Moving home is never fun, and that seems to go double in a big city. First, you ask around your friends, and most likely, although none of them are moving, they all know a couple of people who have a room or are looking for someone else to move with. One-by-one you talk to these pre-vetted strangers and quickly discover the vast range of requirements that people have. This time around I saw all the usual suspects: people who want to pay more or less rent; live in a different area; move on a different date; or keep a vegan-only kitchen. But I also saw one novel factor: a whole bunch of people who only wanted to live with girls.

New York is a girl-heavy city (unlike say, Middlesborough, a heavy-girl city,) but it’s also a very liberal place. So, I was intrigued as to this trend, and I wondered if it was simply a statistical blip from my small sample size. I threw together a Perl script and scraped tens of thousands of postings on the rooms & shares section of New York’s Craigslist, which is my the most common method of finding roommates in a big city (Gumtree is an alternative in other countries.) I then fed the results into a number of data-mining suites I’ve built or used as part of my day job and waited for the results.

The first thing to do, was to normalise the data. All capitalisation and punctuation is removed, carefully. It’s not enough to simply convert all hyphens and colons to spaces, for example. The code for marijuana on such sites is ’4-20′ or ’420′ (as in “must be 420 friendly”), so splitting this to ’4 20′ is meaningless, but the oddly popular ‘washing-machine’ should clearly be split into two words. Data from Princeton’s WordNet together with some manual effort took care of this. Contractions such as “you’re” need to be expanded, and there are many other boring details to be taken care of. Fortunately, I already have the means to do most of this automatically.

Even after this, the initial data was, always, useless. First, I looked at popular ngrams – sets of n concurrent words – and found predictably boring results.

Most popular trigrams
21% ‘looking for a’
18% ‘to move in’
14% ‘the room is’
11% ‘if you are’
10% ‘i am a’

But, even these had value. It seems the posters aren’t particularly egotistical, with first-person utterances only slightly more common than those in the second-person. Interesting to note was that the first-person statements were definitive – “I am a” or “we will be” – and the second-person statements were surprisingly imperative — “you are a” or “you must”. Despite this, posters were polite, with ‘please’ occurring in 35% of posts. The most common, non-mundane ngram followed by exclamation marks was “Washer and dryer” or “Washer/dryer”, as in “Washer and dryer in building!!” — a rare and desirable treat in New York.

Requirements similar to ‘you must be…’
1. Respectful
2. Clean
3. A good fit (with current roommates/with us)
4. Financially stable/have a job
5. Female
…
9. Normal

A lexicon unique to the site has also appeared, with the aforementioned “420 friendly” (generally positive, i.e. required) and “bring the party home/back” (generally negative) being rarely frequent. Posters cared far more that a potential roommate was ‘respectful’ than ‘clean’ and were far more likely to identify themselves as ‘white’ than any other ethnicity (and yes, I checked they weren’t simply talking about the colour of the walls). The word ‘lesbian’ was used only twice, compared to hundreds of occurrences of ‘gay female’, a trend of which I was previously unaware, and the word ‘chill’ (laid-back, cool, fun) was very likely to be found within a post littered with errors of grammar and spelling.

Initial exploration over with, I turned my attention to the question that had started this all off: do New Yorkers really find the idea of living with men unbearable? It seems they do. Somewhere between 11-13% of posts explicitly required a female roommate and around 17% preferred one. The only stricter requirement was no pets (around 35%), but this is a tickbox on Craigslist. Only about 10% specified ‘no smoking’, 8% specified ‘no hard drugs’ and 6% ‘no drugs’. Male roommates were required in less than 1% of posts, by comparison.

A ‘female preferred’ example
…NON SMOKER >>>NO DRUGS no exceptions …GAY friendly bc I am Gay . So no closed minds . I am female chef would perfer female bout will consider the right very clean guy.

The above is one of the few examples where any reason whatsoever is given, as for the rest I can simply guess and relate the guesses of some New York friends. One has suggested that there exists a stereotype that gentlemen are ‘obnoxious and slobbish’, which is somewhat corroborated by the above quote. I can’t argue against that one, but we do have redeeming qualities too. Another notes that ‘what if he turns out to be a rapist?’ is, apparently, a question to be asked of a potential new roommate. Whilst Craigslist indeed contains the odd posting warning seekers away from a certain location, including one recent set of posts warning that a man on the Upper East Side tried to rape a girl who came to view an apartment, I feel the average rapist doesn’t sign a 1-year lease with his victims prior to the act.

It isn’t even just girls that only want to live with girls either. One male German banker posts:

I want to live with more girls, my roommates that are leaving are girls and I want to keep the vibe.

Another chap who I’m simply assuming without evidence to be a banker is looking for:

A girl to live in my downtown apartment, very cheap rent, must be willing to be a tease and if we hit it off, maybe more.

Charming, I’m sure.

So, whilst the handful of guys with the ‘females only’ requirement have explained themselves, I’m still at a loss as to the motivations of the girls that specify this. But, as I’m not particularly interested in living with someone who’d make such a distinction, or those who specify ‘straight people only’ (very rare) or ‘I’m not racist but I prefer to live with people from the same background’ (just the one), I can live without knowing.

Next, I intend to continue gathering data from New York Craigslist and, when I have enough, to compare the attitudes in the various boroughs. I also intend on getting data from other cities and seeing how they differ. Are the Californian cities really much more liberal? Do the Aussies care about anything other than the quality of barbecuing facilities on offer?

First though, I need to stop procrastinating and find myself somewhere to live.

Xx