Wesley Crushes Ratings
April 19, 2016
Dear Wesley Crusher,
You are, by far, one of the most annoying characters in the entire Star Trek Universe. Ok, sure, you have a few close runner-ups (Neelix, Lwaxana, Keiko, Alexander) but I’d still say you take the prize. You are an obnoxious know-it-all kid who had a tendency to screw things up, whether it be by appointing yourself “acting captain” and taking over engineering, letting nanites loose to destroy the ship, lying at Starfleet hearings, or getting a death sentence on a foreign planet. We’re talking second only to Star Wars’ Jar Jar Binks level of annoyance.
I recently found myself wondering if I could measure how annoying you are; how could I test, with numbers, that my dislike towards you is not felt alone, but shared by many? And beyond that, a collective dislike so powerful, that solely your existence in an episode would bring ratings crashing to the ground like the Enterprise in Nemesis.
How’d that turn out, you ask?
Ratings with and without you, Wesley Crusher
To begin, I gathered every line from every episode of Next Generation from Chrissie’s Transcripts Site, along with the corresponding IMDB ratings.
There are 176 Next Generation episodes which have a median IMDB rating of 7.4. If one separates the episodes that contain absolutely no Wesley from the episodes that contain at least one line of Wesley, a statistically significant difference is immediately noticeable: the no-Wesley median is 7.5 compared to a with-Wesley median of 6.95.
In order to determine the statistical significance of the difference in medians between the no-Wesley and with-Wesley episodes, I performed the bootstrap. Might sound like a dance, but it just means that I created a simulation that went through the following steps a million times:
- Randomly sampled values, with replacement, from all of the episode ratings to create two groups: one of size 108 (the amount of episodes without Wesley) and another of size 68 (the amount of episodes containing Wesley).
- Calculated the median of each group
- Calculated the difference between the two medians
Since I sampling ratings from all episodes regardless of whether Wesley talks, I know that there is actually no difference between our two sampled groups, so it’s a good way of measuring the variability in the difference in medians just due to to “chance”. The distribution of the one million median differences returned by the simulation can be seen in the histogram below.
We can see that the probability of the difference in medians being at least 0.53 just by chance is low. In fact, the chance that would happens falls under one percent.
But there’s more. Take a look at this:
This chart is showing how the median rating drops as the number of lines Wesley Crusher has in the episode increases. Sadly for you, Wesley, the more you talk, the less people like the episode.
So, is the continuing drop in ratings as you utter more lines also statistically significant?
To test this, I did another bootstrap, this time randomly sampling values from the Wesley-included ratings list to create groups of 68 (number of episodes where Wesley speaks at least once), 39 (number of episodes where Wesley speaks over ten times), 18 (over twenty times), and 13 (over thirty times). For each iteration, I found the best fit line for those values, and it’s associated slope. After repeating each iteration a million times, the histogram of slopes looked like this:
About 9% of these slopes are at or below the observed slope. Basically, when we assume there is no relationship between the number of lines you speak and the ratings, the slope would look like this or worse about 9% of the time. So, it’s not statistically significant at typical 5% level but, let’s be serious here, you still probably suck.
Wanna see some of the best fit lines generated by the bootstrap?? Of course you do!
Would you expect the actual best-fit line (the red one) to look like it does if Wesley Crusher didn’t suck (the blue ones)? Probably not.
Now, no matter how annoying I find you, Wesley, I feel an obligation to be fair to you. After all, you did save Picard from dying of thirst, save the crew from mind control devices, and you were the first to recognize that Lore was really Data’s evil brother and not Data himself. As much as I wanted this relationship between your words and ratings to be perfect, there are a few episodes where your presence didn’t drag down the rating. The episode with the most Wesley lines (“The Game”, with 119 lines spoken by you) has a respectable rating of 7.4. But, let’s be serious, how awesome would that episode have been without you?
Either way, in the end, I (and the data) side with Lore when he called you a “troublesome little man-child”. I have but one last thing I’d like to say to you, echoing the words of Captain Jean-Luc Picard and your mother (Beverly Crusher) in Datalore…
Very Truly Yours,
P.S. Just to be clear — this is not about you, Wil Wheaton. You seem like a pretty cool dude with whom it would be fun to grab a beer. It’s not your fault the writers often wrote your character to be so insufferable. Also, you’re fabulous in Big Bang Theory. <3
To the Star Trek fans,
Thanks for reading! Would you like more? I have compiled every line from nearly every episode and movie in the Star Trek Universe (thanks to Chrissie’s Transcripts Site) into a format that makes it easy to find insights within the data and do statistical analysis. There is a lot of data here for me to play around with. I want to know if there’s anything YOU are curious to find out. If you have an idea, please fill out this form and tell us what you’d love to know!
Additionally, I would greatly appreciate anyone taking the time to fill out this Best Trek survey about your favorite Star Trek movies, series, and characters. I’m curious to analyze the results!
- I’m not claiming this data implies a causal relationship between Wesley talking and the rating of the episode: common sense, however, does. ↩