Globally 65-72% of greenhouse gas (GHG) emissions are driven by household consumption with food, shelter, and mobility making the largest contributions (Hertwich & Peters, 2009; Ivanova et al., 2016). For this reason, demand-side mitigation targeting individual behavior change is increasingly seen as an essential course of action (Creutzig et al., 2016, 2018; Dietz et al., 2009; Girod et al., 2014; Ivanova et al., 2020; Moran et al., 2020; Wynes & Nicholas, 2017). In North America, the average annual carbon emissions are 13.4 tCO2e per capita (Ivanova et al., 2020) and both U.S. and Canadian governments have called for a 50% reduction of GHG emissions by 2030 (Schmidt, 2021). Some of the most impactful actions include minimizing air travel, living car-free, driving electric or high-efficiency vehicles, shifting to renewable energy, and eating vegetarian/vegan diets, each of which is estimated at reducing personal emissions by at least 500kg of CO2e per year (Akenji et al., 2019; Girod et al., 2014; Ivanova et al., 2020; Koide et al., 2021; Wynes & Nicholas, 2017).
Despite the urgency to mitigate climate change (Capstick et al., 2015), most people lack the knowledge of GHG emissions associated with daily behaviors and have difficulties in determining which behavior changes are effective at reducing emissions (Gifford, 2011; Thøgersen, 2021). For example, people underestimate the impact of air travel, meat consumption, and high-energy appliances, while overestimating the impact of littering, recycling, and switching to canvas shopping bags (Attari et al., 2010; Camilleri et al., 2019; Lazzarini et al., 2016; Shi et al., 2018; Truelove & Parks, 2012; Wynes et al., 2020), possibly because of disproportionate education and government communication on these actions and conflation of pro-environmental and pro-climate behaviors (Wynes et al., 2020). Consumers are relatively unaware of what changes will reduce food emissions, with buying local and organic often incorrectly suggested as more impactful than reducing animal-product consumption (de Boer et al., 2016; Kause et al., 2019). Despite these knowledge gaps, the perceived effectiveness of mitigation measures for mitigating climate change is positively correlated with the willingness to adopt climate actions (de Boer et al., 2016; Pickering et al., 2021; Truelove & Parks, 2012).
To the extent that governments continue to rely on people’s voluntary reduction, it is essential to enable individual behavior change to reduce emissions by making one’s carbon footprint more transparent. To this end, personal carbon footprint calculators inform people about GHG emissions based on their consumption and lifestyle choices. In a sense, these calculators have been used as a public education tool to raise people’s awareness of their carbon footprint. Existing versions—available for use on the internet and within smartphone applications—have extensive variability in the domains covered, level of detail included, constant factors assumed, and additional features provided. Despite their common use and the growing literature about ways to improve the calculators (e.g., Birnik, 2013; Mulrow et al., 2019; West et al., 2016), there is a dearth of research investigating behavioral outcomes associated with the calculators.
Carbon calculators have become an increasingly prevalent tool to quantify GHG emissions in the last decade, along with substantial criticisms regarding transparency, consistency, and data quality (Padgett et al., 2008). As climate mitigation has become increasingly urgent, there are few large-scale attempts to quantify the impact of these calculators. A recent review identified only seven papers that measured the effect of calculators on mitigation behaviors and intentions and found support for the use of calculators to increase knowledge and awareness about carbon emissions but not to change behavior (Dreijerink & Paradies, 2020). For example, one randomized control trial examined the effect of face-to-face carbon calculator interviews on people’s attitudes, electricity consumption, vehicle mileage, and flights over two years (Büchs et al., 2018). The study found that climate awareness and concern increased in the calculator condition relative to the control condition, but there were no significant differences in any of the behavioral measures. However, this study may have suffered from a self-selection bias where participants have already adopted most of the lifestyle adjustments before the study. Another study compared the use of a carbon calculator to a personal carbon footprint management system and found that the footprint management led to a significant reduction in emissions over time, but there was no control condition in the study and the convenience sample of 66 undergraduate students was not randomly assigned to the two conditions (Lin, 2016). Other experiments tested carbon calculators by giving participants bogus, randomly assigned feedback, showing that “worse than peers” or “negative” feedback produced greater pro-environmental intentions driven by social comparisons and mediated by eco-guilt (Adams et al., 2020; Armenta et al., 2020; Mallett et al., 2013; Toner et al., 2014). Other studies used within-subject longitudinal designs but without a control condition, and found improved knowledge and awareness of emissions, increased reduction intentions, and lower emissions below the baseline (Aichholzer et al., 2012; Dowd et al., 2012; Gram-Hanssen & Christensen, 2012; Sutcliffe et al., 2008).
Limitations in the existing studies on carbon calculators include small sample sizes, convenience sampling of undergraduate students and people with pre-existing interests in climate change, absence of control conditions and randomization, lack of consideration for which features of the calculator are most effective, and limited effect on behavioral outcomes. These limitations preclude drawing causal conclusions on the impact of the carbon calculator on intended behavior change.
Despite these methodological limitations, there are more serious reservations on using personal carbon calculators as a tool to mitigate climate change. The reasoning is that calculators put too much emphasis on individual behavior change and reduce people’s support for broader system change (Chater & Loewenstein, 2022). This reasoning is supported by past studies that demonstrated backfiring effects of a green nudge that increased individual pro-environmental behaviors but simultaneously reduced support for climate policy. For example, one study showed that nudging people to choose renewable energy by using defaults reduced their support for a carbon tax policy (Hagmann et al., 2019). However, a recent meta-analysis found positive spillovers for most pro-environmental behaviors where engaging in the first pro-environmental behavior increased the likelihood to engage in a second behavior, but there was a small negative spillover if the second behavior was policy support (Maki et al., 2019). This said, no study in the past has examined the spillover effect of carbon calculators’ features on civic climate action (e.g., policy support, voting). Thus far, there is no empirical evidence to support the hesitations to use personal carbon calculators as a tool to mitigate climate change.
The current pre-registered study aims to examine the causal impact of personalized feedback on emissions from a carbon calculator on intended climate action. This study addresses the limitations in the previous literature by using a randomized controlled trial with participants from the general public that targets a specific feature of the calculator (i.e., personalized feedback) and measures change in intended individual climate action as well as civic climate action on system change. We hypothesize that people who receive personalized feedback from a carbon calculator (including their past carbon footprint in 2019, a breakdown of emissions by consumption domain, a personal reduction target, and tailored recommendations) will have a greater reduction in intended emissions in 2023, compared to those who do not receive this feedback. In addition to measuring private behavioral intentions, we will also examine potential spillover effects to pro-climate civic intentions and carbon offsetting. This study thus will provide the first evidence of the spillover effect of carbon calculators’ features on civic climate action. Using a validated eco-guilt measure (Mallett, 2012), we will also examine whether footprint information increases eco-guilt and how eco-guilt is related to climate action. This is building upon past work that showed eco-guilt is associated with pro-environmental intentions and actions (Ferguson & Branscombe, 2010; Mallett, 2012; Moore & Yang, 2020; Wyss et al., 2021) and may be a strong predictor for climate mitigation behavior and policy support (Brosch, 2021). Finally, we want to see if the personalized feedback would change perceived climate mitigation responsibilities for consumers, businesses, and governments. In particular, highlighting personal carbon footprint may increase perceived mitigation responsibilities for consumers.