Recycling in Belfast

I've been reading a recent paper from another conservation psychologist, Tim Kurz, who has done some recent work in Australia and here in the UK. He's a friend of my supervisor and I've exchanged a few emails with him. This particular paper is focused on recycling in Belfast. Now that may seem like a pretty mundane topic, but this paper is useful for a few reasons, but in particular because it illustrates many of the counter-intuitive, initially baffling nature of most enviro-psych research.

As I've said before, a major flaw in most enviro-psych research is the complete dependence on self-report data. That is, the researcher asks people how much they recycle or save energy or whatever (usually through a written or online questionnaire), and that person's response is taken as valid. The problem with that is that people are really bad at making estimates about their own behaviour, and the data is often not very accurate. Checking actual behaviour - measuring energy used or weighing recyclables - is very intensive in terms of time and money, but it is the gold standard. To get enough statistical power to run any decent analyses, we need to have data on many people, usually at least a hundred and often many more, just depending on the design. Weighing 100 recycling bins every week for a few months is a big pain in the neck, but that is how it has been done before. The Belfast study got around this by using a new technology I've been drooling over for months: every recycling bin has a unique bar code. The recycling truck automatically weighs and scans every bin, logging the data in a database. It's the kind of system conservation psychologists daydream about while in the shower. I've been hoping we'll get a similar system in Exeter soon, but it's at least years off if it ever happens here. Anyway, this Belfast study has some pretty impeccable behavioural data. They also took psychometric data on concern towards the environment, attitudes towards recycling, and sense of community. These psychological variables can be statistically compared with people's real recycling rates to see exactly what relationships are there, and what predicts what.

Before I tell you the results, are there any guesses? Raise your hand if you think general concern towards the environment is related to recycling. Mmm, ok. What? No, keep those hands up. Ok, now raise your hand if you think attitude towards recycling is related to actual recycling. Ok. Now raise your hand if you think sense of community plays a role. You can put your hands down. Now, get a piece of paper, and write down how much you think these three things are related to actual recycling. Do this in a percentage, which we will call "shared variance." It's a kind of measure of how much two measures overlap, or how well you can predict one if you know the other. For example, altitude and air pressure have about 100% shared variance because it is a virtually perfectly correlated relationship.

And now... the answers:
General concern for the environment and actual recycling : 0.0% shared variance
Attitude towards recycling and actual recycling: 3.6% shared variance
Sense of community and actual recycling: 4.0% shared variance

Surprised? That's what a lot of enviro-psych results look like. It is actually incredibly difficult to predict behaviour, and people's attitudes do not go as far in helping with that prediction as our intuition would suggest. So the holy grail has been to find better predictors of behaviour; to an extent this can be improved by just asking more and more specific attitudes: "How do you feel about recycling glass?" is more predictive than "How do you feel about recycling?" and "How do you feel about recycling this particular glass bottle right now?" is most predictive of all. Unfortunately, the more specific you get, the less practical it is in an applied setting. The other best predictors of pro-environmental behaviour are: gender (women do more pro-environmental behaviour than men) and past behaviour (people have done an action in the past are more likely to do it in the future). Of course, both of these are profoundly unhelpful in applied work, as one is unchangeable and the other circular. There are other factors that play a role: how much neighbours recycle, perceived risk of not taking action, and others, but they generally do not account for more than 5% of shared variance, leaving nearly all of the behaviour unexplainable. Environmental education, which I think is a terrific thing, generally accounts for 2% or less of the variance in actual behaviour. This is the tough nut to crack. Every intuitive measure clocks out at very low levels (low for applied value, high enough for proof-of-concept value), leaving nearly all of the behaviour unexplained. On one hand, that's pretty poo-ey, as influencing human behaviour is the most critical part of tackling climate change (more important than technology, or anything else.) On the other hand, it is exciting because these are mostly uncharted psychological waters, with most of the treasure still undiscovered (ughh, pirate metaphors are never good), which can seem like a very rare thing when searching through thousands of journal articles.