Broken windows and litter: tidying up INCPEN’s arguments

The Industry Council on Packaging and the Environment (INCPEN) is putting a lot of effort into communicating the idea that a deposit refund system (DRS) on beverage containers would be ineffective in addressing litter. This line has formed a major part of INCPEN’s response to my recent article highlighting how they cherry-picked litter data for a press release, which has been widely discussed on Twitter. INCPEN’s argument, encapsulated in a tweet, is that:

There is no point addressing some of it as litter breeds litter

They expand on the point on their website, explaining that:

There is no point addressing just some items because there is a huge amount of evidence that even one piece of litter begins to make an area look dirty and uncared for which encourages more irresponsible behaviour and more litter (the Broken Windows syndrome).

INCPEN has been vigorously deploying this argument as part of their campaign to put policy makers off the idea of a DRS on drinks containers. They have even amusingly illustrated it in a video, which shows the littering of cigarette butts and used chewing gum (i.e. non-packaging items) instigating a flurry of further littering – initially more butts and gum, with packaging items appearing some time later. However, given the potential for butts and gum to accumulate, and the greater ease with which packaging can be removed by periodic street cleansing, this presentation of the way littering develops, while conveniently shifting responsibility away from cans and bottles, may not be entirely fair.

Out of the window

Wilson and Kelling’s broken windows theory has been influencing crime policy since the early 1980s, although there is some contention over just how far examples of reductions in crime attributed to the scheme may be accounted for by other factors. Findings from the academic literature also lend support to the intuitive notion that people are more likely to litter in spaces that are already littered. Research undertaken by Cialdini et al. as far back as 1990 found that only 11% of individuals observed littered in a clean environment, while 41% littered in a littered environment. So does this mean that INCPEN’s claim is right?

That rather depends on what we interpret INCPEN’s claim that there is ‘no point addressing just some items’ to mean? Does it mean:

• It will result in no net reduction in litter; or
• It will result in only a very small reduction in litter; or
• It will not entirely eliminate litter?

No-one expects a DRS to wholly eliminate litter, and this would be an unreasonable standard against which to assess it. So INCPEN must be saying either that a DRS would have no impact on litter, or that any reduction would be so small that its costs would outweigh the benefits.

All or nothing

For a DRS to have little or no impact on the amount of litter in the environment, one would have to assume that the expected reductions in littering of deposit-bearing items would be partially or wholly offset by increases in littering of non-deposit bearing items. There is no evidence that this happens. On this basis alone it seems hard to defend a claim that addressing specific items via a DRS is pointless.

INCPEN’s position, however, appears to be that any reduction a measure like a DRS might cause in the level of litter in the environment would have no effect on the background littering rate – at least until the (somewhat mythical, and arguably unattainable) point at which a ‘clean environment’ is achieved. Only at this point would individuals change their behaviour, in line with the Cialdini et al. findings.

Does INCPEN’s analysis of a DRS’s impact on liter hit the target? Photo by ChileFoto.com (CC BY-NC 2.0), via Flickr.

Again, this doesn’t feel right. A more reasonable expectation might be that a lightly littered environment breeds litter at a slower rate than a more heavily littered environment. Of course, as shown in the INCPEN video, a lightly littered environment can gradually become a more heavily littered environment, but it might still be a preferable starting point.

This expectation is supported by the academic literature. In fact, Cialdini et al. report this finding in their 1990 paper, as does a 2013 study by Schultz et al. This latter research identified that the level of pre-existing litter (which the researchers rated on a scale from 0-10) was predictive of observed littering behaviour. For every ‘unit’ increase in existing litter, the observed rate of littering increased by 2%.

Double dividend

INCPEN’s line of reasoning would lead one to the conclusion that, unless we can eliminate litter entirely, we might as well do nothing at all to address the problem. Without some pretty despotic levels of intervention, few areas will achieve a state of absolutely no litter, so INCPEN’s recommendation would appear to be that we shouldn’t even bother trying: if even one littered item remains, it will inexorably lead to a plethora of others joining it.

Fortunately, the evidence from the academic literature says something different. A DRS for beverage containers could realistically be expected to cut the quantity of these items that are littered – and the resulting reduction in the amount of packaging litter in the environment would also be predicted to cut the rate of littering of non-deposit bearing items.

Litter does breed litter, but a little less litter breeds litter a little less. Try saying that with a mouthful of gum!