We selected two target behaviors using the results of our previous study, in which we selected 58 daily PEBs and surveyed the practice rate of and reasons for each behavior in Seoul and Tokyo [1] . Among those 58 behaviors, we selected the target behaviors for the present study based on the following criteria: a) the behavior shows a relatively low practice rate (<60%), b) lack of information determines the non-practice of the behavior, and c) the effect of information provision on the behavior can be easily tested over a short-term period. Two behaviors that fulfilled these criteria were selected: 1) use of one’s own cup or tumbler at cafés, instead of disposable cups (“My Cup”; B27 in [1] ) and 2) purchase of Carbon Cashbag products (“Carbon Cashbag”; B49 in [1] ). “My Cup” behavior had a 59.1% practice rate in the first study. The main reasons not to conduct the behavior were that respondents considered it “inconvenient” and “bothersome” [1] , but this behavior was selected for the present study because it can easily be changed during a short period via information provision. The “Carbon Cashbag” program was begun by Korea’s Ministry of Knowledge and Economy in 2008. When people buy environmentally friendly products with Carbon Cashbag labels, they gain points that are redeemable for various goods and services. The practice rate of this behavior was 28.9% in the first study. The main reason not to conduct the behavior was cited by respondents as “no recognition” [1] , which indicates the important effects of information provision on PEBs. In addition, neither of these two behaviors requires equipment installation and both can be implemented within a short time period.

The content of the information provided in the questionnaire was chosen based on three psychological factors: risk arousal, knowledge, and descriptive norms.

“Risk arousal” is the enhancement of awareness through the provision of risk information with a pessimistic tone. The effects of risk arousal have been discussed in various studies [15] - [17] . Reference [15] proposed a “two-phase model” for environmental behaviors, in which two intention phases―“goal intention” and “implementation intention”―determine behavior. In his model, risk arousal is considered one of the determinants of goal intention.

According to [18] , environmental knowledge can be categorized into three types: 1) declarative knowledge, 2) knowledge about procedure (i.e. action-related), and 3) effectiveness. Declarative knowledge is defined as knowledge about facts, such as knowledge about what global warming is. Knowledge about how to conduct behavior is called “procedure” knowledge, which is one of the key predictors of behavior [6] - [8] . Reference [8] showed that the combination of procedural and declarative information has significantly positive effects on recycling behavior. “Effectiveness” knowledge related to effects driven by behavior, such as knowledge that CO2 or fuel costs can be reduced by switching from a gasoline-fueled car to a hybrid car. This concept is similar to the combination of perceived effectiveness and benefit evaluation in [19] two-phase model, mentioned above. Reference [20] hypothesized that objective information about effectiveness can change people’s perception of effectiveness, which inspires more efficient PEBs. They provided information about CO₂ reduction driven by electricity savings and other environmental behaviors, and demonstrated that objective information about effectiveness raised people’s PEB awareness.

According to [3] , social norms are categorized as descriptive norms and injunctive norms. Descriptive norms are defined as the cognition of the behavior of others. This is reflected in people’s behavior in response to the behavior of others. Reference [4] demonstrated the effect of descriptive and personal norms on recycling behavior and found that personal norms had a short-term effect, whereas descriptive norms worked more effectively in the long term. Reference [3] showed that the effect of descriptive norms was more significant than that of personal norms for recycling behaviors in public places.

With the above information in mind, we designed three types of information for the two target behaviors: a) risk arousal, b) knowledge, and c) descriptive norms.

Regarding “My Cup”, we provided information about waste generation and the lack of landfill sites in Seoul (as shown on Seoul’s metropolitan homepage) as risk arousal information (A). The volume of waste reduction that results from using personal cups or tumblers instead of disposable cups [21] was provided as effectiveness information (B1), while a list of shops that provided discounted drinks to people who bringing personal cups or tumblers was provided as procedure information (B2). The practice rate (59.1%) was provided as descriptive norm information (C).

Regarding “Carbon Cashbag,” we provided information on the seriousness of greenhouse gas emissions in Seoul (as shown on Seoul’s metropolitan homepage) as risk arousal information (A). Methods through which to accumulate Carbon Cashbag points (B3) and lists of Carbon Cashbag products on the market (B4) were provided as procedure information. We also provided declarative information (B5) due to the low degree of recognition of the Carbon Cashbag program. The practice rate (28.9%) was also provided as descriptive norm information (C).

Details on the information provided in the study are found in Table 1.

1) Seoul

Figure 1 shows the average scores of 12 respondent groups for “My Cup” in Seoul. The table below the graphs shows the combinations of information provided to each group. The scores range from 1.0 to 5.0 for intention and from 0.0 to 1.0 for practice.

The group that was given no information had the lowest score (2.44) and the group to which risk arousal and effectiveness information was given had the highest (A + B1, 3.30). A tendency for scores to increase with information provision was observed. When a one-way-ANOVA test was applied to the intention results, a significant difference was found among the groups (F (11,828) = 2.986, p < 0.01).

To identify significant differences among groups, we carried out a post-hoc (multiple comparison) test― Tukey-Kramer Hochberg’s GT2―as an analysis of equal-variance samples. Significant differences were found between [No Information] and [A + B1], between [No Information] and [A + B1 + C], and between [No Information] and [A + B2 + C] (p < 0.01).

Regarding practice, five groups ([A], [A + B2 + C], [B1 + C], [A + B1], and [A + B2]) showed higher scores than the [No Information] group. In particular, the group to which risk arousal and effectiveness information was given had the highest score (A + B1, 0.34). Yet, while this trend was observed, statistically significant differences were not observed, mainly due to the smaller sample size of each group (F (11,828) = 0.996, p = 0.448, n = 70 × 12 groups).

2) Tokyo

Figure 2 shows the average scores of the 13 groups for “My Cup” in Tokyo. In addition to the 12-group set in Seoul, one additional group was given all of the information [A + B1 + B2 + C] in Tokyo in order to analyze the effect of this combination.

The intention scores in Tokyo were relatively lower than those in Seoul. The [No Information] group had the lowest (1.78), and an upward trend was observed along with information provision. The same trend was observed in Seoul. Higher intention scores were seen in the [A + B1 + C], [A + B2 + C], and [A + B1 + B2 + C] groups (2.92, 2.91, and 2.95, respectively). In addition to the combination of risk arousal (A) and knowledge (B), the effect of descriptive norms (C) was significant in Tokyo, though not in Seoul. Significantly different intention scores were also found among the groups (F (12,542) = 4.31, p < 0.001).

Dunnett’s T3 was used as a post-hoc test due to the unequal variance of the samples. A significant difference from the [No Information] group was found for five groups: [A + B1], [A + B2], [A + B1 + C], [A + B2 + C], and [A + B1 + B2 + C] (p < 0.05). The difference between [A + C] and [A + B1 + B2 + C] was also significant (p = 0.032). The tendency for the addition of knowledge information (B1, B2) to increase intention scores was observed.

In the case of practice, [A] (0.15), [B1 + C] (0.20), [A + B1 + C] (0.15), [A + B2 + C] (0.21), and [A + B1 + B2 + C] (0.14) showed slightly higher scores than the [No Information] group (0.05). The highest score was observed by [A + B2 + C] (0.21). Yet, although these trends were observed, the difference was not statistically significant (F (12,542) = 1.33, p = 0.195).

3) Comparison between Seoul and Tokyo

Overall, the intention and practice scores in Seoul were higher than those in Tokyo. Our previous study also showed that the practice rate of “My Cup” in Seoul was significantly higher than that in Tokyo, which is attributed to the fact that intensive waste reduction programs, such as charging for waste bags, were begun earlier and have become more widespread in Korea than in Japan [1] .

In terms of the influence of information provision, some differences between the two cities were observed. Risk arousal information [A] had more influence on intention in Seoul, while knowledge information [B] had more influence on intention in Tokyo. Based on the practice rate of “My Cup”, we can assume that a lack of information acted as more of a barrier to practice in Tokyo than in Seoul. Yet, knowledge information was accepted more effectively, and had more influence on intention, in Tokyo.

Figure 3 shows the average scores of 16 groups for “Carbon Cashbag” in Seoul. In addition to risk arousal [A], and procedure [B3, B4] information, we also provided declarative information [B5] to all of the groups because the original degree of recognition of “Carbon Cashbag” was not high.

As seen in Figure 3, even when only declarative information was provided, respondents’ intention to perform “Carbon Cashbag” behavior increased, and no statistical significance was observed among the groups (F (15,1036) = 1.176, p = 0.284). In our previous survey [1] , we revealed that the practice rate of “Carbon Cashbag” was quite low (28.9%) and that one of the main reasons not to conduct the behavior was that respondents did not know about the behavior itself; therefore, it was indicated that people’s intention would increase simply through declarative information provision. In addition, the combination of risk arousal, procedure, and descriptive norm information [A + B4 + B5 + C] yielded a slightly higher intention score (4.07) than the other combinations.

In the case of practice, although the difference was not statistically significant (F (15,1036) = 1.314, p = 0.186), relatively higher scores were observed in groups [B3 + B5 + C] (0.33), [B4 + B5 + C] (0.37), [B3 + B4 + B5 + C] (0.33), [A + B3 + B5 + C] (0.34), and [A + B4 + B5 + C] (0.30).

We selected the information that would be provided through the free paper Focus in Seoul on the basis of the above results. The information that yielded the highest intention scores, as well as relatively high practice scores, was selected; these were [A + B1] for “My Cup” and [A + B4 + B5 + C] for “Carbon Cashbag”. The information was shown on the cartoon page―as explained in Section 2.3 and shown in Figure 4―using the same explanation text as in the online survey. The figures shown with the text were designed by a professional designer.

There were 101 respondents for the “My Cup” intention survey and 74 for “Carbon Cashbag”. Fifty of the “My Cup” respondents and 21 of the “Carbon Cashbag” respondents took the additional survey about behavior practice three weeks later.

In the “My Cup” survey, the average scores for intention and practice were 3.78 (n = 101) and 0.40 (n = 50), respectively. Respondents who answered the intention question with the responses “I want to do this very much”, “I want to do this somewhat”, and “I want to do this” (n = 85) were also asked about the usefulness of the provided information. For instance, we asked questions such as, “Was the risk information useful to you,” to which subjects responded on a 6-point scale, from 6 (very much) to 1 (not at all).

The results are shown in Figure 5. The average usefulness score was observed to be significantly higher for [A] (5.26) than for [B1] (5.13) (χ² ＝ 88.34, df = 9, p < 0.001).

In the “Carbon Cashbag” survey, the average scores of intention and practice were 3.85 (n = 74) and 0.43 (n = 21), respectively. Respondents who answered the intention question with the responses “I want to do this very much”, “I want to do this somewhat”, and “I want to do this” (n = 65) were also asked about the usefulness of the provided information. The results are shown in Figure 6. The average usefulness score for [B5] (4.98) was significantly higher than for [B4] (4.83) and [C] (4.52) (χ²-test, p < 0.001). This result again indicates that descriptive information can be useful for increasing intention, especially for behavior to which lack of basic information is a barrier.