What Makes a "Bad" Ad? User Perceptions of Problematic Online Advertising

This is the dataset used in Survey 2 of our paper:

Eric Zeng, Tadayoshi Kohno, Franziska Roesner. "What Makes a 'Bad' Ad? User Perceptions of Problematic Online Advertising." In CHI '21: Proceedings of the 2021 CHI Conference on Human Factors in Computing Systems.

PDF https://doi.org/10.1145/3411764.3445459

We surveyed over 1000 participants and asked for their opinions on ads in our dataset - 500 ads scraped from the top sites on the web. Participants labeled each ad with one or more opinion labels describing their subjective opinion of the ad, with 10 participants assigned to each ad. Here, we show the opinion label distributions for each ad (the percentage of participants who agreed with each label).

Try clicking on any column label to sort by the column values. Click on the Ad ID to see the screenshot and participant comments about the ad.

Description of columns

Ad ID: identifier and filename for the screenshot of the ad
Content Labels: Researcher-created labels describing the format and topic of the ad (and any misleading techniques used)
Cluster: Cluster ID generated from the distribution of participants' opinion labels (i.e. ads that the population of people felt similarly about)
Avg. Rating: The average ratings given to the ad by participants, on a 7 point Likert scale
Opinion Labels: The percentage of participants who labeled the ad with each opinion label

Explore Ad Opinion Clusters Download Dataset on GitHub