Post date: Oct 19, 2014 2:57:45 PM

The title is not a typo, here are some stats that are interesting to the singles out there. I do narrow down to stats about singles eventually though, so the last couple of graphs are Sydney's Singles' Sydney Singles Stats.

Though not a typo, the title is a lie. I have results for all of Australia's capital cities, not just Sydney.

What and why?

Someone told me that Sydney has more single women than men (in a "there's more fish in the sea" context). I thought that the claim itself was extraordinary - how could a city of this size have a meaningful gender imbalance?

I've just completed a couple of courses in data wrangling and "Exploratory Data Analysis" (otherwise known as "making graphs, and then looking at them") so I made some graphs and then I looked at them.

Some of the results are really interesting, I put them here so you can look at them.

Are there more single women than men?

Wild claim, 0, Skepticism, 0.

Census Data

Most of the raw data from Australia's 2011 census is available online for free^{Note 1}. That's an amazing resource.

It seems like a daft question, but what/where is Sydney exactly? There are a bunch of different classifications and definitions of where Sydney stops, I settled on a definition called "UCL: Urban Centre/Locality. It's smaller than the default option that the census website shows you (SUA: "Significant Urban Area"), I wanted to target the working heart of the city. To see how different things get how quickly, the UCL has 3.9 million, the SUA is about twice the size but only adds 0.1 million.

The census data website shows you a lot of prepared information for different regions. Here are the Urban Centre/Locality pages for Sydney, Melbourne, Brisbane, Adelaide, Perth, Hobart, Darwin, and Canberra. Here are the maps for the Urban Centre/Localities of Australia's capital cities:

Making graphs, and then looking at them: The General Population

First-pass at the wild claim: What does the overall population look like?

The chart below is called a "Population Pyramid", but usually those are shown with the oldest at the top and you count backwards as you read down, and that's stupid. So all of my graphs are going to be this way up.

Overall, Australia seems to have slightly more boys than girls, all the way up to the mid 20s, and generally speaking more women than men from ~30 to ~60. Some googling confirms this. Worldwide, slightly more boys are born than girls, and after that men move to work overseas and in rural areas (to be honest I don't trust that second source much, but I didn't find any better).

Look at how the plot for Sydney (the business-driven city where I live) bulges out for the working population. Compare it to how uniform the plot for Hobart is (the high-quality-of-life but low-employment city I'm from).

About one in four Australians were not born in Australia! I already knew that but I'd forgotten how it would affect the graphs. I was wondering why we had some kind of strange population drop for kids age 8-18, where the plots narrow in, but I had it wrong; actually what we really have is a huge migrant population, who tend to be ~20-60 (obviously, if you think about it), which make the middle part bulge out.

At 70+, the trend everywhere is simple: Men die earlier.

(Darwin is utterly weird at all ages)

I'm... older than I want to be, but say my dating range is a ten-year slice, of ages 26 to 36 (it would be way more if I used the standard creepiness rule^{Note 3} . It's a good example range because any trend that appears will be similar for any age bracket up to ~60. So in my range:

Making graphs, and then looking at them: Relationships

The census data gives you relationship information. I had a devil of a time making sure I had it right, though. I found the problems when I tried to get the de-facto data. It's interesting, really, but it's not part of the main story, so I moved it to the appendix.

The rest of this page is about ages 18-80, by the way.

Who Are They Married To?

Take a look at this plot for Sydney (I'll get to the other cities soon). Look at the Available column, completely dominated by men in my age group! The correct statistical term for this is a "sausage fest".

Wild claim, 1, Skepticism, 1.

There's a reason I didn't stop there though - take a look at the Married column.
Wow, that's weird, isn't it? How can there be so many more married women than men, age ~20-40? Who are they married to?

Got any theories?
What if I tell you that it's the same across all the cities. Any idea now?

It had me stumped for a few days.

I was showing that graph to friends, as a curiosity and a puzzle, asking if anyone had any good ideas. A few people mentioned that perhaps the graph was skewed because typically men are a little older than their wives / women are a little younger than their husbands (I've been told it's because women mature faster!).

But even if that's true, it's hard to imagine it skewing the plots so much. What was causing it?

One friend suggested that I make some graphs, and then look at them. Good idea. See what showed up:

Well that's amazingly convincing.

That, the perfect alignment at a two-year shift, was the single most surprising thing for me. It changes everything about just matching ages (or groups of ages) and talking about relationship statistics. It's even true for for the de facto stats. While the general population stats are relatively even, once people start pairing-off things look really different.

Next time you hear any reports talking about where the singles are, you've got some clever questions to ask!

Lets take a look at the relationship breakdowns for all the cities. I'm showing both the actual demographics, and the shifted versions.

Well, if you correct for different ages, things even-out between genders, and there's no excess of men after all.

Wild claim, 1, Skepticism, 1 ^O.

Whoops! "No extra men" is not good enough, I forgot to check if there were more women. Nope.

Wild claim, 1, Skepticism, 1 ^Ø 1.

By the way, above, I've plotted a breakdown of relationship trends in all the major cities in Australia, and there are heaps of differences between cities! All sorts of things to look at in there, if you take the time to. I noticed:

• Darwin... don't get me started on Darwin, the stats are weird.

• Compare Brisbane, Perth, and Canberra, where de facto is common, to Sydney especially (but Melbourne too), where de facto is relatively uncommon.

• De facto-wise, the quickest uptake of de facto status and marriage in all of Australia is Darwin, and the greatest relative proportion of de facto relationships, which is nice, very progressive. Darwin is the de facto capital of Oz. But of course it's the worst place to be, proportionally, if you're a straight single guy.

• My sleepy wonderful home-town of Hobart, has propotionally the most singles after 40 (it really is the best place to retire!), and the slowest rate of marriage in all of Australia.

• Older-singles-wise, the (proportionally) fewest older singes are in Darwin. "Darwin has a youthful population with an average age of 33 years (compared to the national average of around 37 years) assisted to a large extent by the military presence and the fact that many people opt to retire elsewhere." - so says Wikipedia. Burn.

• Bloody Darwin... It doesn't even deserve to be called Darwin! Charles Darwin was never even there, he spent time in my towns of Hobart and Sydney, not in Darwin. Darwin wasn't even called Darwin originally! (The town I mean). It was Palmerston.
  ...Okay, okay, I'm way off-topic now.

• One of the most interesting consistent trends, is that in the married column, there are more older men around. It's interesting because women live longer generally - look at the older single women. Married men fare better, because being married is good for men's health and lifespans!

Making graphs, and then looking at them: Singles

The article about married men had a note about education, which lets me segue into my last major point: Who would I date?

I am fortunate to be educated. I work in research. I am surrounded by the culture that comes with that. I like smart people - which is hard to measure so I can't plot it - but I also think I need someone who has spent time at a university, and has been exposed to that culture, because it's such a big part of who I am.

When you start breaking things down by relationship and education, weird trends pop out. I'm not going to side-track into the full story for all states, but I'll just leave these two plots of Sydney here, if you're interested.

I looked at the differences between cities in these graphs. Firstly, I noticed that the general shapes and proportions were similar across all of Australia (there aren't really more-educated or less-educated cities). Secondly, I noticed that looking at the charts and trying to make meaningful judgements gets really hard when you're looking at divided portions of a population that already varies in shape.

So take a look above for yourself, if you want, but here is one important thing that's true for all of Australia:

There are more women at universities than men!

Here's a nice write-up about it, with a couple more graphs.

That's good news for the wild claim! So here's the last multi-city graph: What's the age breakdown of single (non-married, non-defacto) people, who have a bachelor's degree or higher?

Constant bounds again, so changes in shape indicate relative proportions of the available, university-educated, population

More educated women than men, in every city!

Wild claim, 2, Skepticism, 1.

Even, (grudgingly), Darwin is okay. No, look, I'm not being sarcastic. Look at the graph and the scale and you'll see, in my age group the imbalance itself is several tens of women.

Age 22-26 there's a huge difference between men and women, and again in the mid-30s and onwards. Right in my age-group somehow, the trend shrinks and nearly disappears. But to finish the story, let's zoom-in closer to where I originally said I was looking:

There are a lot of educated singles in Sydney, whether there are more of one gender or the other, I'm in good company. Within that 26-36 gap:

• • 44,125 Men,

  • 47,836 Women.

Kewl.

So the wild claim was right.

The connotation was that competition might make one gender more desperate than the other, and I'm not so sure that's true (except in Darwin). But that's a technicality which misses how the claim was right in its central message: If you're looking, then no matter who you're looking for, there are a lot of potentially interesting people that you can meet, anywhere in Australia. The right person for you could even be in Darwin. Yes, really. Because after looking at all the many hundreds of thousands of people across the entire country, remember:

You only need to find one.

Appendix: Dealing with Dodgy Data (or: Why Relationship Stats Are Crap)

First up: None of the census data or graphs say anything about how many "singles" are actually in stable (but not-yet de-facto) relationships.

Second-up: The census had two separate questions dealing with relationships, and the answers don't all align, which is a pain.

Here are the two questions:

And here is a table of how those panned-out for Sydney.

The ABS (Australian Bureau of Statistics) used the Q5 data to calculate thata that they call "MDCP: Social Marital Status" (click the link for more information about it). That's what the rows are.

The data looks bad, right?

• 63,594 people answered "Separated but not divorced" but got classified as "Not married"?

• 38,474 are just plain "Married" and "Not married".

• 2,985 are "Married in a registered marriage" but "Never married"

But numbers like this are deceiving - the tens of thousands of dodgy-looking results need to weigh up against the millions of people in this dataset (2.9 million people here, for Sydney, aged 18-80). All real-world data has a bunch of noise, this is actually okay data. Here, I'll show you, converting the numbers to percentages

• 48% - Yep, makes sense

• 24% - Never entered a marriage, I'm surprised it's not more people actually. Australia is still pretty conservative about getting married.

• 6% - Divorced and not remarried, and 5% - never married but de-facto instead, reasonable data

• The "Not Applicable" row is defined here, I think it's mostly people in a house not theirs (travelling) - 8%.

• These make sense too - and we're getting down to just 4%

• Only 3% of the data is dodgy,

• ...and the rest is negligible - all rounds down to 0%.

And actually, it gets better than that!

The data for "Separated" is the biggest source of suspicious data here,

but,

the category of "Separated but not divorced" is neither "Married"/"De facto", nor "Single", so it doesn't belong in our graphs. The whole Separated column can just go. That only leaves 1% suspicious data. For all the graphs above, you wouldn't notice it even if every single one of the 1% got classified wrong (and of course, about half of them will be correct anyway by chance, no matter which way you class them).

So what I did was: