The Impact of Screen Size on Crowdsourced Image Classification

Screen size has potential to affect the results of crowdsourced image classification and impact data quality, even for simple tasks like determining whether an animal is present in an image or not. We evaluated the rates at which volunteers judged images to be blank for six Snapshot Safari citizen science projects on the Zooniverse platform, and found that in all but one project, people using mobile devices were more likely to say an image was blank than those using a desktop computer. Qualitative evaluation further demonstrated that screen brightness can also play a role in reliability for volunteers using mobile devices. These findings can be taken into consideration in the design and testing of apps and platforms for crowdsourced image classification to support reliability and data quality.