Skip to content

How funny were the jokes?

October 2, 2018

I ran a survey recently asking people to rate some jokes so that I could use the results with my Core Maths classes. Several people said they were interested in the results and I thought I’d share them along with what I plan to do with them here.

The jokes

There were 10 of them and they’re presented here in no particular order:

  • I can’t stand Russian dolls. They’re full of themselves.
  • When I worked as a librarian, if anyone ever asked where the books on paranoia were, I’d always whisper, “They’re behind you.”
  • Someone complimented me on my driving the other day. I got back to the car and there was a little sign saying “Parking: Fine”, which was nice of them.
  • I used to train racing snails. One day, I took the shells off to see if they’d go faster. It didn’t really work and, if anything, it made them more sluggish.
  • I had a job drilling holes for water – it was well boring.
  • Cheer leading exams are easy. You go in and shout “Give me an A”.
  • I got really emotional this morning at the petrol station. I don’t know why, I just started filling up.
  • I’ve started a business selling ejector seats to holy people. Prophets are through the roof!
  • Thanks for explaining the word ‘many’ to me. It means a lot.
  • I’ve always wanted a job putting up mirrors. It’s something I can really see myself doing.

Assuming you’re still with me and not uncontrollably laughing, here’s a bit more detail about my survey design.

Survey design and distribution

  1. I deliberately made it an anonymous survey and didn’t ask for any details other than “How funny are these?” I think this probably means that people are more happy to answer but it does mean I can’t do as much about drilling into the data.
  2. A scale of 0 to 4 allows a sensible “This isn’t funny = 0” rating and also allows enough scope for differentiating between jokes without forcing people to decide “Is this an 8 or a 7?”
  3. Each person that answered the survey had the jokes presented in a random order to avoid the potential issue of skewing results. For example if a mediocre joke followed a poor joke, then the mediocre one may get an unfairly high score.
  4. I shared my survey mostly via Twitter so it has been answered by the kind of people that follow me (and the people that follow them).
  5. I asked people to retweet the link just to get more responses.

The results

There were 585 responses and you can see a spreadsheet of the raw results here. (You’re welcome to use them however you like.)

What am I going to do with them?


One main reason for doing this was to be able to look into correlation. For that, I’m going to do a statistically very dodgy move of finding the mean average rating for each joke. This is something you shouldn’t do with Likert scale data as the scale isn’t linear (ie 4 isn’t twice as funny as 2). However, I do think it’ll give a sense of which jokes were rated more highly and I’ll have to hope the Stats Gods will let me off. I promise I’ll discuss this in class.

I’ve also asked each of my classes to rate the same jokes and I’ll get them to see if there’s a correlation between the rating they gave and my followers. I’ll also get them to see if there’s a better correlation between their average ratings and those of the other class.

I did consider giving them one fewer joke to rate and use regression to predict their rating. However, I suspect the correlation will be weak and therefore the regression will be a poor predictor so I’m a little wary of showing them a scenario where regression doesn’t seem to work.

Averages and Spreads

I’m also going to use the data later in the course as the basis for a discussion about the types of averages and pros/cons of each. Modal response is probably not very helpful as they’re mostly just ‘3’ (on Twitter responses at least) and Median has a similar problem.

I will also (committing the same Likert scale crime) use the data to find the standard deviation of each joke to find the most ‘marmite’ one. Those with lower standard deviations may well have been more consistent in their ratings while those with a larger sd may have been more polarising.

Here are the mean average results in case you’re interested:

Mean Twitter Class C Class M
Russian Dolls 2.356 2.333 1.778
Paranoia 2.471 1.889 1.222
Park Fine 2.132 2.111 2.333
Racing Snails 2.535 2.111 1.556
Well Boring 2.047 1.111 1.333
Cheer Leading 1.941 2.222 1.556
Filling up 1.925 1.333 1
Ejector Seats 2.502 2.111 2.444
Means a Lot 2.55 1.556 2.333
Mirrors 2.341 1.333 1.667

These make some fairly nice bar charts that are interesting to compare and explore/discuss. I might see if there’s anything dodgy I can do to make my favourite joke/s appear more popular than they actually were and let the students play detectives with my misleading graphs!

Using computers

I may also use these as a ‘large data set’ for students to play with on computers. That’s only just occurred to me so I can’t say I’ve given it much thought but I’m sure there’s something there! Will probably be worth looking into if there are any weird results or any cases of people just rating some of the jokes and how we might deal with that.

Comparing to last time

I have done this experiment before (you can see the results here). It wasn’t all the same jokes although you’ll see that there are some repeats. There’s likely to be something we could do along the lines of “Did the jokes that featured both times seem to rank in approximately the same place each time?”

Thank you

Finally, thanks to everyone that read the jokes and rated them. Thanks also to everyone that retweeted the link so it could be seen by a wider audience. If you do something interesting with this data, do let me know!

The ‘actual’ results

Given all of the Likert caveats I mentioned above, here’s my tentative suggestion for top three performing jokes:

  1. Thanks for explaining the word ‘many’ to me. It means a lot.
  2. I used to train racing snails. One day, I took the shells off to see if they’d go faster. It didn’t really work and, if anything, it made them more sluggish.
  3. I’ve started a business selling ejector seats to holy people. Prophets are through the roof!

Their bar charts are below and their mean averages are above. See if you agree.

ejector seats

Ejector seats

Means a lot

It means ‘a lot’

racing snails

Racing Snails


Leave a Comment

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: