Work Text:
POPULARITY, WORD COUNT, AND RATINGS ON AO3.
[FAQ] [More fandom stats]
TL;DR: SUMMARY
- The most popular fanworks to produce are General Audiences and Teen And Up Audiences.
- The most popular fanworks to read are Explicit. In general, the higher the rating, the more hits and kudos a work gets.
- This is partly a side effect of the average word count in these categories; I've written before about how longer fanfic gets more hits, and we can see from the above graphs that the more explicit the fanwork, the higher the average word count, as well.
- When you control for word count, there remains a smaller effect of rating -- in particular, fanworks rated "Explicit" get far more hits on average than others.
- I'll need to reevaluate my previous post about categories and popularity in light of the word count effect.
EXPT 1: POPULARITY OF DIFFERENT RATINGS
Methods. I used the same methods as in my previous post on popularity metrics, relationship categories, and fandoms. Because of the problems with the median kudos/median hits metric (and its inconclusiveness), I omitted that metric this time around.
Ratings categories:
- Not Rated
- General Audiences
- Teen And Up Audiences
- Mature
- Explicit
Results and discussion. Figure 1 shows the number of fanworks produced in each rating category. Because these are mutually exclusive categories, I also displayed this data as a pie chart in Figure 2, showing the percentages of AO3 fanworks that have each rating.
The most popular fanworks to produce are General Audiences and Teen And Up Audiences. However, when we look at the hits and kudos graphs (which show very similar patterns, as they did in my previous analysis), we see that Explicit fanworks get more attention than all the rest, and in general, the higher the rating, the more hits & kudos. (Note: "Not rated" is on the left hand side of the graph, but is not the least explicit category.)
However, it occurred to me this time around to wonder about word count. In the past I found that longer fanfic gets more hits. When I looked at the average word counts for the different categories (Fig 5), the graph looked suspiciously similar to the hits/kudos. Could that be partially or entirely responsible for this effect?
EXPT 2: CONTROLLING FOR WORD COUNT
Methods. I decided to repeat the analysis looking at groups of fanworks that were approximately the same length. This should help tease apart the effect of word count vs. rating. E.g., if I were to only look at fanfic that is exactly 1000 words long, any differences would have to be due to something other than word count.
In order to have very many fanworks returned by my searches, I allowed the word count to vary by 10%. I redid the analysis for the following word count slices:
- 1000-1100 words (very light purple graph)
- 5000-5500 words (light purple)
- 10000-11000 words (purple)
- 50000-55000 words (dark purple)
Within these slices, the minimum number of works returned was 390 works (General Audiences, 50-55K words). The maximum was 10545 works (General Audiences, 1000-1100 words).
Results and discussion. The final four graphs show the comparative number of hits for each of the ratings in the four different word count slices. (Note: the y axes are the same for the first three graphs, but different for the final graph -- necessary because there are so many more hits in that case.)
Even once word count is factored out, Explict-rated fics always get more hits. And there is still an effect of rating on hit rate -- the more explicit ratings get more hits. The effect seems to increase for larger word counts (I would need more data to be certain of that trend, however).
FUTURE WORK
I'll need to go back and try controlling for word count in my previous analysis of popularity metrics. And in the future, I'll need to take word count into account whenever looking at popularity metrics.
Edit: I’ve since compared ratings on AO3 vs. FFN.