In this post Dr Caryn Peiffer discusses her research on the impact of anti-corruption awareness raising programs. Findings, she argues, defy intuitive assumptions.

Corruption is the talk of the town. It is a popular news item and billboards, posters and murals, common across the global south, broadcast the ills rooted in corruption. They often tell commuters which hotline to call if they’re asked for a bribe, or remind them of the effects of corruption on their community. The idea – and it seems logical enough – is that if we understand the damage corruption does, and if we know who to report it to, then we’ll all get involved in tackling it. We’ll defeat corruption together.

Yet findings so far from my research examining this intuitive assumption are not confirming it. Anti-corruption messaging doesn’t seem to work like that. What is more likely is that a lot of anticorruption messaging may be ineffective, and some of it may even be making matters worse.

Starting to fill the evidence gap

For some time now, I have been looking at what the data can tell us about corruption – and how we might come up with some effective solutions. Back in 2012, one of my early studies for the Developmental Leadership Program looked at how a deeper understanding of the politics of corruption reduction might be approached by identifying successful efforts to reduce it and then working out what had gone right.

But the corollary of this was to ask why what was already being done to reduce corruption wasn’t working. Anti-corruption messaging was an obvious focus. There was so much of it, and yet no clear research on it to show that any of it was working. In 2012 Johnson, Taxell and Zaum published their exhaustive review of research on the effectiveness of anti-corruption programming; they found not a single study that focused on the effectiveness of awareness-raising campaigns.

In 2015, in Jakarta, and last March in Papua New Guinea (PNG) with Dr Grant Walton, I carried out field survey-experiments designed to detect changes in attitudes towards corruption among citizens who had been exposed to anti-corruption messages. And, if all goes to plan, next year, with Professor Nic Cheeseman of the University of Birmingham’s International Development Department, I will be putting a third experiment into the field in Lagos.  At this early stage in analysing the PNG data, it’s not possible to give chapter and verse on what the results say. However, I presented early findings from Jakarta at December’s International Anti-Corruption Conference, and we now have the full picture.  As this new paper explains, it has some potentially strong recommendations for anti-corruption messengers.

A cause for concern

The findings of the Jakarta experiment warn that, if you talk to people about corruption – even to set out straightforward ways they can help fight it – they can become more fearful of it. They seem to feel overwhelmed. Their confidence in the idea that they can do something about it can diminish. This, especially, makes them likely to back off from actively getting involved, as citizens, in tackling it. Respondents exposed to anti-corruption messages tell us that they are less willing to report an official who asked them for a bribe.

In short, instead of smoothing the way towards a lasting solution, the message can become another hump in the anti-corruption road.

Are some types of messages less alarming than others? Not necessarily

So drawing on the databases that I’d used to identify ‘islands of integrity’, I devised an original survey-experiment that made it possible for the data collected to be analysed using a structural equation model (SEM). This statistical technique, a combination of factor analysis and multiple regression analysis, makes it possible to analyse structural relationships between both measured variables – for instance being more, or less, likely to report corruption – and latent constructs, such as attitudes towards corruption, after exposure to an anti-corruption message.

Respondents from 1,000 households took part in the Jakarta survey. They were split into five groups; one – the control group – was simply asked to answer the survey questions on how they felt about certain aspects of corruption. The remaining four groups answered the questions after reading one of four messages about corruption. Two of these messages were ‘negative’ – they emphasised corruption as being a huge problem – and two were ‘positive’ – one on how citizens can easily join the fight against corruption, and the other on how the government is making headway in its efforts to reduce it.

The early findings showed that exposure to all the messages, negative or positive, tended to make people more worried about corruption’s consequences, less confident in their government’s efforts to tackle it and, perhaps most importantly, less confident in their own ability to do much to fight corruption. The latter finding is particularly striking, given that one of the messages was designed to achieve the opposite.

An increased will to fight corruption coupled with decreased confidence in success

The SEM analysis, detailed in full in the second paper from this research, focused on what all this meant for citizens’ willingness to get involved in the fight against corruption. It revealed that messages were generally not that influential, but not because they were ignored. Instead, by making people more worried about corruption’s consequences, the messages prompted people to want to fight it. Yet, at the same time the messages undercut people’s confidence that they could do so effectively. In other words, the effects sometimes cancelled each other out. If found to be influential, the analysis detected that the messages tended to ‘backfire’—exposure to a message put the respondents off engaging in anti-corruption efforts. This ‘backfiring’, the analysis showed, was mostly because, regardless of tone or content, the messages had reduced citizens’ confidence in their ability to fight corruption.

What might this mean for anti-corruption messaging programs?

The implications of this for anti-corruption work could be worrying, and we clearly need more evidence. One practical way forward while further research is being carried out could be to use more pilot studies before rolling out anti-corruption awareness-raising campaigns to check likely effects in the particular context. The main takeaway, however, is simply this: before more resources are poured into anti-corruption messaging, we need to actually test what effect those messages might have.