Tackling the Algorithmic Power of Social Media: A Randomized Controlled Trial Examining how Instagram’s Chronological Feed Affects Users’ Daily Well-being.
by Marios Georgiou
Thesis supervisor: Dr. Susanne Baumgartner
Over the past few years, academic scholars and practioners have pointed to several negative aspects of social media algorithms relating to power, commercialization, socialization and surveillance culture. In response to the growing concern about social media algorithms, Instagram has recently relaunched the chronological feed option. Communication scholars and practitioners have argued that replacing the algorithmic feed with a chronological feed is one way to make social media platforms healthier and ensure responsible design. However, there are no empirical findings to support these theoretical claims. With this study, I investigated the effects of Instagram’s chronological feed compared to the algorithmic feed on users’ time spent on Instagram and their daily digital well-being.
A randomized controlled trial experiment was conducted. It was based on a two-week longitudinal field experiment (pre vs post measures), which was fielded in May-June 2022. Participants were randomly assigned to one of the two conditions (control vs experimental). For measuring the dependent variables, the study used a combination of a data donations and self-reported measures, both daily and weekly, for two weeks.
The study consisted of the following parts:
Day 1: Intake survey assessing demographic info
Days 2-7: These days served as a baseline measure (pre-measure), where both conditions used Instagram as usual; the default option is the Instagram algorithmic feed. Participants received daily surveys at 8:00 pm via the app ExpiWell.
Day 8: Participants were redirected to Qualtrics survey instead, which assessed the dependent measures for the past 7 days on longer scales. Also, participants were asked to upload a screenshot of Instagram’s Time spent function which displays the average time one spends on Instagram in the last 7 days (see Figure 1 for examples).
Day 9-14: These days served as an intervention measure (post-measure). During these days, participants in the experimental conditions were instructed and received daily push notifications to change their default Instagram feed to the chronological feed every time they would use Instagram. Whereas participants in the control condition didn’t receive any daily push notifications and used Instagram as usual. Participants again received daily surveys at 8:00 pm via the app ExpiWell.
Day 15: Exit survey assessing the dependent measures on longer scale and asked participants to upload again a screenshot of Instagram’s Time spent function.
To account for the data structure of observations (level 1) nested within conditions (level 2), who were nested within persons (level 3), multilevel modelling (i.e., linear mixed model) was applied. For the daily assessments, there was not any effect of Instagram’s chronological feed versus algorithmic feed on daily time spent on Instagram, the appeal of the feed, digital well-being, and perceived stress. For the weekly assessment, there was only an effect on perceived stress: on average using the chronological feed led to lower perceived stress compared to the use of the algorithmic feed.
This study provided first evidence that the daily use of chronological feed might be beneficial to reduce user’s stress. Key limitations of the study were that the sample size was rather small (N = 45) and the compliance with the use of chronological feed was rather low from participants in the experimental condition (M = 66.70%). Future studies should explore the effects of chronological feed versus algorithmic feed with higher sample size and in a more controlled study settings, as such a study might have provided more significant effects.
The Digital Communication Methods Lab is an initiative of the Research Priority Area Commmunication, at the University of Amsterdam.