Back from a couple of days in Sydney last week. I reckon this must be the first time I’ve been there and not visited the harbour. As my work and hotel were I’m the same complex I didn’t get out much. I had some time in the evening but I didn’t fancy the Quay at night. I did get to the state library for a bit for some more family history research.

I’m not sure if I like Sydney or not. It seems a lot more down-beat and definitely grottier than Brisbane. Then again, not many cities are pretty at 9 o’clock at night!

Sydney Airtrain was very much appreciated though. City to airport in 15 mins every 10. Why can’t Brisbane’s do that?

Around Australia, each state Department of Transport or equivalent conducts regular household activity and travel surveys. These are massive affairs surveying thousands of households to understand their travel behaviour. In part the huge amounts of data are needed to calibrate the equally massive strategic transport models that (more or less) predict future travel demands.

Mostly, the data from these surveys is kept confidential which means it is not available to outsiders without special permission. An exception to this is the NSW Transport Data Centre which is brilliant in that lots of the data is released, so you can examine behaviour by local government area. (The outputs of the travel models are, of course, terribly commercially sensitive and almost never shared with anyone)

Not long ago I downloaded a batch of their Sydney data and whacked it into Excel to see what I could do with it.

One of the first things I did with it was to take the figure which showed the average mode share for car driver by LGA (1997-2001 average) and did a simple linear regression against the 1996 ‘transit access’ figures that I had previously calculated for my thesis in 2005.

Transit Access (a percentage) is defined as the proportion of an area that is within 800m of a train station or 400m of a bus route that runs at least every 15 minutes during the day and at least every 30 minute at night and on Sundays. (I had previously developed this measure and found that it was about the minimum public transport level needed to generate a reduction in private vehicle VKT).

This time I was more interested in overall number of car trips, because this can be very important in understanding traffic impacts (and hence infrastructure requirements) of a urban development.

The result of this regression is shown below (my apologies if the graph is a bit blurry)

In short, it shows a 70% correlation between increase in ‘full time’ public transport coverage and decrease in car driving. I am so happy with this, I am now using it as a convenient ‘rule of thumb':

Each 1 percentage point coverage in full time public transport generates a 0.25 percentage point drop in car driver use.

It’s not perfect, but it’s a start.