In my post last week, I took a look at how the most popular pairings in Harry Potter fanfiction changed over time. Specifically, I looked at the rate at which fics were published. This time, I wanted to look more closely at a wider range of pairings, as well as other measures of popularity (i.e. favorites, follows, and reviews).

## The data

The dataset is the same from last week, e.g. the descriptive info from all 724,315 Harry Potter fanfics I could scrape from Fanfiction.net. From these, I looked only at fics with either (a) two characters, or (b) fics with more than two characters where pairings were explicitly marked with brackets. This shortens the number of fics to 505,406 consisting of 5,414 unique pairings.

Below are the 20 most popular of these pairings (in terms of number of fics). At the top are Draco/Hermione and Draco/Harry, which have more than 50,000 fics each. To give you a sense of that number, this means that there were, on average, more than 8 fics published per day*. ***For 17 years**.

To take a closer look at other popular pairings, I decided to consider the top 100 ranked pairings. These pairing each have more than 500 fics apiece and make up a large majority of all the fics with pairings. Here are some stats for the top 100 pairings.

Number of pairings considered | 100 |

Percentage of all pairings | 1.8% |

Number of fics with these pairings | 431,337 |

Percentage of all fics | 85.3% |

Therefore, the bottom 98% of pairings only account for ~15% of fics. This may be some serious inequality, but sadly I don’t think anybody’s going to be protesting for more Harry/Stan Shunpike fics. Even though Starry is an amazing portmanteau for a pairing. #OccupyDiagonAlley #MakeStarryHappen.

## Pairings with the most favorites

The next thing I looked at were average number of favorites, follows, and reviews that each pairing receives. These average values were highly correlated with each other, so for the sake of simplicity I’m only showing the number of favorites. Sorting the pairings by follows or reviews does switch up some of the rankings, but there is a large degree of overlap.

Here are the top 20 pairings with the most favorites per fic, i.e. the highest average values.

I think that there are a few interesting things to point out

- Daphne/Harry and Fleur/Harry have ~800 and ~600 fics, respectively, but still have averages of >500 favorites. This is impressive, considering that over 98% of all fics have fewer than 500 favorites.
- Two of the top 10 most popular pairings made it to this list: Harry/Severus (#9) and Harry/Hermione (#4). This is also impressive, since their average values are from ~16,000 and ~25,000 fics, respectively.
- Every pairing on this list has either Harry or Hermione in it. Also, many of these are with significantly older men, which is, well, interesting. Of course, fics with pairings may not necessarily be romantic in nature, but I’m guessing a fair amount still are.

## Case study: Daphne/Harry vs. Bellatrix/Rodolphus

It may seem like high average values are inflated by a small number of fics with thousands of favorites, and it is true that these distributions are highly skewed, e.g. most fics have only a few favorites. To look at this most closely, below I’m showing the distribution of favorites for two pairings with very similar fic counts: Harry/Daphne (#72) and Bellatrix/Rodolphus (#71). Besides their similar fic counts, I picked Bellatrix/Rodolphus because its average is quite low: ~10 favorites per fic.

This plot is a cumulative distribution plot, i.e. for a given x-value (number of favorites), the corresponding y-value is the fraction of fics with that number of favorites or less. So for a value of 10 favorites, Daphne/Harry has a value of 0.04, meaning 4% of Daphne/Harry fics have 10 or fewer favorites. The value for Bellatrix/Rodolphus is 0.8, so a full 80% of fics have 10 or fewer favorites. Also, note the x-axis has a log scale, so the values range from 1 to 10,000.

You can see the distribution of favorites for Daphne/Harry is shifted to the right (i.e. it has more favorites). Around 99% of Bellatrix/Rodolphus fics have <100 favorites, whereas only 40% do for Daphne/Harry. In fact, ~20% of Daphne/Harry fics (~170 fics) have more favorites than highest-rated fic for Bellatrix/Rodolphus (634 favorites). So there are many highly-rated fics for Daphne/Harry that help to increase its average.

## Why are certain pairings so highly rated?

There are many valid answers for why one pairing might receive more favorites/follows/reviews than another. There’s always the obvious one: readers probably enjoy some pairings more than others, at least enough to favorite those fics more often (I certainly do). But there are two other aspects I want to explore: **length** and **timing.**

### Possibility 1: Length

This one is fairly self-explanatory. Across all fics, there is a significant correlation between length (in words) and the number of favorites. This also holds true when considering the average values for these top 100 pairings. So below I’m showing the top 20 pairings by average fic length. There is significant overlap between this list and the one sorted by most favorites.

As a sidenote, I think it’s interesting to point out that half of these pairings are OC pairings (original characters). I wonder if OC fics tend to be longer because they have to include more character development instead of starting with established characters.

### Possibility 2: Timing

In my last post, I showed that the popularity of pairings can change over time, sometimes to the point where previously popular pairings seem to disappear. Conversely, there is something interesting when you look at pairings that have been popular only recently. Below I have 20 top pairings ranked by what I call date “center-of mass”, i.e. it’s a weighted sum of the publication times for each pairing. The higher the “center-of-mass”, the more recently popular that pairing has been.

One thing to notice here are all the next-generation Harry Potter characters, which makes sense as these characters were only introduced in the epilogue of the final book. However, there are several pairings on both this list and the highest-rated list, namely the top two of Daphne/Harry and Fleur/Harry.

## Case study: Daphne/Harry vs. Cho/Harry

To see these timing differences more clearly, I have two pairings compared below with similar numbers of total fics but very different trends over time: Daphne/Harry (#72) and Cho/Harry (#73). This graph shows the number of fics published each year, so you can see how the popularity has changed over time.

The Cho/Harry pairing peaked in popularity in 2003, likely due to the release of Harry Potter and the Order of the Phoenix, which was the first (and last) book which had this pair together romantically. On the other hand, Daphne/Harry has only become popular very recently, leading to its high “center-of-mass” value.

It’s possible that there is some degree of favorite “inflation,” i.e. readers are more likely to favorite stories in recent years than in the early 2000s. On the other hand, older fics have had more time to accumulate favorites if we assume the overall rate were constant.

Overall, there isn’t much of a correlation between publication date and number of favorites (especially when considering the full population of fics), but I still found it interesting that the pairing with the most favorites (Daphne/Harry) also tended to have the newest fics.

## Final Thoughts

To briefly summarize, there are a lot of popular pairings in Harry Potter fanfiction, but a popular pairing doesn’t necessarily translate to higher ratings (in terms of favorites/follows/reviews). Two factors that might partially explain higher ratings are length (more likely) and timing (less likely), but they certainly don’t explain everything. The true predictor for an individual fic’s ratings is always going to be how much readers like it, but I hope this discussion and my previous post show that there are always interesting trends to see if you look closely enough. Until next time!