A YouTube Shorts study found an effect only at the very top

The test

A preprint titled 'Shorts on the Rise: Assessing the Effects of YouTube Shorts on Long-Form Video Content', posted to arXiv on 28 February 2024 by Prajit T. Rajendran, Kevin Creusy and Vivien Garnes, studies whether adding YouTube Shorts to a channel changes the performance of that channel's existing long-form videos.

What the evidence says

The paper's methodology states that the authors collected data on 250 YouTube creators with at least 100,000 subscribers each, each having published at least eight Shorts and eight long-form videos before 31 December 2023, drawing on 170,690 videos in total. For each creator, the authors used the publication date of that channel's first Shorts video as a before-and-after marker, then normalized view counts for the time videos had been online. The abstract reports 'a significant decrease in both view counts and engagement in long-form videos' following a channel's adoption of Shorts, but the results section narrows that claim considerably: using a Wilcoxon signed-rank test, the decrease reached statistical significance only for creators with more than 5 million subscribers, while the effect was not statistically significant for the other subscriber tiers the study defined. The paper is a preprint; no confirmed peer-reviewed publication is noted on its arXiv listing, and arXiv itself states that its hosted materials are not peer-reviewed.

The sample and the variance

The 250-creator sample was drawn from named content categories, entertainment, education, lifestyle and technology among them, all on YouTube specifically, and the finding does not extend to TikTok, Instagram Reels or Snapchat Spotlight, none of which the paper examines. Because the statistically significant effect the authors found was concentrated among the largest channels, a claim that Shorts adoption reduces long-form views, drawn from this paper's abstract alone, overstates what the underlying subscriber-tier breakdown actually supports for a typical mid-sized channel.

What to try next

A creator or channel strategist reading this paper should look at the subscriber-tier results, not just the abstract, before drawing conclusions about their own channel's likely experience, and should treat the 2023 cutoff date as a real boundary on how current the observed pattern is. This is a reading suggestion grounded in the paper's own tables, not growth advice.

• Does a claim drawn from this paper describe the full 250-creator sample or only the largest, over-5-million-subscriber tier where the effect was significant?
• Does the channel or claim in question fall inside the YouTube-only, pre-2024 dataset this paper studied?
• Has a peer-reviewed version of this preprint since appeared, and does it change the reported findings?

The paper's own subscriber-tier breakdown is more informative than its headline finding, showing a real but narrowly concentrated pattern rather than a universal trade-off between short and long-form video on YouTube.