Does Information about Personal Emissions of Carbon Dioxide Improve Individual Environmental Friendliness? A Survey Experiment

: The purpose of this study is to identify factors that can change the environmental friendliness of individuals in the context of climate change issues in terms of values, beliefs, controllability, concern, attitude, intention, and behavior through a survey experiment, and to test the hypothesis that providing information about the amount of carbon dioxide (CO 2 ) emissions attributable to an individual with its threshold value motivates him/her to reduce that amount using statistical analyses (the Mann–Whitney test) and multivariate regressions (the ordered logit model). It is crucial to change the behavior of individuals as well as organizations to reduce the emissions of CO 2 for solving climate change issues, because the aggregate amount of individual CO 2 emissions is too large to ignore. We conducted a survey experiment to detect factors affecting the environmental friendliness of individuals. Subjects of the experiment were 102 students at Shiga University in Japan. They were randomly provided with communication opportunities, information about individual or group CO 2 emissions, and information about their threshold value. The ﬁnding is that provision of information about the amount of individual and group CO 2 emissions may be able to improve that person’s environmental friendliness in terms of values, beliefs, concern, attitude, intention, and behavior.


Introduction
Climate change due to emissions of greenhouse gases, such as carbon dioxide (CO 2 ), is damaging natural and human systems on Earth [1]. The data for the global land and ocean surface average temperature shows warming of 0.85 • C from 1880 to 2012 [1] (Pachauri et al., 2014), and global warming is expected to increase by 1.5 • C between 2040 and 2050 if it continues to rise at the current pace [2]. The current atmospheric concentration of CO 2 (398.5 ppm) has already exceeded a planetary boundary (350 ppm) [3,4]. We need to reduce CO 2 emissions by a minimum of 33% by 2055 in order for our emissions to stay below the high risk zone (550 ppm) in 2100 [5].
Climate change issues are a problem of the tragedy of the commons, because the use of atmospheric sinks for greenhouse gases is non-rival and non-excludable [6][7][8].
In the tragedy of the commons, freeriding pays from the viewpoint of each economic entity, because monitoring individual actions is almost infeasible in global environmental problems like climate change.
Multiple governing authorities working as a part of polycentric governance are likely to be effective in solving climate change problems on different scales [9][10][11]. Each unit in such a polycentric system independently develops norms and rules within various The definitions of the seven categories in our study are provided as follows. Values (sense of value) are norms that an individual has accepted about nature and human action. Beliefs are individual thoughts or feelings regarding the relationships between human activities and the environment based on his/her individual experiences, contexts, and values. Controllability describes individual judgements on people's capacity to control environmental issues. Concern encompasses individual interests in climate change issues, environmental conditions, pro-environmental technologies, environmental policies, and so on. Attitude describes individual ways of evaluating environmental problems, climate change issues, and one's own lifestyle. Intention is an individual commitment to behave in an environmentally friendly manner. Behavior is individual action on climate change issues.
Some previous studies in the field of environmental psychology and behavior have attempted to clarify the relationships among the psychological categories [32][33][34]. However, we should note that the analysis of the correlation or causality between those psychological categories is outside of the scope of this paper. Experimental designs of the previous studies are different from ours.

Subjects
The subjects of the experiment were 102 students majoring in either economics or data science at Shiga University, a national university in Japan. Similar experiments targeting university students have been conducted in the past [35][36][37][38][39]. The discrepancy in responses from students and relevant professionals is not large [40].
To solicit participation, we advertised the study by putting up posters on campus and sending out emails. We originally collected the contact information of 119 interested persons before the start of the experiment. All of these 119 persons were asked to complete the entry survey online, which 109 persons out of the 119 did. We set the sample size to at least 10 respondents in each of the ten treatment groups in advance and randomly distributed the 109 persons to the ten groups. However, 102 persons out of 109 participated in the treatment stage of the experiment. They were treated as the subjects of this experiment.
Before the experiment commenced, we told the participants that private information would be kept confidential and came to an agreement with all subjects in terms of the conditions of the experiment. Then, we obtained written, informed consent for their participation.

Experimental Design
In this experiment, participants completed four steps: (i) entry survey, (ii) pre-survey, (iii) treatment, and (iv) post-survey.
Entry survey: All subjects were asked to complete an entry survey prior to the day of the experiment. Although gasoline consumption should be included, we target the following six items instead, considering the difficulty of response and calculation. They answered questions about their monthly consumption of electricity (kWh), water (m 3 ), city gas (m 3 ), liquefied petroleum gas (LPG; kg), diesel ( ), and kerosene ( ), as well as their socioeconomic characteristics. We used these data to calculate each participant's individual CO 2 emissions based on the CO 2 emission factor for each energy source as follows: 0.496 (kg CO 2 /kWh) for electricity, 0.54 (kg CO 2 /m 3 ) for water, 2.23 (kg CO 2 /m 3 ) for city gas, 3.00 (kg CO 2 /kg) for LPG, 2.58 (kg CO 2 / ) for diesel, and 2.49 (kg CO 2 / ) for kerosene [41,42]. For participants who were unable to provide data about part of their monthly energy consumption, we used the average energy consumption of all participants who answered that question after cutting both the upper and lower 10% of the figures to decrease the effects of outliers.
Pre-survey: To measure environmental friendliness in the context of climate change issues in the seven psychological categories, we administered a pre-survey consisting of 70 questions (Tables A1-A7 in Appendix A) to each subject between January 20 and 22, 2020. We asked 10 questions in each of the following seven categories that are commonly used in studies of environmental psychology and behavior: values (sense of values), beliefs, controllability, concern, attitude, intention, and behavior [27][28][29][30][31]. The original questions and categories are based on previous research conducted in different contexts [29,[43][44][45][46][47][48][49], but were modified, customized, and adapted to the current context of Japanese society (Tables A1-A7 in Appendix A). Answers to each question were provided based on a five-point Likert scale.
Treatment: For the treatment stage of the experiment, the 102 participants were randomly divided into ten groups with different treatments ( Table 1). One of the experimental treatments is the provision of information about the threshold value. We used 85.7 kg per person per month as the threshold value based on the following information. The target amount of CO 2 emissions in the household sector is 0.601 times smaller than the amount of CO 2 emissions in 2013 [50]. We calculated the amount of monthly CO 2 emissions for each energy source of the Japanese household sector in 2013 using the data provided by the Greenhouse Gas Inventory Office of Japan [51]. We then multiplied the aggregate amount by 0.601 to obtain 85.7 kg. Subjects in Groups 5 and 6 ( Table 1) compared their individual emissions with this threshold. Furthermore, we multiplied the figure, 85.7 kg, by the number of subjects in a group to obtain the group threshold. Thus, subjects in Groups 9 and 10 ( Table 1) compared their group emissions with that group threshold in the experiment. When we gave participants the information about the threshold value, we explained it as follows: "We are providing the threshold value so that you can compare the amount of your (or your group's) CO 2 emissions with it. The threshold value indicates the amount of CO 2 emissions that we must not exceed. Thus, you can simply compare them. If the amount of your (or your group's) CO 2 emissions exceeds the threshold value, it is considered unsustainable." Figure A1 in Appendix A shows a sample feedback sheet with information about the amount of CO 2 emissions per month with the threshold value. Subjects also received information about its component percentages ( Figure A2 in Appendix A).  1  9  2  7  3  11  4  10  5  10  6  11  7  10  8  10  9 12 10 12 Members of Group 1 (the control group) had no opportunity for communication and were given no information about their level of CO 2 emissions or the related threshold value. Participants in some groups had the opportunity to communicate with each other for 25 min to discuss potential measures for reducing their CO 2 emissions. Additionally, participants in some groups were provided with information about their individual emission levels or the total amount of emissions of their group, and the threshold of the emissions was provided to some groups. For example, members of Group 6 had the opportunity to communicate about possible measures for reducing individual CO 2 emissions and were given information about their individual CO 2 emission levels, as well as the CO 2 emission threshold. Participants were able to compare their emission levels with the threshold value and find out whether their amount was above or below the threshold.

Post-survey:
In the last step of the experiment, each subject completed the post-survey between January 27 and 29, 2020 (one week after participating in the treatment phase of the experiment). The post-survey contains the pre-survey questions but in a different order. Each subject received 2000 Japanese Yen (19.35 US dollars) as a reward for completing all steps of the experiment.

Data Analysis
We obtained each participant's responses to the 70 questions of the pre-and postsurveys. The five-point Likert scale values of the answers were sorted in descending order for each question, with higher values indicating greater environmental friendliness. Using the results of the pre-and the post-surveys, we conducted two kinds of analyses: the Mann-Whitney test on the difference of average changes in the aggregate score in each category (values, beliefs, controllability, concern, attitude, intention, and behavior) between different treatment groups, and an ordered logit model to explain changes in responses between the pre-survey and the post-survey for each question.
In preparation for the Mann-Whitney test, allocating an integer number (1)(2)(3)(4)(5) to the sorted Likert scale values for each question in ascending order, we calculated the aggregate scores of the integer numbers in each of the seven categories. Then, deducting the total score of the pre-survey from that of the post-survey, we obtained the amount of change in the aggregate scores in each category. To be precise, those Likert scale values are ordinal scales, and thus it is not considered correct to aggregate or deduct the answered values in the categories. However, as is often the case with this type of survey analysis, we conducted the calculations for analyzing the tendency of the impacts that can be attributed to different experimental treatments.
We carried out the statistical tests on the difference in the average amount of change between distinct treatment groups (Table 2). First, we tested the difference between the groups with and without a communication opportunity (Groups 2, 4, 6, 8, and 10 vs. Groups 1, 3, 5, 7, and 9 in Table 1), using the Mann-Whitney test. Second, we tested the difference among groups without any information about emissions (Groups 1 and 2), with information about individual emissions (Groups 3, 4, 5, and 6), and with information about group emissions (Groups 7, 8, 9, and 10), using the Mann-Whitney test for specifying the difference between any two groups out of the three. Third, we tested the difference between the groups having information about individual emissions with their threshold value (Groups 5 and 6) and the groups having information about individual emissions without their threshold value (Groups 3 and 4), using the Mann-Whitney test. Fourth, we tested the difference between the groups having information about group emissions with their threshold value (Groups 9 and 10) and the groups having information about group emissions without their threshold value (Groups 7 and 8), using the Mann-Whitney test.  Next, we applied an ordered logit model to the changes in responses to each question between the pre-survey and the post-survey. The changes were measured as an ordinal scale, because we set the five-point Likert scale values for the answers in the survey. Thus, an ordered logit model was used. In these regressions, the dependent variable is the degree of the change in response, measured as an ordinal scale. If the answer that an individual provided to a question in the pre-survey is 1 and the corresponding answer in the postsurvey is 4, the environmental friendliness in terms of the question increases by 3 points. The degree of change ranges from −4 to +4 points for each question. The independent variables are the conditions of the treatments and some socio-economic characteristics (Table A8 in Appendix A). Figure 1 shows statistically significant differences between distinct treatment groups in terms of changes in respondents' scores of environmental friendliness in the context of climate change issues from pre-survey to post-survey in the seven psychological categories. We identify five meaningful differences with statistical significance out of 42 tests. information about group emissions without their threshold value 20 7 and 8

Results
Next, we applied an ordered logit model to the changes in responses to each question between the pre-survey and the post-survey. The changes were measured as an ordinal scale, because we set the five-point Likert scale values for the answers in the survey. Thus, an ordered logit model was used. In these regressions, the dependent variable is the degree of the change in response, measured as an ordinal scale. If the answer that an individual provided to a question in the pre-survey is 1 and the corresponding answer in the post-survey is 4, the environmental friendliness in terms of the question increases by 3 points. The degree of change ranges from −4 to +4 points for each question. The independent variables are the conditions of the treatments and some socio-economic characteristics ( Table A8 in Appendix). Figure 1 shows statistically significant differences between distinct treatment groups in terms of changes in respondents' scores of environmental friendliness in the context of climate change issues from pre-survey to post-survey in the seven psychological categories. We identify five meaningful differences with statistical significance out of 42 tests.      . Statistically significant differences between groups in changes of scores. Statistically significant differences between groups in changes of scores (continued). Note: ***, **, and * represent a 1%, 5%, and 10% significance, respectively. Each test evaluates the difference in changes of respondents' scores from pre-survey to post-survey between groups with different treatments.  Figure 1 shows the results of statistical tests on the difference in average values of change in the aggregate score of environmental friendliness. The average value of changes in the score of attitude category for subjects having information about individual emissions is larger than that for subjects not having any information about their emissions, with a 1% statistical significance (Figure 1a). The average value of changes in the score of attitude category for subjects being provided with information about group emissions is larger than that for subjects not having any information about their emissions, with a 5% statistical significance (Figure 1b). The average value of changes in the score of belief category for subjects who have information about the threshold value in addition to information about group emissions is larger than that for subjects who do not have information about the threshold value, with a 10% statistical significance (Figure 1c). The average value of changes in the score of concern category for subjects having information about individual emissions is larger than that for subjects not having any information about their emissions, with a 10% statistical significance (Figure 1d). The average value of changes in the score of concern category for subjects who are provided with information about group emissions is larger than that for subjects who are not provided with any information about their emissions, with a 10% statistical significance ( Figure 1e). Table 3 shows the results of the regressions of the ordered logit model that include statistically significant estimated parameters in the category of values (VAL). Provision of information about individual emissions improves environmental friendliness in question VAL-3, which asks how much the participant agrees with the statement "all countries including developed and developing ones should equally make efforts to mitigate climate change". Provision of information about group emissions improves environmental friendliness in question VAL-8. Provision of information about the threshold value in addition to information about individual or group emissions decreases the degree of environmental friendliness in question VAL-6, which asks how much the participant agrees with the statement "firms should aim to solve environmental issues". The opportunity of communication positively affects environmental friendliness in question VAL-8, which asks how much the participant agrees with the statement "you should be responsible for nature, considering the impact of what you are doing on nature and the environment". In addition to those experimental treatments above, some socio-economic factors also affect changes in the degree of environmental friendliness due to the participation in the experiment. Individuals majoring in economics show decreased environmental friendliness in question VAL-3 and improve it in question VAL-9, which asks how much they agree with the statement "extinction of a species is inevitable". An increase in monthly income enhances environmental friendliness in question VAL-5, which asks how much the participant agrees with the statement "Governments should aim to solve environmental issues."

Results
The experience of learning about environmental studies also augments environmental friendliness in question VAL-5. It should be noted that the impacts of these socio-economic factors, not the experimental treatments, stimulated environmental friendliness due to the participation in the experiment. For example, people with higher income tended to have higher pro-environmental values in question VAL-5, probably because the experiment provided the impetus for promoting their pro-environmental consideration.  Note: Standard errors in parentheses. **, and * represent a 5%, and 10% significance, respectively. Table 4 shows the results of the regressions of the ordered logit model that include statistically significant estimated parameters in the category of beliefs (BEL). Provision of information about individual emissions improves environmental friendliness in question BEL-2 (how much the participant agrees with the statement "human beings can change natural environments, but they cannot perfectly control them"), BEL-3, and BEL-10 (how much they agree with the statement "my action affects neighbors' and friends' happiness"). Provision of information about group emissions enhances environmental friendliness in question BEL-3. Provision of information about the threshold value in addition to information about individual emissions increases the degree of environmental friendliness in question BEL-6, but it decreases the degree of environmental friendliness in question BEL-9, which asks how much the participant agrees with the statement "resilience of nature is strong enough to endure environmental burdens generated by human economic activities". Provision of information about the threshold value in addition to information about group emissions increases the degree of environmental friendliness in questions BEL-2 and BEL-10. The opportunity of communication positively affects environmental friendliness in question BEL-3 (how much the participant agrees with the statement "rules are beneficial to everyone") and BEL-6 (how much they agree with the statement "climate change is caused by CO 2 emissions due to human economic activities"). In addition, some socioeconomic factors acted on environmental friendliness. Compared to male participants, female participants show improved environmental friendliness in question BEL-7, which asks how much they agree with the statement "when human beings change nature, they often face dreadful outcomes". Participants who major in economics show decreased environmental friendliness in questions BEL-6, BEL-7, and BEL-10. An increase in monthly income enhances environmental friendliness in questions BEL-3 and BEL-6. Having learned about environmental studies lowers environmentally friendliness in question BEL-8, which asks how much the participant agrees with the statement "technological innovation can solve environmental issues".  Note: Standard errors in parentheses. **, and * represent a 5%, and 10% significance, respectively. Table 5 shows the results of the regressions of the ordered logit model that include statistically significant estimated parameters in the category of controllability (CTL). Provision of information about individual and group emissions worsens environmentally friendliness in question CTL-9, which asks how much the participant agrees with the statement "it is almost impossible to reduce the amount of emissions of greenhouse gases, considering the current economic activities". Provision of information about group emissions improves environmental friendliness in question CTL-3, which asks how much they agree with the statement "changes in your lifestyle could solve climate change issues". It decreases the degree of environmental friendliness in question CTL-6, which asks how much the participant agrees with the statement "it is unlikely to conserve biodiversity because of uncertainty in nature". Furthermore, some socio-economic factors influence environmental friendliness. Compared to male participants, female participants show enhanced environmental friendliness in questions CTL-7 (measuring how much they agree with the statement "it is too costly to address environmental issues"), CTL-8 (how much they agree with the statement "refusing plastic bags is not enough to contribute to solving global environmental issues"), and CTL-9. Those who major in economics show worsened environmental friendliness in question CTL-7. An increase in monthly income improves environmental friendliness in question CTL-1 (measuring how much the participant agrees with the statement "you can reduce the amount of electricity consumption, being careful about it") and CTL-4 (how much they agree with the statement "cooperative actions could solve climate change issues"), but lowers it in question CTL-6.  Note: Standard errors in parentheses. ***, **, and * represent a 1%, 5%, and 10% significance, respectively. Table 6 shows the results of the regressions of the ordered logit model that include statistically significant estimated parameters in the category of concern (CON). Provision of information about group emissions enhances environmental friendliness in question CON-5, which asks how much the participant is concerned about "the amount of plastic rubbish (e.g., plastic bottles)". Provision of information about the threshold value in addition to information about individual emissions increases the degree of environmental friendliness in question CON-7, which asks how much they are interested in "new technologies to reduce the amount of CO 2 emissions and the development of such technologies." On the other hand, provision of information about the threshold value in addition to information about group emissions lowers the degree of environmental friendliness in question CON-4, which asks how much the participant is interested in "renewable energy". In addition, some socio-economic factors influence changes in the degree of environmental friendliness because of the participation in the experiment. Compared with male participants, female participants show improved environmental friendliness in question CON-1 (asking how much they are interested in "services and goods that generate as small an amount of CO 2 emissions as possible") and CON-3 (asking how much they are interested in "electric, hydrogen, and hybrid vehicles"). However, female participants show a decreased degree of environmental friendliness in question CON-2, which asks how much they are interested in "the amount of my electricity consumption". Participants majoring in economics show enhanced environmental friendliness in question CON-2, as well. The experience of learning about environmental studies augments environmental friendliness in question CON-7. Table 6. Estimation results of regressions in the category of concern (CON).  Note: Standard errors in parentheses. ***, **, and * represent a 1%, 5%, and 10% significance, respectively. Table 7 shows the results of the regressions of the ordered logit model that include statistically significant estimated parameters in the category of attitude (ATT). Provision of information about individual emissions has a positive impact on environmental friendliness in question ATT-3 (how similar the following statement is to their attitude: "I disapprove of government's inaction on environmental issues"), ATT-7 (how similar the following statement is to their attitude: "it is inevitable that typhoons often come"), and ATT-8 (how similar the following statement is to their attitude: "I feel sad, seeing the Earth polluted by human activities"). Provision of information about group emissions improves environmental friendliness in questions ATT-7 and ATT-8. It, however, lowers the degree of environmental friendliness in question ATT-2, which asks how similar the following statement is to the participant's attitude: "I do not want the temperature in summer to rise more". Adding information about the threshold value to information about individual emissions worsens environmental friendliness in questions ATT-7 and ATT-8. Including the threshold value in information about group emissions enhances environmental friendliness in questions ATT-2 and ATT-10 (how similar the following statement is to their attitude: "I am willing to change my lifestyle for environmental conservation and preservation"). Compared to male participants, female participants' degree of environmental friendliness decreases in question ATT-4 (how similar the following statement is to their attitude: "I am sad about a decrease in beaches due to sea-level rise"). For those who major in economics, the degree of environmental friendliness increases in question ATT-4 and decreases in question ATT-8. An increase in monthly income enhances environmental friendliness in question ATT-2.  Note: Standard errors in parentheses. ***, **, and * represent a 1%, 5%, and 10% significance respectively. Table 8   Note: Standard errors in parentheses. **, and * represent a 5%, and 10% significance, respectively. Table 9 shows the results of the regressions of the ordered logit model that include statistically significant estimated parameters in the category of behavior (BEH). Provision of information about individual emissions enhances environmental friendliness in question BEH-6, which asks how similar the following statement is to the participant's behavior: "I use public transportation in urban areas". Provision of information about group emissions lowers the degree of environmental friendliness in question BEH-1, which asks how similar the following statement is to their behavior: "I move by bicycle and/or on foot as frequently as possible". Provision of information about the threshold value in addition to information about individual emissions decreases the degree of environmental friendliness in question BEH-6. Provision of information about the threshold value in addition to information about group emissions increases the degree of environmental friendliness in question BEH-1 and decreases it in question BEH-3, which asks how similar the following statement is to the participant's behavior: "I do not buy beverages in a plastic bottle". Provision of a communication opportunity improves environmental friendliness in question BEH-1. However, it reduces the degree of environmental friendliness in questions BEH-2 (how similar the following statement is to their behavior: "I buy goods that have low food mileage"), BEH-3, BEH-6, and BEH-7 (how similar the following statement is to their behavior: "I point out other's environmentally unfriendly behavior"). In addition, some socio-economic factors have an impact on changes in the degree of environmental friendliness due to the participation in the experiment. Female participants show worsened environmental friendliness in question BEH-2. Those who major in economics improve their environmental friendliness in question BEH-10, which asks how similar the following statement is to their behavior: "I get plastic and/or paper bags when I go shopping". An increase in monthly income enhances environmental friendliness in question BEH-1 and lowers it in question BEH-10. The experience of studying environmental subjects elevates the degree of environmental friendliness in question BEH-2 and BEH-6. However, it reduces it in question BEH-5, which asks how similar the following statement is to their behavior: "I avoid buying plastic products". Note: Standard errors in parentheses. ***, **, and * represent a 1%, 5%, and 10% significance, respectively.

Importance of Information about Individual CO 2 Emissions
One of the interesting results is that information about individual CO 2 emissions could enhance individual environmental friendliness in the aggregate score of measures of attitude ( Figure 1a) and concern (Figure 1d), and in some questions of values (Table 3), beliefs (Table 4), attitude (Table 7), intention (Table 8), and behavior (Table 9). This implies that an individual can be motivated to take pro-environmental actions if he/she obtains information about the environmental burdens he/she imposes. For example, many individuals might be interested in making contributions to the reduction of CO 2 emissions. However, they might not know how to do so without relevant information about their current level of individual CO 2 emissions.
In many cases, an individual is able to reduce their CO 2 emissions by changing their patterns of consumption. Changes in individual behavior also alter the processes of industrial production to be done in a more environmentally friendly manner. However, information about the emissions related to the consumption of commodities is often unavailable. If every individual had access to such information, they could choose to use a lower carbon commodity among substitutes. Governments also need to appropriately incentivize producers to provide information to consumers about the amount of CO 2 emissions that occurred in the production process of a product, because carbon-labelling might enable individuals to make eco-friendly purchases [52][53][54]. The carbon footprint of commodities should be evaluated and available to every individual, ideally based on the life cycle assessment of the commodities.
We should note that information about individual emissions negatively affects individual environmental friendliness only in question CTL-9 (how much the participant agrees with the statement: "it is almost impossible to reduce the amount of emissions of greenhouse gases, considering the current economic activities") in the category of controllability (Table 5). Understanding how much CO 2 an individual emits may make him/her feel that it is too difficult to sufficiently reduce these emissions, although people need to know how much CO 2 they emit and what activity is responsible for most of the issues related to climate change in order to take specific actions to reduce their CO 2 emissions.
Every individual might be latently willing to take action to reduce CO 2 emissions. However, people generally tend to believe that organizations, such as firms, municipalities, central governments, and international organizations are to blame for climate change and have the responsibility to alleviate it, because an individual's potential contribution is small compared to the size of the problem. This is because individuals might think that it is impossible to resolve the problem of climate change in question CTL-9 from the individual point of view. They may rely on governments and firms for solutions. Although school climate strikes and movements to spur governmental actions on climate change issues are now common, every individual needs to change his/her behavior to alleviate climate change problems, because the aggregate changes in individual behavior could be impactful. In the polycentric system, both individual and organizational actions on climate change are required to resolve the issue.
On the other hand, the provision of information about group CO 2 emissions can have positive or negative impacts on individual environmental friendliness (Tables 3-9), although positive impacts were observed in the aggregate score of measure of attitude ( Figure 1b) and concern (Figure 1e). This depends on the questions. Even if people are provided with information about group emissions, its positive effects on changes in pro-environmental motivation are not necessarily obtained. We should note that information about group emissions lowers individual environmental friendliness in the category of behavior, although information about individual emissions enhances it in the same category (Table 9). It is not necessarily effective to provide information about group emissions, but information about individual emissions seems more important for a positive behavior change.

Effects of Threshold Value on Individual Environmental Friendliness
We obtain complicated results about the effects of the threshold value on changes in individual environmental friendliness. In some cases, the provision of thresholds affects participants' values, beliefs, concern, attitude, and behavior in both positive and negative directions (Tables 3, 4, 6, 7 and 9). What we need to focus on is that the addition of information about the threshold value sometimes lowers individual environmental friendliness, while providing information about individual emissions positively affects it in terms of questions ATT-7 (how similar the following statement is to the participant's attitude: "it is inevitable that typhoons often come"), ATT-8 (how similar the following statement is to their attitude: "I feel sad, seeing the Earth polluted by human activities"), and BEH-6 (how similar the following statement is to their behavior: "I use public transportation in urban areas"). This means that participants tend to feel that it is too difficult to reduce the amount of CO 2 emissions sufficiently when they know the ratio of the emission amount relative to the standard. In many cases, the actual emission amount exceeded the threshold stated in the Paris Agreement. For this reason, people should not be informed of the threshold if quick changes in individual behavior are needed. We should be cautious when we add information about the threshold value to information about individual emissions, as this may offset the positive effect of information about individual emissions on individual environmental friendliness in some cases.
Some studies have reported that an individual increases the extent of their proenvironmental behavior if he/she is given information about the relative size of their environmental burden within a group (e.g., an individual's relative proportion of the entire electricity consumption), as this information enables him/her to compare their value with that of others [20,[22][23][24][25]. In our study, we provided threshold values instead of a relative value, which enabled participants to compare their CO 2 emission with the Paris Agreement standard. Thus, our results are distinct from those of previous studies in that changes in pro-environmental behavior depend on the relationship between the amount of CO 2 emissions and the threshold value. We believe that the standard, based on the Paris Agreement, is more significant as a criterion for individual behavior than relative values within a group. This is because the amount of individual emissions that is smaller within a group may still exceed the standard provided in the Paris Agreement. In this respect, our results are more meaningful, because they illustrate that individuals must urgently abide by the standard in order to solve climate change issues effectively.

Impact of Communication on Environmental Friendliness
The provision of a communication opportunity might be able to improve environmental friendliness in the category of values and beliefs (Tables 3 and 4). Considering the questions in which communication has a positive effect with statistical significance (VAL-8: how much the participant agrees with the statement: "you should be responsible for nature, considering the impact of what you do on nature and the environment", BEL-3: how much they agree with the statement "rules are beneficial to everyone", and BEL-6: how much they agree with the statement "climate change is caused by CO 2 emissions due to human economic activities"), subjects could gain a correct understanding of some aspects of climate change issues through communication. As a result, they might change their recognition of the current state related to these problems.
However, it decreases the degree of environmental friendliness in the category of intention and behavior, except for one question (Tables 8 and 9). Through communication, subjects may have clarified what they can do and what they cannot do to alleviate the issues of climate change. Based on their current lifestyles, they may have learned that they have no choice other than accepting what they currently do and that they cannot easily change their current behavior. These results tell us that we need to consider how and what we should discuss during a communication opportunity in order to change individuals' behavior to be more in line with the solutions for climate change problems.

Income Effects on Environmental Friendliness
The results of the regression analysis show that environmental friendliness, measured in 10 questions addressing each category of values, beliefs, controllability, attitude, intention, and behavior, improves more significantly after participating in the experiment and as monthly income becomes larger, except for two questions (Tables 3-5 and 7-9). This implies that people with a higher income would be more motivated to change their behavior in a pro-environmental manner if they reviewed their current environmental friendliness and lifestyle. People with a higher income can also more easily improve their intentions, because a large monthly income expands their range of choices in terms of pro-environmental behavior (Table 8). Obviously, it is possible for people with a lower income to reduce their living costs by making their behavior environmentally friendly as well. For example, the cost of walking is cheaper than that of using a car. However, it is often the case that such low-cost options are inconvenient and provide disutility. While their income is not sufficiently high, people tend to focus on their own welfare, which is closely linked to consumption. As their income increases, people may be more altruistic and inclined to form pro-environmental beliefs, focusing on the relationships between human activities and the environment. In addition, people with a high income may have grown in a rich family with wider options such as education opportunities and life/travel experiences. Thus, they are inclined to have more ideas and options to change their behavior in a pro-environmental manner.
As mentioned above, an increase in monthly income lowers environmental friendliness due to the participation in the experiment in questions CTL-6 (how much the participant agrees with the statement "it is unlikely to conserve biodiversity because of uncertainty in nature") and BEH-10 (how similar the following statement is to their behavior: "I get plastic or/and paper bags when I go shopping"). These two exceptions have not been well explained up to now. This finding needs to be further researched, including not only university students, but also the general public as experimental subjects.

Other Significant Findings
The ordered logit regression analysis reveals five additional findings (Tables 3-9). First, economics students tend to give low priority to environmental consideration in seven questions, while they increase their environmental friendliness in four questions. This may be because economics students are more self-interested and sensitive to economic profits [55][56][57].
Second, female participants enhance their environmental friendliness because of the participation in the experiment in seven questions, while they lower it in three questions. As compared with male participants, female participants tend to consider environmental aspects more important than other socio-economic aspects. This coincides with the results of similar studies [58][59][60].
Third, learning about environmental studies improves environmental friendliness after the experiment in five questions, while it reduces the degree of environmental friendliness in two questions. On the whole, studying environmental subjects seems to be effective for enhancing pro-environmental motivations. This may depend on the contents of environmental studies.

Conclusions
We carried out an experiment with surveys to detect factors that can change the environmental friendliness of individuals in the context of climate change issues in terms of values, beliefs, controllability, concern, attitude, intention, and behavior. Then, we analyze the data obtained from the experiment by using the statistical tests and the ordered logit model. The results of the analysis show that information about individual CO 2 emissions might mostly be able to enhance individual environmental friendliness in terms of values, beliefs, concern, attitude, intention, and behavior, even though the time scale of this experiment was short term. Based on these results, providing relevant information about individual CO 2 emissions may be able to contribute to mitigating climate change issues. In other words, individuals would not change their behavior to reduce CO 2 emissions without relevant information about their individual emissions, even though they may be motivated to do so. However, the provision of information about group CO 2 emissions has positive or negative impacts on individual environmental friendliness. It positively affects pro-environmental attitude and concern.
On the other hand, we obtain complex results about the provision of information regarding the threshold value. In some cases, we observe a positive impact of it on environmental friendliness, but in other cases, a negative impact is observed. The addition of information about the threshold value lowers individual environmental friendliness while providing information about individual emissions positively affects it in terms of some questions related to global environmental issues: how similar the following statement is to the participant's attitude: "it is inevitable that typhoons often come" (ATT-7); how similar the following statement is to their attitude: "I feel sad, seeing the earth polluted by human activities" (ATT-8); and how similar the following statement is to their behavior: "I use public transportation in urban areas" (BEH-6). The big gap between the current emission amount and the threshold might make subjects feel that it is too difficult to reduce the amount of CO 2 emissions to a substantial extent. We believe that information about the comparison between our current emissions and the threshold is crucial, and thus, we need to avoid providing such a negative mental effect while showing information about the threshold.
Furthermore, we have obtained some other findings. Communication opportunities may be able to improve environmental friendliness in terms of values and beliefs, while they lower it in the category of intention and behavior. Environmental friendliness in values, beliefs, controllability, attitude, intention, and behavior improves in the experiment as income level increases. Economics students tend to place less importance on environmental consideration. Female participants are likely to give a high priority to environmental friendliness, compared to male ones. Learning about environmental studies is inclined to enhance environmental friendliness.
Based on the main results of this study, it is urgently necessary to devise policies that enable every individual to access information about their carbon footprint (i.e., carbon footprint labelling), at least in their usual life. Every individual should obtain information about his/her daily and/or monthly carbon footprint based on the life cycle assessment and the impacts of his/her efforts to change specific actions regarding CO 2 emission reduction. Most people are currently addicted to smart phones, and thus it may be a promising idea to pre-install some software programs to automatically calculate and record carbon footprint in smart phones. If people glanced at the data of their own carbon footprint as compared to the threshold value whenever they used a smart phone, they might change their behavior in a more sustainable manner. Another practical option is to set up meters of monthly carbon footprint in many places in each municipality, like thermometers erected at corners of roads. They show a monthly carbon footprint that is clearly compared to the threshold value at the same time. They may work like information about the amount of group emissions.
In our current world, we have been managing CO 2 emissions through macro-entities such as countries, municipalities, and firms, but we have not currently accomplished the necessary reduction of CO 2 emissions to be in the target zone. This is because we cannot attribute the aggregate emission amount to every individual as a crucial constituent member in a macro-entity. It is necessary to consider behavior changes of all the stakeholders in the polycentric system in terms of a CO 2 emission target. We can manage only macroentities if we do not consider the clear responsibility of every individual or motivate them to reduce their emissions. Under this condition, we have no effective measure, because the total amount of emissions of micro-entities is too large to ignore. Every individual should be practically responsible in order to make the polycentric governance work well. Improvement of individual behavior would positively affect changes in the other stakeholders' actions, as well. The interactions between them would lead to the target reduction of CO 2 emissions.
Several limitations of this study should be considered in future research. First, we need to expand the experiment so that the results are more generalizable to other locales. In the current study, the sample size and participants' diversity are limited, although some studies taking an experimental approach are similar to ours in terms of sample size and subjects' diversity [61][62][63]. However, we should note that the experiment in this paper treated students in a university in Japan. They might make more collective decisions in comparison with European and American people and be more pro-environmental as compared with people in low-income countries. Second, the long-term effects of the intervention should be observed. In the current study, only short-term effects were analyzed, and it is unclear how long the effects observed in this study remain. We need to conduct a follow-up study after the initial experiment in order to investigate whether long-term changes occur in environmental friendliness in terms of the seven categories. Furthermore, we want to analyze the impact of the frequency of experimental treatments over a certain period of time on changes in environmental friendliness. This should also be carried out in a longterm study. Third, the impacts of the treatments provided in the experiment were measured via questionnaires, which were based on subjective evaluations, even regarding behavior. In addition to psychological changes based on subjective assessments, we need to capture real behavioral changes that can be objectively measured. Fourth, we cannot measure or observe to what extent subjects understand information about their CO 2 emissions in this experiment. Variation in the level of understanding may make differences in changes of their environmental friendliness. An explanatory variable of educational background is included in the ordered logit model as a proxy to partly measure the level of understanding. On the other hand, we think that people would be motivated to change their behavior in a more pro-environmental manner if they perceived a connotative message in the data of their own emissions and noticed the gap between their personal emissions and the threshold value. Fifth, the mechanisms for enhancing the environmental friendliness of every individual need to be linked to macro governmental policies to tackle issues of climate change. We need both individual and collective pro-environmental actions in polycentric systems to find effective solutions. Responsibility should not be imposed only on governments but also on every individual. We need to specify how we can, or should, inspire their responsibility with relevant information about their CO 2 emissions. Sixth, we need to develop specific methods to provide every individual with information about their personal CO 2 emissions, considering the results we have obtained in this study. This is obviously a future research topic in the sustainability field.
Finally, we need to find factors that will have changed people's non-sustainable repeated behavior and lasting routines in their daily life for dealing with climate change problems in the future research. Certainly, individuals have the incentive to freeride other people's pro-environmental behavior [6], but they merely repeat their usual activities as their lifestyles without intention of freeriding in most cases. Unless we can change our unsustainable continued behavior in our daily life, we will be confronted with calamities due to climate change sooner or later.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author. The data are not publicly available due to privacy protection.   Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author. The data are not publicly available due to privacy protection.

Acknowledgments:
The authors would like to thank Ms. Matsuoka and Ms. Yamazaki for clerical work. The authors would like to appreciate the participation of Shiga University students in the experiment. The authors would like to give many thanks to three anonymous reviewers for their productive comments.

Conflicts of Interest:
The authors declare no competing interests. Figure A1. A sample feedback sheet. Feedback sheets were provided in Japanese. Participant's CO2 emissions with the threshold are provided. Figure A2. A sample feedback sheet. Feedback sheets were provided in Japanese. Proportions of CO2 emissions in categories are provided. Figure A2. A sample feedback sheet. Feedback sheets were provided in Japanese. Proportions of CO 2 emissions in categories are provided.  slightly disagree, 5: disagree Note: All the questions above were translated from Japanese. The questions were given to subjects in Japanese. The sentence in parentheses is the original Japanese question. slightly disagree, 5: disagree Note: All the questions above were translated from Japanese. The questions were given to subjects in Japanese. The sentence in parentheses is the original Japanese question. Response alternatives 1: agree, 2: slightly agree, 3: neither agree nor disagree, 4: slightly disagree, 5: disagree Note: All the questions above were translated from Japanese. The questions were given to subjects in Japanese. The sentence in parentheses is the original Japanese question.   I am little interested in, 5: I am not interested in Note: All the questions above were translated from Japanese. The questions were given to subjects in Japanese. The sentence in parentheses is the original Japanese question.   Note: All the questions above were translated from Japanese. The questions were given to subjects in Japanese. The sentence in parentheses is the original Japanese question. Note: All the questions above were translated from Japanese. The questions were given to subjects in Japanese. The sentence in parentheses is the original Japanese question.