Hi all.

I am trying to find out if it's possible to calculate an r^2 value (% of variation explained) when performing a linear bivariate regression using dummy variables.

Let me provide an example of what I'm working on. I'm trying to find if a correlation exists between the type of house people live in and the amount of electricity they use, so my house types are:

Detached, semi-detached, terraced and flats.

I have monitoring data from several hundred houses and have worked out the average for each house type - let's suppose in terms of average kWh per year we have:

Detached - 4000 kWh/yr

Semi-detached - 3600 kWh/yr

Terraced - 3300 kWh/yr

Flats - 3200 kWh/yr

I've done ANOVA on the groups and found that there is a statistically significant differnce between at least two of them. My issue now is: how do I put these into a single equation that gives me some kind of r^2 value so that I can show how much variability is explained by the house type?

Is this even possible? How do I go about this?

Any advice greatly appreciated!

Thanks in advance

-Rob