Hauke Roggenkamp

Institute of Behavioral Science and Technology, University of St. Gallen


October 10, 2023

oFeeds Logo

1 Preface

The project oFeeds contains three apps

  1. oCom,
  2. oNovitas and
  3. oTweet

that mimic shopping, news as well as social media feeds, respectively. This document describes how to configure each app to your needs.

oFeeds Screenshots

All of these apps are based on oTree (Chen, Schonger, and Wickens 2016). Hence, all of these apps share some common ground that will be described in the first few chapters: Chapter 2 first describes how to deploy oTree apps in general. Chapter 3 describes how to create and configure a session. Subsequently, Chapter 4, Chapter 5 and Chapter 6 describe the session configurations that apply to the respective apps in detail. The document closes with instructions on how to run an oTree experiment on Prolific.

1.1 Features

Improve realism (and thus, external validity) without sacrificing internal validity:

  • Manipulate the content of whole feeds instead of single posts. This allows you not only to investigate the composition but also sorting ‘algorithms’ of feeds.
  • Display ads as skyscrapers or native ads with perfect control as there are no competing advertisers (as in Platform A/B-Tests [PABTs]).
  • You can thus, not only manipulate and randomize the feed’s content but also the ads (both: copy and creatives).
  • You can gather implicit measures (e.g. scrolling behavior, which screenshots usually do not provide).
  • oFeeds provides individual-level data that allows you to further investigate mechanisms (in contrast to PABTs’ aggregate data).
  • Randomized and stable treatment assignment (no divergent delivery, see e.g., Braun and Schwartz 2023).
  • As of today, oFeeds is a single player game without interaction (i.e. no social network) between participants to avoid spillovers between treatments and SUTVA (Angrist, Imbens, and Rubin 1996, 446) violations.
  • Based on oTree: even though we provide a graphical admin user interface to adapt sessions to your needs, you can go one step further and implement far-reaching changes easily, because there are oTree workshops, a good documentation and a vivid community. If you can’t do it, your lab manager surely can.
  • One can combine oCom with oTweet or oNovitas to simulate a shop as a landing page of some campaign.

Please note that both the apps as well as the documentation are under development (in fact, Chapter 4, Chapter 6 and Chapter 7 don’t even exist). Hence, feedback is much appreciated. If you have any comments, questions or requests, don’t hesitate to reach out to me via email or create an issue in the corresponding github repository.

1.2 License & Citation

oFeeds is licensed under the MIT open source license with the added requirement of a citation of Roggenkamp (2023). This licensing approach provides users with the liberty to utilize, adapt, and share the software with minimal constraints to foster collaborative development in line with open-source and open science principles.

  author={Roggenkamp, Hauke C},