BigSurv18 program


Wednesday 24th October Thursday 25th October Friday 26th October Saturday 27th October





Smartphone Sensor Measurement and Other Tasks in Mobile Web Surveys I

Chair Dr Bella Struminskaya (Utrecht University)
TimeSaturday 27th October, 11:00 - 12:30
Room: 40.002

Smartphones allow researchers to collect data through sensors such as GPS and accelerometers to study movement, and passively collect data such as browsing history and smartphone and app usage in addition to self-reports. Passive mobile data collection potentially decreases measurement errors and reduces respondent burden. However, respondents have to be willing to provide access to sensor data or perform additional tasks (e.g., download apps, take pictures). If willing respondents differ from nonwilling respondents, results might be biased. This session brings together empirical evidence on the state-of-the-art use of sensor measurement and other additional tasks on smartphones. It combines presentations of results from (large-scale) studies with diverse sensors and tasks from multiple countries and research settings. Presentations discuss current practice in collecting these new types of data focusing on the willingness to allow sensor measurement and perform additional tasks and its implications for nonparticipation bias.

Emergent Issues in the Combined Collection of Self-Reports and Passive Data Using Smartphones

Professor Frederick Conrad (University of Michigan) - Presenting Author
Professor Florian Keusch (University of Mannheim)

Download presentation

The integration of smartphones into everyday life has made it possible to capture information about users’ movements and activities via sensors such as GPS, accelerometers, gyroscopes and pedometers. In addition, it is possible to measure users’ behavior using apps as they interact with their smartphones and engage in online activities, enabling measurement of social interaction and social networks, e.g., by identifying other smartphone users in physical proximity or who the user has phoned, texted, etc. Social researchers are enthusiastic about the availability of this type of data for several reasons. First, the information can be collected almost continuously and at relatively low -- or even no -- burden to participants. Second, because smartphones have become central to daily life they are, in effect, extensions of people’s bodies which means the devices are present in the same contexts as their users, enabling in situ measurement which is generally not possible through traditional self-report. Third, automatically collected sensor data and traditional self-reports have complementary strengths and weaknesses so integrating the two can potentially provide a richer picture of social phenomena than can be constructed from either type of data alone. Depending on how the participants are recruited, the benefits of data collected via smartphone sensors and apps can potentially be paired with representative samples of sufficient statistical power to produce population estimates.

In this presentation we summarize the existing research that combines data from self-reports with data from sensors and apps, that is, two or more types of data from the same participants collected on their smartphones. We have developed a taxonomy of this research that consists of about 15 dimensions such as the type of sample (probability vs. convenience sample), which type of data is primary (self-reports or passively collected data), the type of passively collected data (e.g., location, nearby devices, movement, phone call logs), the frequency with which they are passively collected (e.g., around the clock or when the app is on), whether users need to classify the data (e.g., indicating which GPS coordinates correspond to home, work, and other locations), how much user action is required to collect data (e.g., pressing a button on a camera vs. walking), the field of study, application domain, etc. By using this taxonomy we are able to characterize the state-of-the-art combining self-report and passively collected data.

Our goal is to help tame the combinatorial explosion that is created by crossing the many distinctions among survey data -- even just survey data collected on smartphones -- with the many distinctions among types of passively collected data. We will report themes that emerge from our review and classification of this research into the taxonomy. The presentation will provide a snapshot of progress to date and a blueprint for where future research might go.


Combining Active and Passive Mobile Data Collection: A Survey of Concerns

Final candidate for the monograph

Professor Florian Keusch (University of Mannheim) - Presenting Author
Professor Frauke Kreuter (University of Mannheim, University of Maryland, Institute for Employment Research)
Dr Bella Struminskaya (Utrecht University)
Professor Martin Weichbold (University of Salzburg)

Download presentation

Smartphone use is on the rise worldwide, and researchers are exploring novel ways to leverage the capabilities of smartphones for data collection. Mobile surveys, i.e., surveys that are filled out on a smartphone web browser or through an app, are already extensively studied. On the other hand, research on the use of other features of smartphones that allow researchers to automatically measure an even broader set of characteristics and behaviors of their users that go far beyond the collection of mere self-reports is still in its infancy. For example, smartphone users can now be asked to take pictures of receipts to better measure expenditure, to agree to tracking of movements to create exact measures of mobility and transportation, or to automatically log app use, Internet searches, and phone calling and text messaging behavior to measure social interaction. These forms of data collection provide richer data (because it can be collected in much higher frequencies compared to self-reports) and have the potential to decrease respondent burden (because fewer survey questions need to be asked) and measurement error (because of reduction in recall errors and social desirability). However, agreeing to engage in these forms of data collection from smartphones is an additional step in the consent process, and participants might feel uncomfortable sharing specific data with researchers due to security, privacy, and confidentiality concerns. Moreover, users might have differential concerns with different types of data collection on smartphones, and thus be more willing to engage in some of these data collection tasks than in others. In addition, participants might differ in their skills of smartphone use and thus feel more or less comfortable using smartphones for research, leading to bias due to differential nonparticipation of specific subgroups.

Over the course of the last year we have asked a set of questions related to concerns when collecting different types of data on smartphones in four surveys among smartphone owners; two surveys in German non-probability online panels (n1~2,700; n2~1,200), one survey in a German probability online panel (n3~2,300), and one probability-based web survey in Austria (n4~500). This presentation will report on the results of these four surveys and answer the following research questions:

• Does concern about participating in smartphone data collection differ by type of data collected on a smartphone?
• Do general and specific concerns about privacy and data security correlate with concerns about smartphone data collection?
• What is the role of trust in data collecting organizations when it comes to concerns about smartphone data collection?
• Does concern about smartphone data collection vary across subgroups of smartphone users with different levels of smartphone skills and smartphone use habits?


Collecting Smartphone Sensor Measurements in the General Population: Willingness and Nonparticipation Bias

Dr Bella Struminskaya (Utrecht University) - Presenting Author
Dr Peter Lugtig (Utrecht University)
Professor Barry Schouten (Statistics Netherlands, Utrecht University)
Dr Vera Toepoel (Utrecht University)
Dr Marieke Haan (University of Groningen)
Mr Ralph Dolmans (Statistics Netherlands)
Ms Vivian Meertens (Statistics Netherlands)
Ms Deirdre Giesen (Statistics Netherlands)
Dr Annemieke Luiten (Statistics Netherlands)

Using smartphone sensors to collect data in addition to traditional methods of data collection can offer social science researchers detailed data about human behavior, reduce respondent burden by eliminating certain survey questions and improve measurement accuracy by replacing or augmenting self-reports. However, respondents have to be willing and able to use their smartphones to collect sensor data. If those not willing and/or not able differ from those that are, the results will be biased. Growing number of studies investigates the willingness to collect sensor data either in-browser or through apps that need to be downloaded, however, the research on the mechanisms of willingness and actual participation is still much needed.

We study performing actual in-browser smartphone sensor measurements, and, specifically, the role of (1) the wording of the consent question, (2) the assurance of confidentiality, and (3) the ability to control the data being collected. We randomly assign smartphone and tablet users from the general population survey, a methodological study conducted by Statistics Netherlands (N about 2000), to three conditions (gain/loss framing, additional assurance of confidentiality, and ability to revoke measurements) and ask them to share the GPS location of their smartphone and take photos and videos. For those not willing to collect smartphone sensor data, we ask for the reasons of nonwillingness and under what conditions they would change their behavior. We will answer the following research questions: 1) what are the participation rates for different types of sensors (GPS, camera) and do they differ by specific tasks (taking photos with different content, taking videos), 2) what is the optimal way to ask respondents for consent to smartphone sensor measurement, 3) what factors (privacy concerns, smartphone use skills, technical problems) identified by respondents could be addressed to improve the consent rates?

This presentation will advance the understanding of mechanisms of participation in smartphone sensor measurement and provide advice for practitioners on wording of the consent questions for smartphone sensor data collection.


Data Collection Using Mobile Technologies: Changes Over Time in the Barriers to Participation

Professor Annette Jäckle (University of Essex) - Presenting Author
Mr Alexander Wenz (University of Essex)
Professor Mick Couper (University of Michigan)

Download presentation

Survey practitioners are increasingly looking at new measurement tools that go beyond standardised survey questions. For example, the inbuilt technologies in mobile devices, including smartphones and tablets, allow capturing new forms of data such as accelerometry, GPS positioning, or images. The potential value of such technologies for research is that they can be used to measure concepts that cannot be measured with survey questions, or to increase the level of detail or accuracy of measures.

To date, surveys that have trialled data collection using mobile technologies have generally had low participation rates, and, for example, found that participation is related to age and education, usage and familiarity with mobile devices, and attitudes towards data security. Ownership of mobile devices is, however, steadily increasing amongst all age groups, and people are using their devices for an increasing array of different tasks.

Previous research also suggests that people are more willing to do some types of tasks than others: when asked whether they would participate in a range of different mobile data collection tasks, respondents, for example, state higher willingness to participate in tasks where they can control the content of the data transmitted than in tasks that capture data passively. Respondents also state lower willingness to participate in tasks that require downloading apps onto their devices.

In this paper we examine what the increasing ownership and familiarity with mobile devices implies for the future scope of data collection. We examine the following research questions:

(1) How are the barriers to participation in mobile data collection, represented by access to devices, skills, willingness to participate in different types of tasks, and data security concerns, changing year on year?
(2) Are people consistent over time in their stated willingness to do different types of tasks?
(3) Are the correlates of stated willingness to participate in different types of tasks changing over time?
(4) How does asking respondents in the first wave of a panel versus a later wave affect their stated willingness to participate in mobile data collection tasks?

We use data from the UK Household Longitudinal Study Innovation Panel, a probability sample of households in Great Britain, that interviews all household members aged 16+ annually. The 2016 and 2017 interviews included a module of questions about the potential barriers to participation in mobile data collection. This included questions about the ownership and usage of smartphones and tablets, the frequency of use, self-rated skills, the type of tasks done with the device, attitudes towards sharing information via mobile technologies, and stated willingness to participate in different types of data collection using their mobile devices. The 2017 survey also included a refreshment sample interviewed for the first time.