As an experimental psychologist I am interested in the intersection of human-computer interaction, the science of unconscious attitude forming, behavior and cognition, online persuasion and persuasive technology. The principal subjects of my research are:

  1. investigate how to apply priming research to trigger or change online behavior
  2. explore and optimize persuasive strategies implemented with interactive technologies
  3. advance survey methods and behavior observation tecniques to increase participant engagement and data quality

Most of my research focused on online experiments and surveys, primarily using crowdsourcing platforms such as Amazon Mechanical Turk (AMT). In my research I value modern and robust statistical methods and approaches such as mixed-effects modelling (lmer4) or non-parametric analyses. I am also interested in efficient survey design to maximize participants’ motivation, engagement and creativity.

I taught several workshops and courses in introductory statistics, advanced survey design with Qualtrics, R-programming and how to do research on Mechanical Turk.

Metaphor priming and online behavior (thesis project)

Abstract: Recent research on embodiment suggests that priming within a physical domain such as sensation of hardness or rigidity affects cognition, emotion, and behavior related to a psychological domain that is metaphorically linked, such as ‘rigidity in negotiation’. Referring to the metaphor ‘sticking together’, we demonstrate that exposure to products used for physical adhesion, such as magnets, can increase feelings of interpersonal closeness, self-reported phone call behavior (Experiment 1 and 2), increase preference for social-media advertisement measured through click rates (Experiment 3) and increase interest in social networks measured through browsing an interaction information box on social networks (Experiment 4). We hereby demonstrate how priming can manipulate online behavior among internet samples similar to the US internet population by using crowdsourcing platforms such as Amazon Mechanical Turk. We discuss the practical implications of our findings and derive concrete suggestions for how to apply our results to online marketing strategies.

Collaborators: John Bargh, Yale University, Peter Gollwitzer, New York University

Glueckert, K. C., Song, H., & Bargh, J. A. (2013). Adhesion priming

Persuasion profiling and persuasive strategies

Recent research suggests that the effectiveness of persuasive strategies known from social psychology strongly depends on individual preferences. Researchers were able to create individual persuasion profiles for website users in order to dynamically adapt webdesign and product presentations to maximize their persuasive intent. In several studies we attempted to derive a theory and create survey scales that explain why these individual differences can occur.

Collaborators: Dean Eckless, Facebook, Maurits Kaptein, Founder of persuasionapi.com

Increasing motivation and creativity among survey participants

Pre-eliminary research confirmed that survey participants show more engagement, higher creativity and are less distracted when they start a survey in a maximized new window, not a new browser tab. Those results most likely occur because subjects are not distracted by other browser tabs that may prime them to engage in other activites. More studies to confirm and generalize the effect are conducted.

Using Amazon Mechanical Turk (AMT) to collect high quality data

More and more researchers use crowdsourcing tools like AMT to gather high quality experimental data at low cost. Several papers have tested and examined data quality on AMT and the general consensus is that AMT samples provide higher data quality at much lower cost compared to college samples. However, closer analyses revealed that an AMT sub-population called “Superturkers” do most of the research tasks on AMT. This might comprimise certain experimental paradigms as they are already known. In my research I consistently try to improve screening and filtering methods to idenfity outliers, classify workers and eliminate random respondents. More on MIT Deneme Blog and Experimental Turk Blog.