-completion context with an address-based sampling frame. The studies were identified as part of a larger evidence review on targeted procedures (Sladka & Lynn, 2025). This review consisted of an online search for publications on targeted survey procedures in English, which was conducted between autumn 2023 and spring 2024. Google Scholar was used to search for papers published in academic journals, using various combinations of keywords such as “targeted survey procedures”, “targeted survey designs” and “non response”. Seeing as the review was expanding on the work of Lynn (2017a), which had already reviewed earlier publications, the additional search was limited to work published since 2017. With the aim to review all currently existing experimental evidence, the search was not limited by the geographic location of the survey, nor to any particular type of survey – both longitudinal and cross-sectional surveys were included, regardless of the data collection mode used, but was restricted to English language publications. The evidence drawn upon in this section therefore covers all publications up to early 2024.

 

Table 2: Summary of studies of targeted features relevant to self-completion surveys

Intervention	No. of studies	Studies showing effectiveness	Studies showing no effects
Invitation letter	4	Lynn (2016); Atchison et al. (2025)	Einarsson et al. (2024); Zhang et al. (2024)
Reminder letter	0	–	–
Landing page	0	–	–
Additional mailings	2	Fumagalli et al. (2013); Cleary & Balmer (2015)	–
Questionnaire design and length	0	–	–
Contact information request	0	–	–
Timing/number of reminders	0	–	–
Use of registered/special mail	0	–	–
Knock-to-nudge visit	1	Fraser (2024)	–
Contact mode	0	–	–
Reminder mode	0	–	–
Mode assignment	1	Luiten & Schouten (2013)	–
Provision of alternative mode	1	Kantar Public (2021)	Zhang et al. (2024)
Incentive value or form	3	Jackson et al. (2020); Beste et al. (2023)	Brown et al. (2023)
Extra incentive	3	Zukerberg et al. (2007); Atchison (2025)	Zhang et al. (2024)
Incentive timing	1	McGonagle et al. (2023)	–

4.1 Invitation letter

Lynn (2016), Einarsson et al. (2024) and Zhang et al. (2024) on mixed-mode surveys, as well as Atchison et al. (2025) on a push-to-web survey, experimented with targeted invitation letters. All four based their targeted subgroups on respondent characteristics and adjusted the wording of the letter to appeal to each subgroup. Lynn (2016) and Atchison et al. (2025) both used data from previous waves of longitudinal surveys to inform their subgroup selection. The former targeted young respondents (aged 16-29), pensionable respondents (aged 60+ for female and 65+ for male), respondents with children below 16, London-dwellers and those labelled as ‘employment busy’ (who work long hours or have long commutes). The latter targeted the youngest (aged 5-10) and the oldest (aged 70+) sample members. Einarsson et al. used data available in the sampling frame (Icelandic National Register) and targeted respondents living in Reykjavik, those living in the Southern district, as well as young respondents (aged 18-35). Zhang et al. used commercial geodemographic data (Esri Tapestry Segmentation: https://www.esri.com/en-us/arcgis/products/arcgis-data/explore/tapestry-data) to identify subgroups of interest and targeted respondents living in areas with a large proportion of Hispanic residents. Lynn (2016) and Atchison et al. (2025) both reported improved response rates among the treated respondents as a result of targeting and hence reduced overall variation in response propensities.

4.2 Additional reminders

The UK Community Life Survey targets a third reminder mailing to sample addresses estimated to contain solely residents aged under 35 years and to addresses in the most deprived quintile of small areas in the UK. This group was identified through analysis of response data from earlier surveys (DCMS, 2023). There is no data to assess the treatment’s effectiveness as it was not implemented experimentally.

4.3 Additional mailings

Fumagalli et al. (2013) and Cleary & Balmer (2015) both experimented with targeted between-wave communications on longitudinal surveys. Using data from previous waves of their respective surveys, the former targeted two different subgroups with low predicted response propensities- namely young people (aged 16-24) and busy people (who work long hours or have long commutes), while the latter targeted respondents who were co-habiting or married, employed or unemployed but economically active, and respondents who reported experiencing at least one of fifteen social issues that were the subject of the survey. Both studies reported improved response rates in the targeted groups, with Cleary & Balmer also seeing an overall increase in response rate.

4.4 Knock-to-nudge visit

Fraser (2024) used Office for National Statistics Population Estimates, Urban/Rural classification and the Index of Multiple Deprivation – all at lower super output area level – to calculate response propensities in order to inform a targeted knock-to-nudge intervention in the context of a self-completion web survey, the Transformed Labour Force Survey. Sample households were sent a pre-notification letter, a letter inviting them to the web survey, and a reminder letter 12 days later.  Sample households living in urban areas, more deprived areas, or where the average age is under 45, were selected to receive knock-to-nudge visits if they had not responded two weeks after the invitation was expected to have arrived. The data shows improved representativeness and an increase in return rates compared to previous cohorts that did not use targeting.

4.5 Mode assignment

At the initial self-completion stage of a sequential mixed mode survey, Luiten & Schouten (2013) assigned sample members with high cooperation propensities to web mode, while those with low propensities were sent a paper questionnaire, with the aim of reducing variation in response propensities. The final response rates of the two propensity groups were closer to each other than in the web-first mixed-mode reference survey, with the high propensity group actually having a marginally significant lower response rate than the low propensity group. Understanding Society, the UK Household Longitudinal Study, used mode-specific response propensity modelling to target a face-to-face initial approach to the small subset of sample households (less than 10%) whose overall response propensity was estimated to be lower if initially invited to complete online (Lynn, 2017b). All other households were initially invited to a self-completion web survey. While there is no final data available, partial data presented while the study was still ongoing suggested that the stated objectives were being met and showed no negative effect of targeting.

4.6 Provision of alternative mode

In an otherwise web-first mixed-mode survey, Zhang et al. (2024) provided respondents living in areas with older populations and lower internet speeds with a paper version of the questionnaire concurrently with the web invitation. The design was not shown to have any effect. The Department for Culture, Media and Sport Participation Survey (DCMS, 2022) and the Community Life Survey (DCMS, 2023), both run by Verian in the UK, include a paper version of the questionnaire along with the second reminder mailing in an otherwise web-only design, but only to a targeted minority of sample addresses, namely those in the most deprived quintile of small areas and addresses where all residents are estimated to be aged 65 or older. This design was informed by an experiment carried out on the 2020-21 Community Life Survey (Kantar Public, 2021), which found that provision of a paper questionnaire to this subgroup should reduce overall variation in response propensities. Earlier analysis (Williams, 2019) had shown that paper questionnaires disproportionately improved response rates for certain subgroups.

4.7 Incentive value

Jackson et al. (2020) on a mail survey and Beste et al. (2023) on a mixed-mode survey both used predicted response propensities to identify subgroups to receive higher-value incentive(s) and report an increase in response rate among the treated low propensity cases. In Jackson et al. (2020), however, this increase does not offset the decrease in response rates in the two highest propensity groups, which was caused by lowering their incentives in order to compensate for the increased incentive value in the low propensity group – leading to an overall decrease in response rate. Brown et al. (2023) also experimented with differential incentives on a mixed-mode survey, offering prior wave nonrespondents higher incentives than prior wave respondents. No significant improvement was seen among the treated non-respondents.

4.8 Extra incentive

Zukerberg et al. (2007) experimented with offering an incentive to sample members in groups with low response propensity, while other sample members received no incentives, and observed an improved response rate (quoted in Lynn 2017a). Zhang et al. (2024) offered an extra incentive to new sample members who had not been successfully contacted within the first 14 days of data collection. This design was not shown to have any effect. Following the results of an incentives experiment in a previous round of the REACT-1 Study, Atchison et al. (2025) offered targeted incentives (conditional on returning a swab test) to the youngest three age groups, and no incentives for other age groups in the final two rounds of a push-to-web survey. This intervention led to an improved swab response rate among the targeted groups and a decreased variation in response rate by age.

 4.9 Incentive timing

On a CATI survey, McGonagle et al. (2023) experimented with offering sample members labelled as high-effort double the standard incentive if they completed their interview within the first month of fieldwork. High effort was defined as requiring the highest number of contact attempts at the previous wave or being eligible for upcoming supplemental studies. The study saw an increase in response rate among the treated cases.

5. Implications and discussion

We have identified design features and specific interventions that can be used in a targeted way on self-completion surveys (Table 1) and we have provided an overview of existing evidence on the use and effectiveness of these procedures (Table 2). Targeted invitation letters, between wave mailings, knock-to-nudge visits, mode assignment, differential incentive value, extra incentives and incentive timing have all been shown to be effective by at least one experimental study. Furthermore, some of the targeted interventions for which there is no experimental evidence of effectiveness, such as targeted provision of alternative mode, contact information request, and varying number of reminders have been, or are currently being, used on surveys in the UK (see Sladka & Lynn, 2025). The remaining interventions listed in section 2 are ones that – to our knowledge – have neither been tested experimentally nor used in practice but could have potential to be applied in a targeted way on self-completion surveys.

Examining existing methodological research into the relevant design features could help to inform their development for targeted use. Where data from methodological experiments allows for comparison of effects between sample subgroups, secondary analysis of this data could simulate the effect of targeting and thereby help to identify appropriate ways to target design features (Asimov & Blohm, 2024; Collins et al., 2026). For example, experiments with randomised allocation of incentive treatment, of which there are many, could be re-analysed with a view to identifying the optimum allocation of incentive treatment to subgroups in order to achieve minimum variation in response propensities for a fixed budget.

We note additionally that any targeting strategy chosen should be aligned with the objectives of the survey, bearing in mind the relative importance of factors such as measurement error, nonresponse error and costs. The aim should be to achieve an optimum balance between errors and costs. With this in mind, secondary analysis of the kind outlined in the previous paragraph can be particularly helpful as it can be tailored to the specific objectives of interest.

Another area of interest for future research is improving the effectiveness of already tested features. While we know that targeted invitation letters can be effective, for example, there is room to explore what types of messaging work best for which sample subgroups. Initial qualitative research designed to identify messaging that resonates well with different subgroups could inform new quantitative experiments. The risk with such experiments (of damaging survey response) are minimal, so we suggest that such experiments could be incorporated into large-scale surveys.

Finally, we have discussed the challenges of obtaining data about sample members in address-based samples in order to be able to define and identify subgroups to target. In longitudinal studies, this issue only exists at the initial wave, as subsequent waves can utilise data collected at previous waves to inform targeting. The solution most commonly seen in survey practice for cross-sectional surveys is to link data from external sources, such as the census or deprivation indices, to the sampling frame though the availability of such data may differ by country.  In the UK, a wide range of such data is available and can provide an effective basis for the development of targeted designs for self-completion surveys even when the sampling frame is an uninformative list of residential addresses. Further research could usefully explore the relative performance of different data sources and different variables for defining groups to target. A screening stage prior to the start of mainstage fieldwork is another possible way of obtaining data to define targeting groups. This may be most suitable for exploration on local surveys, such as those covering a single city, where in-person collection of observation data for a survey sample (of addresses) may be feasible. While targeting on longitudinal surveys using previous wave data and targeting on cross-sectional surveys with informative frames such as population registers have generally proved more effective than targeting on cross-sectional surveys with address-based samples, we conclude that there is considerable potential for effective targeting when only an address-based frame is available. Consequently, evaluation of methods to link or otherwise obtain data for targeting is likely to be worthwhile.

References

1. Asimov, A., and Blohm, M. (2024), “Sequential and Concurrent Mixed-Mode Designs: A Tailored Approach,” Journal of Survey Statistics and Methodology, 12, 558–577, DOI: 10.1093/jssam/smae016.
2. Atchison, C. J., Gilby, N., Pantelidou, G., Clemens, S., Pickering, K., Chadeau-Hyam, M., Ashby, D., Barclay, W. S., Cooke, G. S., Darzi, A., Riley, S., Donnelly, C. A., Ward, H., Elliott, P. (2025). Strategies to Increase Response Rate and Reduce Nonresponse Bias in Population Health Research: Analysis of a Series of Randomized Controlled Experiments during a Large COVID-19 Study. JMIR Public Health Surveillance, 11.
3. Beste, J., Frodermann, C., Trappmann, M., & Unger, S. (2023). Case Prioritization in a Panel Survey Based on Predicting Hard to Survey Households by Machine Learning Algorithms: An Experimental Study. Survey Research Methods, 17(3), 243-268.
4. Brown, M., Adali, T., Fleetwood, S., Burston, K., & Macleod, K. (2023). Efforts to boost response and maximise representativeness in the Next Steps Age 32 Survey.
5. Charman, C., Mesplie-Escott, S., & Collins, D. (2025) The post-pandemic role of face-to-face interviewers: second report, Survey Futures Report no.6. Colchester UK: University of Essex. https://surveyfutures.net/reports/
6. Cleary, A. & Balmer, N. (2015). Fit for purpose? The impact of between-wave engagement strategies on response to a longitudinal survey. International Journal of Market Research, 57(4), 533–554.
7. Collins, J., Batholomäus, S., Gummer, T., Weiss, B. & Kern, C. (2026), “Prediction-based adaptive designs for reducing wave nonresponse rates and bias in panel surveys,” Sociological Methods and Research. DOI: 10.1177/00491241261492217
8. Couper, M.P. & Wagner, J. (2011) Using paradata and responsive design to manage survey nonresponse. Proceedings of the 58th World Statistical Congress, Dublin, 542- 548. http://2011.isiproceedings.org/papers/450080.pdf
9. Department for Digital, Culture, Media and Sport. (2022). DCMS Participation Survey 2021/22: Annual Technical Note October 2021–March 2022.
10. Department for Digital, Culture, Media and Sport. (2023). Community Life Survey: Technical Report, 2021-22. https://www.gov.uk/government/statistics/community-life-survey-202122/community-life-survey-technical-report
11. Department for Energy Security and Net Zero. (2023). DESNZ Public Attitudes Tracker: Technical Overview Summer 2023, UK.
12. Domarchi C, Maslovskaya O & Smith P W F (2025) Knock-to-nudge methods for recruitment in self-administered surveys: Evidence review. Survey Futures Report no.8. Colchester UK: University of Essex. https://surveyfutures.net/wp-content/uploads/2025/11/report-8-knock-to-nudge-evidence-review.pdf
13. Einarsson, H., Cernat, A., & Shlomo, N. (2024). The effects of framing the survey request and using targeted appeals on participation in cross-sectional surveys. Field Methods, 36(3), 187-205.
14. European Social Survey (2023). ESS round 11 – 2023: Country Documentation. https://ess.sikt.no/en/study/412db4fe-c77a-4e98-8ea4-6c19007f551b/
15. Fraser, O. (2024) Lessons from the Transformation of the Labour Force Survey. Presented at the 3^rd Survey Practice Forum, online, October 2024. https://surveyfutures.net/events/2024/01/17/survey-practice-forum-3/
16. Fumagalli, L., Laurie, H., & Lynn, P. (2013). Experiments with methods to reduce attrition in longitudinal surveys. Journal of the Royal Statistical Society: Series A (Statistics in Society), 176(2), 499–519.
17. Gummer, T., & Blumenstiel, J. E. (2018). Experimental evidence on reducing non-response bias through case prioritization: The allocation of interviewers. Field Methods, 30(2), 124 139.
18. Gummer, T., von Glasenapp, K., Skora, T., Bartholomäus, S., & Naumann, E. (2025). The Impact of the COVID-19 Pandemic on the Design of Repeated Cross-Sectional and Panel Surveys in Germany. International Journal of Public Opinion Research, 37(4).
19. Groves, R. M. & Heeringa, S. G. (2006). Responsive design for household surveys: tools for actively controlling survey errors and costs. Journal of the Royal Statistical Society: Series A (Statistics in Society), 169(3), 439–457.
20. HM Revenue and Customs. (2023). Individuals, Small Businesses and Agents Customer Survey 2022 Technical Annex. https://www.gov.uk/government/publications/hmrc-individuals-small-businesses-and-agents-customer-survey-2022
21. Jackson, M. T., McPhee, C. B., & Lavrakas, P. J. (2020). Using response propensity modelling to allocate noncontingent incentives in an address-based sample: Evidence from a national experiment. Journal of Survey Statistics and Methodology, 8(2), 385-411.
22. Kantar Public (2021) Community Life Survey Technical Report 2020/21. https://doc.ukdataservice.ac.uk/doc/8867/mrdoc/pdf/8867_community_life_technical_report_2020-21.pdf
23. Kantar Public (2023). Attitudes to Mental Illness 2023 Technical Annexes.
24. Luiten, A. & Schouten, B. (2013). Tailored fieldwork design to increase representative household survey response: an experiment in the survey of consumer satisfaction. Journal of the Royal Statistical Society: Series A (Statistics in Society), 176(1), 169–189.
25. Lynn, P. (2012). Failing to locate panel sample members: minimising the risk. Presented at the International Workshop on Household Survey Non-response, Ottawa, Canada, September 2012.
26. Lynn, P. (2016). Targeted appeals for participation in letters to panel survey members. Public Opinion Quarterly, 80(3), 771–782.
27. Lynn, P. (2017a) From standardised to targeted survey procedures for tackling non-response and attrition. Survey Research Methods, 11(1), 93-103.
28. Lynn, P. (2017b) Pushing Household Panel Survey Participants from CAPI to Web. Paper presented at International Workshop on Household Survey Nonresponse, Utrecht, September.
29. McGonagle, K. A., Sastry, N., & Freedman, V. A. (2023). The effects of a targeted “Early Bird” incentive strategy on response rates, fieldwork effort, and costs in a national panel study. Journal of Survey Statistics and Methodology, 11(5), 1032-1053.
30. Schouten, B., Peytchev, A., & Wagner, J. (2018) Adaptive Survey Design. Chapman & Hall / CRC.
31. Sladka, V. & Lynn, P. (2025) Report 5: Targeted procedures for tackling survey non-response: Evidence Review. https://surveyfutures.net/wp-content/uploads/2025/05/report-5-targeted-procedures-evidence-review.pdf
32. Williams, J. (2019). What difference does the provision of paper questionnaires make to the Community Life Survey sample and results? https://assets.publishing.service.gov.uk/media/5ffc54698fa8f563fc973b1c/Impact_of_Paper_Questionnaires_in_the_Community_Life_Survey_V2.pdf
33. Zhang, S., West, B. T., Wagner, J., & Gatward, R. (2024). Incorporating Adaptive Survey Design in a Two-Stage National Web or Mail Mixed-Mode Survey: An Experiment in the American Family Health Study. Journal of Survey Statistics and Methodology, 12(3), 578-592.
34. Zukerberg, A., Hall, D., & Henly, M. (2007). Money can buy me love: experiments to increase response through the use of monetary incentives. Paper presented at the Joint Statistical Meetings of the American Statistical Association. Washington DC: U.S. Census Bureau.