My Research
My research focuses on nonresponse and measurement error in modern data collection methods, in particular (mobile) Web surveys and passive mobile data collection. Below you find a detailed description of my research interests.
Web survey methodology
Collecting survey data over the internet has emerged as one of the main data collection methods in academic, social, marketing, and public opinion research within a very short period of time. This results in a need for additional research that helps fully understanding what drives people to participate in Web surveys, and more recently, in mobile Web surveys. In my research, I have examined the applicability of different survey participation theories to Web surveys (Keusch, 2015). I have applied involvement and leverage salience theory to explain the participation decision and response behavior process in nonprobability online panels and to understand the role of personal survey topic interest during the survey participation request (Keusch, 2013). I have analyzed how intrinsic and extrinsic reasons for joining online panels predict survey participation behavior (Keusch, Batinic, & Mayerhofer, 2014). Additionally, I have used cognitive dissonance theory to explain the influence of prenotification messages on response behavior in list-based online samples (Keusch, 2012). I was also part of a team that developed a typology for the use of paradata in web surveys (McClain et al., 2018).
Using smartphones for data collection
The increasing popularity of smartphones poses new challenges for researchers but also opens up new opportunities for novel ways of data collection. For example, a rising share of respondents access Web surveys that were designed to be taken on large screen computers on their smartphones (a.k.a. “unintentional mobile respondents”), and recent research demonstrates the impact of smartphones on survey participation and data quality (Keusch & Yan, 2017). On the other hand, researchers are now also able to collect data from smartphone users – such as geolocation, online behavior and browser history, app usage – through passive mobile data collection via apps and sensors. Compared to surveys that rely on self-reported data, passive data collection has the potential to provide richer data (because it is almost constantly collected), to decrease respondent burden (because fewer survey questions need to be asked), and to reduce measurement error (because of less forgetting and social desirability). However, to collect passive data from smartphones, participants need to agree to download a research app that tracks their behavior and location over a longer period of time. In a number of projects currently in the field, I address concerns about nonresponse and consent as well as privacy concerns that come with this new form of data collection.
In collaboration with Christopher Antoun (University of Maryland), Bella Struminskaya (Utrecht University), Mick Couper (University of Michigan), and Frauke Kreuter (University of Maryland), I conducting a vignette study in Germany that assess the circumstances under which smartphone users are willing to participate in passive mobile data collection (Keusch et al., 2019). In addition, together with researchers from the Institute for Employment Research, Germany, I am recruited a panel of smartphone owners who actually downloaded such a research app. This allows us to combine behavior and location data collection with short mobile Web surveys to answer both methodological questions and substantial questions about social participation of unemployed and formal and informal ways of job-seeking, among others (Kreuter et al., 2018). Funded by a grant from the Mannheim Centre for European Social Research (MZES), I am also exploring the feasibility to use the same technology to collect data from refugees in German refugee residences (Keusch et al., 2019).
Measurement errors in rating scales
I also study the influence of the visual design of rating scales, in particular scale direction, on response behavior. I am especially interested in the underlying cognitive mechanisms that lead to response order effects. Findings from this research show strong response effects caused by the direction of rating scales – responses are shifted toward the beginning of the scale for attitudinal questions regardless of the content of the item (Yan & Keusch, 2015; Liu & Keusch, 2017). A number of experiments, conducted together with Ting Yan (Westat), help us to identify the influence of scale direction in Web vs. mobile Web surveys (Keusch & Yan, 2016), for different types of rating scales (Yan, Keusch, & He, 2018), and when respondents show satisficing behavior when responding to the survey (Keusch & Yan, 2018).
Survey gamification
I am also interested in the potential influence of gamification (i.e., application of game mechanics) on measurement and nonresponse error in Web surveys. Reports about declining response rates and increasing satisficing behaviors (e.g., speeding and straightlining) among respondents in Web surveys have brought researchers to propose to make the survey experience more game-like to keep respondents engaged. Gamification in surveys is currently a hot topic, especially among commercial survey organizations, and some researchers/companies claim to use gamification successfully to engage respondents in the survey process. However, so far the scientific literature lacks a theoretical framework for gamification in the survey context. A first paper, co-authored by Chan Zhang (Zhejiang University), reviews the current state of the survey literature on gamification and identifies gaps for future research (Keusch & Zhang, 2017).
Visual design effects in questionnaires
In the context of visual design effects on open-ended questions, I have analyzed differences in response behavior in list-style open-ended questions, i.e., questions asking respondents to name all items (e.g., brands) they know of a specific category, between formats that provide the respondent with one large answer box and those that provide ten small answer boxes (Keusch, 2014). I found that the overall number of brands named, the proportion of low accessibility brands named, and the time that the respondents needed to give an answer to the question were significantly higher if ten small answer boxes were presented. These findings confirmed that respondents utilize cues from the design of the survey instrument in their answer process as proposed by visual design theory.
Applied questionnaire design
Apart from my more methodological interests, I have consulted other researchers in the development of questionnaires for their respective research fields. For example, I was involved in the development and testing of the questionnaire for the University of Michigan Energy Survey that measures attitudes toward energy issues, such as reliability, affordability, and environmental impact, on a national level on a quarterly basis (De Cicco et al., 2015).
I also was involved in the development, testing, and analysis of data from a rider to the Survey of Consumers (SCA) on weather and mobility that examined the impact of weather on the daily lives of US adults to understand which populations are most vulnerable to various weather conditions (Clarke et al., 2015; Wagner et al., 2016).
Together with researchers from the University of Michigan Bone Marrow Transplant (BMT) Program, I have worked on the development of a survey among BMT patients about their attitudes and perceptions of clinical trials in BMT. Based on the findings from focus groups and semi-structured interviews with pre- and post-BMT patients (Keusch et al., 2014), the research team will develop a survey that can be administered to patients on iPads in the clinic.
Web survey methodology
Collecting survey data over the internet has emerged as one of the main data collection methods in academic, social, marketing, and public opinion research within a very short period of time. This results in a need for additional research that helps fully understanding what drives people to participate in Web surveys, and more recently, in mobile Web surveys. In my research, I have examined the applicability of different survey participation theories to Web surveys (Keusch, 2015). I have applied involvement and leverage salience theory to explain the participation decision and response behavior process in nonprobability online panels and to understand the role of personal survey topic interest during the survey participation request (Keusch, 2013). I have analyzed how intrinsic and extrinsic reasons for joining online panels predict survey participation behavior (Keusch, Batinic, & Mayerhofer, 2014). Additionally, I have used cognitive dissonance theory to explain the influence of prenotification messages on response behavior in list-based online samples (Keusch, 2012). I was also part of a team that developed a typology for the use of paradata in web surveys (McClain et al., 2018).
Using smartphones for data collection
The increasing popularity of smartphones poses new challenges for researchers but also opens up new opportunities for novel ways of data collection. For example, a rising share of respondents access Web surveys that were designed to be taken on large screen computers on their smartphones (a.k.a. “unintentional mobile respondents”), and recent research demonstrates the impact of smartphones on survey participation and data quality (Keusch & Yan, 2017). On the other hand, researchers are now also able to collect data from smartphone users – such as geolocation, online behavior and browser history, app usage – through passive mobile data collection via apps and sensors. Compared to surveys that rely on self-reported data, passive data collection has the potential to provide richer data (because it is almost constantly collected), to decrease respondent burden (because fewer survey questions need to be asked), and to reduce measurement error (because of less forgetting and social desirability). However, to collect passive data from smartphones, participants need to agree to download a research app that tracks their behavior and location over a longer period of time. In a number of projects currently in the field, I address concerns about nonresponse and consent as well as privacy concerns that come with this new form of data collection.
In collaboration with Christopher Antoun (University of Maryland), Bella Struminskaya (Utrecht University), Mick Couper (University of Michigan), and Frauke Kreuter (University of Maryland), I conducting a vignette study in Germany that assess the circumstances under which smartphone users are willing to participate in passive mobile data collection (Keusch et al., 2019). In addition, together with researchers from the Institute for Employment Research, Germany, I am recruited a panel of smartphone owners who actually downloaded such a research app. This allows us to combine behavior and location data collection with short mobile Web surveys to answer both methodological questions and substantial questions about social participation of unemployed and formal and informal ways of job-seeking, among others (Kreuter et al., 2018). Funded by a grant from the Mannheim Centre for European Social Research (MZES), I am also exploring the feasibility to use the same technology to collect data from refugees in German refugee residences (Keusch et al., 2019).
Measurement errors in rating scales
I also study the influence of the visual design of rating scales, in particular scale direction, on response behavior. I am especially interested in the underlying cognitive mechanisms that lead to response order effects. Findings from this research show strong response effects caused by the direction of rating scales – responses are shifted toward the beginning of the scale for attitudinal questions regardless of the content of the item (Yan & Keusch, 2015; Liu & Keusch, 2017). A number of experiments, conducted together with Ting Yan (Westat), help us to identify the influence of scale direction in Web vs. mobile Web surveys (Keusch & Yan, 2016), for different types of rating scales (Yan, Keusch, & He, 2018), and when respondents show satisficing behavior when responding to the survey (Keusch & Yan, 2018).
Survey gamification
I am also interested in the potential influence of gamification (i.e., application of game mechanics) on measurement and nonresponse error in Web surveys. Reports about declining response rates and increasing satisficing behaviors (e.g., speeding and straightlining) among respondents in Web surveys have brought researchers to propose to make the survey experience more game-like to keep respondents engaged. Gamification in surveys is currently a hot topic, especially among commercial survey organizations, and some researchers/companies claim to use gamification successfully to engage respondents in the survey process. However, so far the scientific literature lacks a theoretical framework for gamification in the survey context. A first paper, co-authored by Chan Zhang (Zhejiang University), reviews the current state of the survey literature on gamification and identifies gaps for future research (Keusch & Zhang, 2017).
Visual design effects in questionnaires
In the context of visual design effects on open-ended questions, I have analyzed differences in response behavior in list-style open-ended questions, i.e., questions asking respondents to name all items (e.g., brands) they know of a specific category, between formats that provide the respondent with one large answer box and those that provide ten small answer boxes (Keusch, 2014). I found that the overall number of brands named, the proportion of low accessibility brands named, and the time that the respondents needed to give an answer to the question were significantly higher if ten small answer boxes were presented. These findings confirmed that respondents utilize cues from the design of the survey instrument in their answer process as proposed by visual design theory.
Applied questionnaire design
Apart from my more methodological interests, I have consulted other researchers in the development of questionnaires for their respective research fields. For example, I was involved in the development and testing of the questionnaire for the University of Michigan Energy Survey that measures attitudes toward energy issues, such as reliability, affordability, and environmental impact, on a national level on a quarterly basis (De Cicco et al., 2015).
I also was involved in the development, testing, and analysis of data from a rider to the Survey of Consumers (SCA) on weather and mobility that examined the impact of weather on the daily lives of US adults to understand which populations are most vulnerable to various weather conditions (Clarke et al., 2015; Wagner et al., 2016).
Together with researchers from the University of Michigan Bone Marrow Transplant (BMT) Program, I have worked on the development of a survey among BMT patients about their attitudes and perceptions of clinical trials in BMT. Based on the findings from focus groups and semi-structured interviews with pre- and post-BMT patients (Keusch et al., 2014), the research team will develop a survey that can be administered to patients on iPads in the clinic.
