Who tweets about xkcd? | Hacker News Reader

Who tweets about xkcd?

thetwoweeker.com

46 pointsthetwoweeker24 comments

Open articleView on HN

That looks like data about

People who read xkcd AND have twitter AND feel compelled to tweet about xkcd.

I suspect that's a small subset of the total people that read xkcd.

Maybe the people at "thetwoweeker" only spent two weeks learning logic and statistics.

you give us too much credit...finals week at our school was only 1 week long

Probably.

Recently I enjoy Abstroose Goose http://abstrusegoose.com/ a bit more than the recent XKCD strips. If you enjoy XKCD it's probably worth for you to check Abstroose Goose out.

Seconded, although they're rarely as concise as xkcd.

Assumptions in this post:

>Who reads xkcd?... In the process, we've scraped 20,000+ xkcd tweets

>Although Twitter doesn't contain gender data for tweeters, we were able to identify the gender of each user by their first name.

>We again applied fuzzy string matching to match a reference set of time zones/countries/continents to each tweeter's self-reported time zone.

>We had to make one critical assumption here - namely, that people will tweet shortly after reading.

>We measured this on Twitter by defining: Popularity = # of Followers Likability Ratio = # Followers / # Following

The likability one takes the cake though. Celebrities would be popular and likable. So would people who bought fake followers. People who like to follow a lot would be unlikeable.

Phew looks like we generated some statistical controversy here....so re: the complaints that our sample doesn't actually capture "who reads xkcd" but rather "who tweets xkcd" - we hear you. Unfortunately the crux of the problem is that there doesn't appear to be an easy way to draw a random sample from the xkcd population, unless you are Randall Munroe or have access to xkcd traffic logs. Any API you use will probably lead to a biased sample, and given that there's no physical location it's not like we can check who checks into xkcd on foursquare either. But that being said, if anyone knows of a better way to sample "who reads xkcd", please hit is up!

I expected to see XKCD-style graph, like this: http://mathematica.stackexchange.com/questions/11350/xkcd-st...

One of the problems with assuming male/female within datasets is that typically more females keep their gender information confidential, which ends up skewing the dataset significantly.

Also it would have been interesting to compare some of these charts to the overall sample set of Twitter, considering that it also skews more towards "geek", "student", "developer" as well.

I'm surprised more people don't find xkcd author's obsession with his ex more disturbing, especially when he makes passive aggressive comments about it in his comics.

I also personally don't find the formula of 'ha ha some science reference I know of!' or graphs in comic form to be humorous.

At first I missed the links "Nerd talk" (as I usually don't click on -potentially external- links). They add quite interesting insights!

>>They are fathers, husbands, and software engineers. They love, and they like. But above all, they read xkcd.

What about female followers?

the "Are they male or female" link is wrong

Thanks for the catch, fixed now

The headline and pitch are misleading (clickbait?). This is an article about people who tweet about xkcd. Which is a combination of people who read xkcd and people who think "Oh, look at this cool comic my friend showed me. I should tweet about it!"

It isn't accurate to extrapolate data from such a specific group to the whole.

[Also, privacy badger blocks squarespace on this site, which breaks the styling ]

Edit: Relevant SMBC comic: http://www.smbc-comics.com/index.php?id=3374#comic

I was going to make the same comment. Not everyone who reads xkcd tweets about it, or even uses twitter.

And?

You could say that the people who tweet about xkcd are not a representative sample of the people who read xkcd. However, I don't see this as something absolutely intuitive -- my intuition is that it's a slightly distorted sample, but should not be _that_ off. I don't see why the people who tweet should have different characteristics to the people who read.

The thread is full of comments complaining about why the OP took this particular set -- I don't have too much problem with it. It's a sample, and quite a large one.

> You could say that the people who tweet about xkcd are not a representative sample of the people who read xkcd. However, I don't see this as something absolutely intuitive

That a sample drawn from a group distinct within the target universe by a common factor (as "people who tweet XKCD comics" is within the universe of "people who read XKCD comics") is otherwise representative of the target group is a point which must be justified when using analysis of the sample group to draw conclusions about the target group, not something to be assumed unless there is evidence to reject it.

Because the distinction is, in and of itself, a manner in which the sample group is not representative of the target group.

Thank you. Title changed.

Thanks for changing it (Sorry if I came across as too harsh. )

For anyone trying to make sense of this post in the future, the permalink for this blog post is http://www.thetwoweeker.com/blog/2014/6/30/kkbbwyicx0nwlprkn...

(I actually had to view source to find that - it's not visibly linked anywhere on the http://www.thetwoweeker.com/blog/ page)

(stable links are important, people!)

A good trick to get this information without viewing the source code is to simply hit the tweet button ('Share' on the bottom of the page). It will open a popup window with the static URL of the page as a template tweet. Works on most pages and without a twitter account.

Thank you. Url changed.

Reply on news.ycombinator.com