If you’re reading this blog, then you’re probably interested in ‘the truth’ – by which I mean that you’re interested in the way the world really is, rather than pretending it’s the way you want it to be. We tend to howl with rage whenever politicians lie to justify injustices, and there’s been a lot of eloquent evidence-based howls in recent weeks, including from Inequalities contributors Declan, Rob, and Lindsey.
But what if the truth doesn’t matter? What do we do if providing people with information doesn’t change their attitudes? And what if a focus on ‘mythbusting’ is actually unhelpful in persuading people to support progressive policies? These are major questions that I’ve written about previously & I’m going to return to over the year (culminating in a working paper), but for the time being there are two great studies that offer food for thought.
Where information about inequality matters…
Hot of the presses comes an amazing March 2013 NBER report by Kuziemko, Norton, Saez & Stantcheva, based on that wonderful staple of political science research: the survey experiment. They advertised a survey at $1.50 for 15mins work among 2000 Americans who use Amazon Mechanical Turk – like other Americans, except younger, more educated, and with fewer African Americans or Hispanic Americans. Some people were just asked the questions, but others were given information on inequality, in two parts (the exact information varied across the six waves):
- On inequality: this typically included an interactive graphic on the current income distribution and what they ‘would have made’ last year if economic growth had been equally shared since 1980.
- On policy: this included one figure that showed that growth in the US was higher during periods of high tax rates than low tax rates, and one slide emphasising that the estate tax only affects the largest 0.1% of estates.
Some visuals of this are given in the appendix of the paper – and if you want to see these yourself, you can do! Kuziemko et al have one wave where they present this information separately, which I’ll return to below.
The limited effects of information
So what impacts did this information have on people? Well, one of the strengths of the paper compared to earlier studies by Sides, Cruces et al and Kuklinski et al is the sheer number of different outcomes they looked at. Rather than put all of the tables from the paper in this post, I’ve tried to summarise the results from their preferred specification in the chart below (all of the differences are significant at least at the 5% level, except for the food stamps comparison).
Two big conclusions come from this. Firstly, giving people information about inequality can change their reported attitudes – this is true for nearly all of the measures here. Secondly, though, the effects are MUCH stronger on inequality-related perceptions (above the key) than they are on policy measures (below the key). One of Kuziemko et al’s favourite measures is how much the liberal-conservative gap is reduced by presenting people with information. For the view that inequality is very serious, information reduces the gap by 40% – that is, the difference between liberals and conservatives is nearly halved. Yet in terms of the support for a higher millionaire tax, only 11% of the liberal-conservative gap is reduced by information. While I’ve not shown it on the figure because of the different units, the same is also true of people’s preferred tax rate on the rich, which rises when people are given information – but only from 29% to 30%.
The only exception to this is the state tax, shown at the bottom, where information makes a huge difference (mirroring the early work of Sides 2011). This seems to be because there is a widespread false belief that many people are affected by it; when it’s clear that it only applies to the richest tenth of the richest 1% of estates, people are much more in favour of a higher rate.
Underneath the bonnet
Before trying to think about what this means, there’s a couple of other interesting points here beyond these headline results.
Firstly, giving people information about inequality made them more likely to distrust government (!) – 15% of the control group trusted government, vs. 13% of the treatment group. My take on this is that people’s opinions on policy may be somewhat fixed, and they try and figure out the rest of their attitudes so that these support their views on e.g. whether taxes should go up. If they get more concerned about inequality, but still dislike higher taxes, then it makes sense for them to distrust government more to square the circle. I’m obviously not arguing that people do this deliberately, nor that this kind of process is restricted to anti-tax views – it just seems a plausible take on all the psychological work on cognitive dissonance.
Secondly, this information seems to have completely different impacts on different people’s support for inequality-reducing policies. For high income/high education groups, information on inequality raises support for raising food stamps and tax credits and the minimum wage. For low income/low education groups, it doesn’t (see Table 10 of their report). This seems to be because it gets people to think about economic struggles, which can lead to status anxiety in low education/income groups. This is the authors take on it anyway, based on a separate survey experiment where giving people positive vs. negative information on the economy has similarly mixed effects on views of inequality and redistribution.
The implications of survey experiments
Pulling this all together, this seems to tell us that (i) information about the state of inequality and tax matters, but (ii) it matters much more for closely-related perceptions than it does for support for concrete policies – with the notable exception of the US estates tax. Sometimes people will change their views of inequality, but then shuffle around some of their other opinions to justify strongly-held beliefs about policies that they held to start with. So to get people to support a higher tax rate, you have to do more than just ‘give them the facts’. And this ties in very neatly with the study that I’ll blog about next week.
Before that, it’s worth ending with a few caveats – because while I love this kind of thing, it’s unclear quite what it’s telling us:
- Information is not ‘just facts’, but it’s always embedded in particular language and a particular worldview – and in this case, these were made deliberately strong. The authors note, “In general, the goal of the information treatments were to provide a large “shock” to individuals’ information about inequality and redistributive policies, rather than to provide a Ph.D-level, nuanced discussion about, say, the underlying causes of inequality or the trade-off between equality and effiency. As such, some of the treatments we display will seem overly-simplified to an economics audience…”
- Connected to this, you could probably produce a parallel series of ‘facts’ that challenge some false left-wing beliefs, with very different effects. As it happens, I think that people’s false views are generally biased towards the right in the US and UK at the present time. But truth is not inherently liberal, and the opposite has been the case in the past.
- Finally, the extent to which this presents ‘facts’ with no uncertainty and no disagreement is unclear. Some of the information is fairly hard (but not impossible) to contest, but other parts of it are more contentious – in particular the evidence on higher growth when tax rates are higher, which from an outsider’s perspective seems to involve some fierce debates.
These might seem like minor details, but I think that facing these issues head-on is important – so more on this later in the year too.