How we talk matters part 2

In partnership with transport for London.

At the beginning of 2018, we set ourselves a question - is it better to promote diversity & inclusion as a matter of rights and fairness or performance and productivity? The answer wasn’t so simple. When we asked people what mattered most (n=267), they reported fairness over performance at a ratio of 4:1. But when we randomised them into two groups and asked them to take a small action within the context of a wider questionnaire, this happened…


Weird, right? Well, maybe not so much. It isn’t the first time that what people think has misaligned with how they behave. But surely we should explore that dynamic a little further when it comes to inclusion? After all, there are plenty of lofty words spoken about equality in the workplace. It seems important to know whether they translate to action.

An opportunity arose with our friends at Transport for London. They’ve committed to build a diverse workforce and inclusive culture, and therefore need to maintain accurate data on the demographics of their employees. The task poses a different but not unrelated question to the above - how do you ask people to voluntarily and anonymously submit information about their identity for the greater good?

We wanted to optimise a message asking people to update their details, so we designed an experiment. Two thousand employees received one of five emails. Each was exactly the same except one line that introduced a motive for our request:

  1. Control - Please give us your demographic data…

  2. Organisational Commitment - …Because we’re committed to reflecting the city we serve.

  3. Prosocial - …Because we support the broader goal of a fair and inclusive workforce.

  4. Social Norms - …Because thousands of your colleagues have already declared.

  5. Loss Aversion - …Because we can’t improve our diversity until we understand it.

Once again, the way we talked mattered. The below charts the percentage of people who added at least one new data point in the employee database, broken down by the email they received.


Percentage respondents by email

So far, so interesting. 10% percentage points or a 25% difference in impact between the highest and lowest performing message. The most unexpected insight (if you didn’t read our last research summary) is perhaps the ineffectiveness of the prosocial message. Going back to those earlier findings, this seems to align to how people think but might be inadvertently discouraging them to act. Above all, we were most surprised about the strength of the control message. Sometimes it really is better to say nothing at all.

The size of our audience allowed us to dig into an open question after our last experiment: Did the strength of the message differ depending on the recipients background and identity?

Percentage respondents by demographic

Here are five key insights to get you thinking:

  1. There were statistically significant differences in message effectiveness* within every subgroup except women. In our White British population, the highest performing was nearly twice as effective as the lowest performing message!

  2. Organisational commitment was a top performing message across all population subgroups.

  3. The prosocial message performed consistently poorly. It come last for every subgroup except those from a minority background, where it was third of five.

  4. The social norms message was the highest performing motive for people from a minority background. This represents a 52% increase in impact when compared to the control.

  5. The performance of the control and loss aversion message declined dramatically for people with minority status. The drop was unique to this demographic group and significantly different from all other groups.



What can we learn from this research?

We should be cautious about generalising our findings. Do they mean all communications about inclusion should lead with organisational commitment or social norms, and there isn’t a place for value-based messages? Absolutely not. Do they mean that words matter and can have a fundamental impact on behaviour? Yes, and we’re proud that our research led to better, higher quality data for Transport for London to continue their pioneering work.

It seems obvious, but if you want people to do something you believe is of value to the culture to which they belong, you should think carefully about how you ask them. TfL could have done what the vast majority of organisations do when facing an important behavioural challenge - they could have guessed. Instead, they’ve created an inclusive, effective messaging strategy, designed specifically for the people of their organisation, with a better understanding of how different motives will appeal to different demographic groups. And uniquely, they’ve allowed us to share this research for you to build upon…

Many questions remain, not least emerging from the insights above. It’s also worth remembering our behaviour didn’t require much time or resource. What if we were asking leaders to join a sponsorship programme or trying to impact people’s attitudes? What if there was more time between the message and the behaviour - would some motives be ‘stickier’ than others if the opportunity to act came a week later?

We’ll never know unless other employers join TfL and start putting the impact of their words and actions to the test. That’s the way to design better interventions, contribute to our collective understanding of ‘what works’, and build a platform for learning, discovery and progress….

We’re ready when you are ;)

Thanks to TfL for their commitment and partnership, and for allowing us to share these findings. A particular mention to Diversity & Inclusion Lead, Frances McAndrew whose support, energy and leadership made it all happen. The above represents a light preliminary analysis and precedes a research paper we hope to produce with our academic partners.

If you’re interested in learning more about the application of behavioural science to equality at work, we’re running a 3-day education programme from the 15th-17th October 2019.

Finally, if you enjoyed this article please consider sharing it on your social channels below.

James Elfer