A few suggestions:

Online self-report of gas/electricity/water meter readings by consumers would be very efficient and potentially offer a very large N. An appropriate percentage of those self-report readings could be independently verified to establish the statistical validity of the measure.

Online purchasing behaviour might be another possibility, with consumers choosing between environmentally friendly or regular products, correlating this with attitude surveys etc. Obviously this would require a partnership with an online store or seller of some kind, but does offer the possiblity of quick and easy data collection. It would even be possible to set up your own store for minimal outlay using drop shipping.

The research paper I've been working on recently uses the Facial Action Coding System to measure emotional reactions based on facial expressions. This could be another avenue for creating a quantitative measure, e.g. looking at facial reactions to short video clips based on environmental issues. Although the manual version of FACS coding is very labour intensive, more and more automated versions are becoming available (although these might not be available for free).

What about measuring the number of housholds that have installed solar panels and are selling energy back to their electrical grid? Or, the number of households that have a hybrid car? Just a thought.

recycling centers, that offer money for recycling of materials by weight such as aluminum cans, they have to keep records. I believe Canada still recycles glass bottles for a small refund.

There's a lot of reporting - and confusion, I think - about veganism being a growing choice. There is data available showing the growth rate. Could there be data showing what percentage of people made this choice for environmental reasons?

Can I ask how 'pro-environmental' is understood by those being used for measurement? I wonder if some changes may have been believed to be positive by the individuals but perhaps not according to the facts / boundaries for the report. How would you factor pro-environmentally mis-informed behaviour? (This question is for my own interest, sorry).

Mark McGee Researchers study it from two main angles, a) what people think affects environmental outcomes, including misconceptions about relative impact (e.g., many erroneously think aerosols --> climate change); and b) what behaviors have the largest impact on environmental outcomes, e.g., individual behaviors like transportation and diet, and collective behaviors like voting and protesting. Hope that helps.

Many of these answers are helpful, and I'm still looking for that golden goose where I could both learn about the individuals or manipulate the situation for an experiment while measuring changes in objective behavior.

Thanks a lot for initiating this discussion, Cameron!

If we assume that all (or many) kinds of pro-environmental behavior have something relevant in common (and only then it makes sense that we study them under the same name, I think), I suggest that we try to identify and isolate this common denominator. We can then create situations in the laboratory that reflect the contingencies that are involved in many field situations in which individuals can behave pro-environmentally.

An example for this is the FISH task created by Gifford and colleagues (https://www.ncbi.nlm.nih.gov/pubmed/11029814) that allows observing participants’ responses to the conflict between short-term individual and long-term collective consequences. This is a conflict that may be relevant to many pro-environmental behaviors. What the task does not involve are actual consequences for the natural environment (so maybe the behavior that is studied here can be described as long-term oriented/cooperative/economically sustainable rather than pro-environmental).

As an alternative, we have developed and validated the Pro-Environmental Behavior Task (PEBT) as a laboratory task that involves actual consequences, not only for the participant, but also for the environment (https://www.sciencedirect.com/science/article/pii/S0272494418301282?via%3Dihub).

The PEBT requires participants to make a number of trips. For each trip, they can choose between an environmentally friendly (e.g., the bike) and an environmentally unfriendly (e.g., the car) mode of transportation. Following their choice, they have to endure a waiting period, which is longer for the environmentally friendly than for the environmentally unfriendly option. However, whenever participants choose the environmentally unfriendly option, a series of lights is illuminated for the duration of the trip. The associated waste of energy makes this option truly environmentally unfriendly and choosing the environmentally friendly PEBT option an actual pro-environmental behavior. Anyone who is interested in the task can download it here: https://osf.io/tcnza/.

I believe that tasks like these are our best (scientific) shot if we want to understand pro-environmental behavior and its determinants. If we then develop interventions based on this knowledge and want to know if the effect of an intervention is strong enough to stand out from the noise typical for a particular pro-environmental behavior, then we need to go back into the field.

One thought with regard to field observations of pro-environmental behavior: when we weigh trash or read meters, we do not observe actual pro-environmental behavior, we observe a product of multiple behaviors of multiple people. That does not have to be a problem, but it might be (think of a visiting cousin who may like to shower for half an hour). One of the field observation methods that I find most interesting is the use of GPS data to measure speeding behavior (https://www.ncbi.nlm.nih.gov/pubmed/21376917) or travel mode choice (https://www.sciencedirect.com/science/article/pii/S0272494416300299).

It might be useful to consider if pro- and anti-environmental behaviors are two sides of the same coin or merely related but distinct from each other. If the former is true, then apart from positive behavior(s) measurement (e.g., energy conservation, etc.), negative behavior(s) measurement (e.g., food wastage, etc.) would be possible. If the latter is true, however, then measurement should be restricted only to positive behavior(s).

To my understanding it makes a big difference if we want to understand pro-environmental behaviour itself or if we intend to use pro-environmental behaviour as a proxy for behaviour with an environmental impact in a more general way.

In the field I am primarily active in (lifestyle research with a focus on energy and climate relevant behaviour), PEB scales are most commonly used as proxy variables for a wider range of behaviour. However, most common PEB-scales do not provide a reliable basis for classifying individuals with regard to their actual environmental behaviour due to being prone to social desirability.

As a part of the project ECHOES (in which we aim at understanding the "driving factors" of energy related behaviour) we designed an alternative approach which still relies on self-reports, but has a stronger focus on considering a larger variety of behaviours (preferably in all areas of life). By doing so, it does not primarily address “low hanging fruits” (like switching off the lights or avoiding plastic bags) but aims at capturing all practices and behaviours that potentially have a significant impact on energy use and Greenhouse Gas emissions.

For example in the area of mobility we ask for Public Transport (daily trip-duration in different types of public transport), for Car use (km per year, fuel consumption and average number of people on board), for Air Travel (hours spent in airplanes during the past year), and for bicycle-use. All of them are asked as fact-oriented and neutral as possible so that we expect the distinction between "good" and "bad" behaviour to be less obvious for respondents. ;-)

A first (application-oriented) pilot-study based on this kind of questions made us optimistic that suchlike questions are more closely related to actual impact-relevant behaviour than PEB-responses. If anyone is interested:Chapter The “Paris Lifestyle”—Bridging the Gap Between Science and C...

An EU-wide survey (EU-28+Norway, Turkey and Switzerland) with an additional emphasis on emission-impacts of consumption and other forms of behaviour is on the way, results should be available by the end of the year. In parallel, colleagues from the University of Graz (Institute of Sociology) just launched a project focusing on the question how validly questions of that kind measure actual behaviour, depending on the respective area of life and the spatial context.

Really looking forward to hearing about your experience with other alternative approaches and/or validation studies!

Best,

Stephan

Stephan Schwarzinger, this sounds excellent. I would love to be sent a copy of anything you publish using this approach. Another option that comes to mind is quantifying the amount of social desirability, and/or controlling for it by measuring it directly. There's a deep history of worrying about this in personality psychology, although it's not my specialty.

Thanks for your great response.

Thank you very much, Cameron Brick !

I agree, having an eye on respondents' tendency to give socially desirable answers (also in our approach) is definitely worth some extra effort. Especially with regard to the question how the influence of soc.-des. answering behaviour might change over time with the likely increasing awareness about the environmental relevance of "normal" everyday behaviour.

I forgot to mention one particularly interesting difference we have noticed: While many PEB-instruments are designed to be one-dimensional, the impact-based approach obtains one-dimensional results only in some special cases - probably more common for those who consistently have very high or very low behavioural impacts. We will have a special focus on such "high" and "low" impact lifestyles in the analysis of the EU survey data to see if that hypothesis is correct. For the comparison between the two different approaches (PEB vs. impact-based), the "low" impact group seems to be of special interest from my point of view - do you agree with that?

Best,

Stephan

Similarly, I have noticed that the dimensionality of PEB scales is different between high- and low-performing groups, such that there is a much clearer and more dominant single factor underlying the behaviors in the high-PEB group. Is that what you meant Stephan Schwarzinger

Yes, exactly - to my understanding this one-dimensionality by design causes a blurred spot regarding the question on the causes for high/low PEB in different areas of life!?

I have a couple papers published that measure actual pro-environmental behavior in the form of a donation, not self-reports of behavior, or intentions of behavior.

A behavioural measure of environmental decision-making for social surveys. Environmental Sociology, 2015. Vol. 1, No. 1, 27-37, Article A behavioural measure of environmental decision-making for s...

Measuring Actual Payment for Biodiversity Protection: The Influence of Religiously and Scientifically Framed Messages. Worldviews, 2018. 22, 263-288. Article Measuring Actual Payment for Biodiversity Protection: The In...

@John_Clements6 , donation is a good tactic. In my experience, the results are messy and difficult, usually with a trimodal distribution peaked at 0, 50, and 100%. How do you choose to analyze that? I see in your abstracts in one paper you used means (assuming normality), and in the other you used donation or not. How did you justify those? I'm only asking because I personally struggle with this!

Hi Cameron, we also had a multimodal distribution in the 2015 paper - 0, 0.5, 0.10, 0.25, and 0.5., and trimodal distribution in the 2018 paper. Both actually used means (but the 2015 paper also compare median, and percent zero responses), and also looked at donation or not (one using ZIP regression, and one using seemingly unrelated regression). In all honesty, we struggled with the best analyses to use too. The first was more exploratory to determine that the method might work, while the second tested specific messaging techniques and I was more interested in determining what influences people to donate (or not). Wish I had a better answer for you on this one!

Hi Guys, I hope that this discussion about investigating real behavior instead of self-report will be interesting for you:

https://spb.psychopen.eu/issue/1339/

Thank you very much, Mr. Dariusz, for the link and this well-collected essential text about the current state of research on behavior. Interesting and stimulating to reflect.

Thank you Cameron Brick for such an important discussion topic!

I agree with you that there's a lot of room for creativity in tackling "the problem" of measuring PEBs.

One thing to additionally consider is the scientific aim of using a particular behavioral measurement, since we could think of two general reasons (to put it simply): 1) testing the straightforward effectiveness of a particular intervention on the occurrence of particular PEB; and/or 2) verifying the underlying mechanisms of a particular intervention or the occurrence of PEB.

From that perspective, as Florian Lange rightly noted - even though many field experiments focus on "measurement of product of multiple behaviors of multiple people" - it's not necessarily a problem if we primarily consider the need of testing the effectiveness of a particular intervention. On the other hand, this can limit our ability to establish theoretical explanations - due to lacking data on dispositional factors (i.e., in field experiments). Therefore it would be fair to assume, that it's best to combine both behavioral and declarative measurements.

However there's still the problem of difficulty in measuring actual behavior (i.e., the amount of resources that we have to spend to conduct it). Even as simple behavior as light-switching require a large amount of resources when relying on observational methods (e.g., using multiple observers and training them; establishing observational protocol; or simply the amount of time to gather the data). We've overcame these obstacles by using occupancy and light data loggers which are highly cost-effective and allow to collect data on behavior of individual participants in a standardized manner (since human observation is prone to errors).

Article The role of normative prompts and norm support cues in promo...

Moreover, data loggers can be used easily in laboratory settings (in which it would be possible to additionally measure dispositional factors of participants).

Just recently we've published an article (regarding a method of measuring light-switching behavior), which presents a step-by-step protocol of using occupancy and light switching logger in field experiments:

Article Measuring Light-Switching Behavior Using an Occupancy and Li...

Additionally, we have created a short video tutorial for using this method (as a supplement for the article):

https://www.jove.com/video/60771

Maybe more solutions to the issue of measuring actual conservation behavior (among many others that have been mentioned) can be found in other technical appliances such as occupancy and light data loggers (i.e., are easy to use and do not require a lot of additional resources)?

Dear Cameron Brick

Re: Measuring actual behavior rather than self-report . Eye tracking is actual behavior. It is clearly becoming quite astonishing what can be "picked up" via eye tracking. In fact (given the measures already generated and used successfully), I believe that fundamental shifts in BASIC perception (NOT intentionally controlled) might be able to be monitored. That would include perceptual shifts at the core of the qualitatively different concept and thought levels/stages of ontogeny. My 1000 pages of writing related to this are all available on RG. Refining and finding evidence for just ONE of my near-clear testable (strictly empirically-based) hypotheses should get a behavioral scientist a Nobel Prize (the first Nobel Prize for behavioral science since 1973 Tinbergen and Lorenz, classical ethology (all behavior) -- their Prize under "Physiology and Medicine" (because of the limited number/natures of Nobel categories)).

I'm very glad to participate in this discussion. I think eye movement movements including Saccade , gaze and blink. These objective behaviors can show the impact of driving tasks on drivers, which is more accurate than a single questionnaire survey.So it is a good way to study the status of drivers.

It is a great honor to participate in this discussion. 90% of the task information of drivers is obtained by eye movement behavior, and studying eye movement behavior is of great help to study the psychology, physiology and behavior of drivers.

I would like to share our recent article on how to ask participants to do an observation of mentally healthy/unhealthy places and photograph in a gated manufacturing plant. Then investigators met the participants for photograph-survey and interview on their feelings of being in those places. The method is called PSPS and it is a good combination of observational and self-reported instruments for a difficult site. I hope it is helpful for you.

Article Quality of sweatshop factory outdoor environments matters fo...