Posts Tagged ‘regression’

How long before the Tories overtake Labour in the polls?

28/10/2013, 07:00:51 AM

by Jonathan Todd

Conservative Home recently highlighted an analysis by YouGov that shows a shortening of Labour’s lead over the Conservatives.

In parallel, the economy has continued its improvement. It grew by 0.8% in the third quarter of this year, building upon 0.7% in the second quarter and 0.4% in the first quarter.

Are the two connected? Logic would suggest so.

The trend identified by YouGov reminded me of one that I have spotted myself in one of the trackers that they run.

Roughly once a week YouGov ask voters whether they think the British economy is doing good, bad or neither. Until 25 July, never more than 10% of the electorate answered good in 2013. Since then, never less than 10% have done.

I resolved to bring some econometrics to Uncut to look more deeply into this.

I put together two time series over 2013: one on the Conservatives lead over Labour, which was my dependent variable, and another on the proportion of the electorate who think the economy is doing well, which was my independent variable. When the dependent was regressed on the independent, the co-efficient on the independent variable was just under 0.6. The p-test indicated that the regression was accurate with more than 99% certainty.

This is telling us that for every 1% increase in the proportion of the electorate that think the economy is doing well, there should be a 0.6% increase in the Conservatives lead over Labour.


Facebook Twitter Digg Delicious StumbleUpon