Durably reducing transphobia: A field experiment on door-to-door canvassing

By David Broockman, Joshua Kalla. Summarised by Tom Grand.

Researchers in America (Miami) looked to see if ‘deep-canvassing’ - aka talking to voters for a long period of time (around 10 minutes on average in their case) about an issue - could significantly change their opinions. The canvasser asked a voter their view on transgender rights, then spent time getting to know the voter and asking open questions to get to know their reasons behind their current view. From there the canvasser asked the voter to think about a time they were judged negatively and use this to help them to empathise with transgender people. The research suggests this longer, more open discussion can have significant long lasting effects in changing people’s beliefs. The research also aimed to challenge the idea that deeply ingrained prejudices could only be tackled and overcome with months of intense intervention.

Key Takeaways

  • Deep conversations seemed to have significant effects in changing peoples views, of all political backgrounds.

  • These more positive feelings translated to greater support for legislation protecting transgender people.

  • Deep conversations seem to have long lasting effects and persist against seeing attack ads - surveys were done 3 months later and people’s more positive feelings remained.

  • Prejudices after often thought to be deeply ingrained and hard to shift - but this may not be so true.

Practical Recommendations

  • Potentially consider more ‘deep-canvassing’ - aka talking to voters for longer periods of time about a particular issue.

  • Starting conversations by asking voters their view about an issue and then asking open questions rather than directly trying to persuade instantly, helps in the longer run to change their mind, but also make these changes last.

  • This sort of canvassing could potentially be done well ahead of elections (or outside election time) as the effects seem to be long lasting.

  • If you do some sort of ‘deep’ discussion, its good to provide some structure, some open questions to ask to help guide the conversation.

  • Also this approach would require more training of volunteers.


The programme visited voters in Miami (Dade County), canvassing about the inclusion of transgender people in anti-discrimination laws. This was in response to an ordinance in December 2014 where transpeople were granted more protections and there was fear of a public backlash against it. The canvassers were from SAVE (a South Florida LGBT organisation.)

The canvassers visited homes unannounced. They then used a range of techniques to encourage ‘active processing’ (getting people to engage and think about the issue at the time) These include telling the voter they may need to make a decision bout the issue (aka voting on a law repealing the decision), ask them their opinions on transgender people being included in nondiscrimination laws and also defined what transgender was. They then tried to encourage “analogic perspective-taking” - in this case asking them to think of a time they were negatively judged as different and then getting them to empathise with a trans individual. Conversations lasted on average 10 minutes. 1825 people were included in the student, and they were split into two groups - half were talked to about recycling half about transgender people’s rights. Follow up surveys were done 3 days, 3 weeks and 3 months after the intervention.

It found these deep conversations left people feeling much more positively about transgender people (10 ‘feeling points’ on a. Thermometer from 0 - 100) higher than the control group. This, according to the researchers is ‘larger than the average increase in positive affect toward gay men and lesbians among Americans between 1998 and 2012.’ So its potentially quite a notable change. Its effect appears to be non-partisan too - it worked for both Republicans and Democrats, and both transgender and non-transgender campaigners were effective using this method. They were also later showed an attack ad against transpeople, whilst the attack video did decrease support for nondiscrimination laws to include transgender people, the attack adds effects disappeared over time, but the interventions did not. This was run on 1825 and the random divisions were based on households.