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

Tuesday 14th July, 11:00 - 12:30 Room: HT-105

Web surveys: challenges and strategies

Convenor Dr Maria Clelia Romano (ISTAT )
Coordinator 1Dr Francesca Gallo (ISTAT)

Session Details

Over the past decade the use of Internet has brought a profound effect on the survey methodology and the CAWI technique has received great attention from social researchers.
Social surveys can greatly benefit of the web techniques for two main reasons: on the one hand, the relative low cost of conducting web surveys makes them very competitive and well suited to the current budget constraints; on the other hand, they give a better chance to reach subgroups of population hardly accessible with other techniques.
Furthermore, the combined use of different survey techniques allows to improve the target population coverage and to overcome the drawbacks related to each single technique.
Nevertheless, the CAWI technique requires a great effort in order to design or redesign the entire survey system: the questionnaire, the interaction mode with the respondents, the reminder system and so on.
Although recent literature has provided interesting contributions, there are still gaps in our knowledge on how to deal with official government surveys, which are often much complex and involve a great deal of editing.
The researchers have to face old and new challenges, in a new context, which is strongly characterized by the spread of new technologies. Some of these challenges include the optimal design for questions that require complex coding (e.g. economic activity or occupation), the development of effective instructions to explain complex statistical concepts to respondents or the navigation improvement for complex household questionnaires.
Moreover, in the case of mixed mode, which techniques to combine in order to get the best from each? In which sequence?
The proposed session aims at sharing and discussing the ongoing experiences on web surveys on individuals and on mixed mode techniques including web. New approaches and best practices to improve data quality of the NSIs web surveys are extremely welcome.

Paper Details

1. Correlates of early and late responses to surveys in an online panel
Dr Salima Douhou (CentERdata, Tilburg University)
Mrs Corrie Vis (CentERdata, Tilburg University)

Keeping the survey open for completion if the net amount a researcher needs for his/her studies is reached involves extra costs in terms of incentives and other processing costs. Some respondents prefer to complete surveys immediately after the invitation is sent out, while others wait until right before the closing time of the survey. Some preliminary results show that early and late responders are different in terms of sociodemographics and personality. We expect that if the variables of interest for the researcher is correlated with characteristics of timing of response this may have an impact on the outcomes.

2. Is there any medicalization in celiac disease?
Dr Beba Molinari (Università degli studi di Genova)
Professor Cleto Corposanto (Unviersità Magna Graecia di Catanzaro)

The paper aims to highlight how new web-survey can become a useful tool for analysis of social reality, not losing in statistical representation if used not only to census studies, but also through sampling conducted ad hoc.
In addition to the presentation of the results of the study conducted on over 5,000 celiac will be discussed also the sampling error of the breack off and drop out and if these measures, used in classical quantitative research, may still be valid for the web-survey

3. Nonresponse and Measurement Bias in Web surveys
Ms Anke Metzler (Darmstadt University of Technology)
Mr Marek Fuchs (Darmstadt University of Technology)

Web surveys suffer from substantial nonresponse inducing nonresponse bias in estimates. Various groups cause nonresponse: initial nonrespondents and survey break-offs. A detailed understanding of the relative impact of initial nonresponse and survey break-offs is necessary to select proper strategies. Reducing survey break-off may cause additional measurement bias. Results indicate that the survey break-off bias and the initial nonresponse bias cancel each other and that responses of survey break-offs are less accurate. Therefore, survey break-off should not only be considered regarding the nonresponse bias prevention but also in terms of the measurement error.

4. The first ISTAT web survey on individuals: main choices and first results
Ms Maria Clelia Romano (ISTAT (Italian National Institute of Statistics))
Ms Francesca Gallo (ISTAT (Italian National Institute of Statistics))

In the context of a wide modernisation process of social surveys, in 2013 the Italian National Institute of Statistics launched the carrying out of the first web survey on individuals. The chance for introducing web data collection in population surveys has been supplied by the second edition of “Doctorate Holders’ Vocational Integration
The designing has been a very complex work with many steps including the pre-test that allowed to tune the pilot survey. The presentation will aim to share the whole experience, highlighting the main methodological choices, strategies and outcomes achieved during the pilot survey and the final survey

5. Comparison of different mixed-mode and face - to face surveys - response rates and costs
Dr Mare Ainsaar (senior research fellow)
Mr Reigo Hendrikson (analyst)

The presentation provides comparative results from two surveys combining web-mode and paper- mode on interview mode. We compare the results of the mixed mode survey with the similar face-to face surveys conducted at the same time and compare the results with results of European Social Survey mixed mode experiment. For both mixed mode surveys a sequential design was implemented and all participants were first invited by mail to complete a survey online or to inform about their inability of doing this. This presentation compares the costs, response rates, non-response structures of all surveys.