How Different Are Telephoning and Canvassing? Results from a 'Get out the Vote' Field Experiment in the British 2005 General Election

By Peter John, Tessa Brennan. Summarised by Jenny Brennan.

Researchers wanted to compare the effectiveness of non-partisan 'get out the vote' (GOTV) canvassing and telephoning on voter turnout in Britain. They took a similar approach to GOTV researchers in the US, to see whether learnings from the US might be generalisable to outside of the US. They found both interventions had positive, strong and statistically significant effects on voter turnout. Canvassing and telephoning were similarly effective. The authors concluded GOTV tactics involving speaking to people can be effective in Britain, and suggests US findings may generalise, at least to a British context.

Key Takeaways

  • There are some concerns in study design here; if you're interested in applying this research, we'd recommend you read the paper in full (it's only 10 pages!)
  • Estimated effect of personal canvassing on voter turnout = turnout diferential(3.6%)/contact rate(53.7%) = 6.7%
  • Estimated effect of telephone contact on voter turnout = turnout differential(3.5%)/contact rate(47.7%) = 7.3%
  • Both interventions had positive, strong and statistically significant impacts. There is a large standard error, because of relatively small sample sizes.
  • Concludes GOTV campaigns using personalised techniques are effective in Britain, and suggests US findings may generalise, at least to a British context. Calls for more research.

Practical Recommendations

  • Recognise that US research may be applicable to a UK context
  • Consider prioritise personalised means of GOTV campaigning, such as canvassing or telephone calls
  • Consider the quality of personal interactions


Study conducted in Wythenshawe and Sale East in Manchester, which had a turnout in the 2001 general election of 48.6% (versus national average of 59.4%) and a safe majority for sitting Labour MP. 9,976 registered voters for whom they could obtain landline telephone numbers were randomly selected into three groups.

One treatment group got a telephone call (telephone group), one got a visit (canvassing group), and the control group received no contact. Everyone in the treatment groups (telephone and canvassing) received a neutral letter warning them they'd be contacted, branded as a non-partisan university 'Get Out the Vote Campaign'.

There are some methodological concerns here: for instance, the letter sending was assumed not to have an effect itself, based on US studies showing letter-based campaigns to be ineffective. However, it included a link to a website which listed reasons to vote and further resources, and the authors highlighted they were uncertain as to whether US evidence was applicable in Britain.