In this section, we first describe the econometric models used to identify the use of the Likert-type environmental awareness index for predicting altruism by comparing models with the environmental awareness index and the SRAS as dependent variables. Second, we explain the details of the survey that was administered to obtain the data used in the econometric models.
2.1. Econometric Model
Our analyses are performed using the ordered probit model. Denoting
as our observed ordinal dependent variable and
as the highest rank order such that
,
is determined by an unobserved latent variable
:
where
are the threshold parameters such that
. Then, the probit model can be expressed as:
where
and
are
vectors of observed explanatory variables and unknown parameters, and
is the random error term. The probability of observing a particular ordinal outcome for a given
:
where
represents the cumulative distribution function of the standard normal distribution.
The
s are estimated by the log-likelihood function written as:
where
is an indicator variable that equals 1 when
and equals 0 otherwise.
Using the above ordered probit model, we estimated a model where
represents the SRAS and two models where
represents the Likert-type environmental awareness indices:
We identify if the environmental awareness and SRAS indices have a bidirectional relationship and test if variables such as demographical characteristics (age, income, education, gender, etc.), degree of political and social awareness, and attitudes toward donation that are likely to have impacts on the SRAS have the same effects on the environmental awareness indices. For these purposes, all the independent variables included in Equations (5)–(7) are the same except for the SRAS and the environmental awareness indices.
Table 1 shows the descriptions of the variables included in the equations. As seen in the table, the SRAS refers to the self-report altruism scale that was developed based on Rushton et al. [
14]. To compare this scale with the Likert-type environmental awareness index, we converted the average SRAS score to an ordinal variable, as described in the table. The details of the items on the SRAS are presented in
Table 2. We administered the SRAS to Japanese respondents, but the original questionnaire created by Rushton et al. [
14] was used for Canadian respondents. Thus, in this study, some of the items included in the original index were modified to fit Japanese people. In this process, we included questionnaire items 12, 15, and 21 of Oda et al. [
21], which developed the Japanese version of SRAS.
The average of the SRAS was calculated based on the respondents’ pick regarding the questions presented in
Table 2. Then, since the environmental awareness indices are ordinal variables, the average SRAS was converted to a similar ordinal variable as shown in
Table 1. As seen in
Table 1, most of the variables are ordinal variables except male, child, pure donation, and gifted donation being dummy variables.
To test the reliability of our modified SRAS, we estimated Cronbach’s
, Revell’s
, and McDonald’s
[
22] using the items shown in
Table 2. Most studies testing the reliability of an index only use Cronbach’s
, but Zinbarg et al. [
22] suggest that Cronbach’s
can result in overestimation of the scale, and thus, it is important to use several indices to determine the validity of the scale. Knight et al. [
23] suggest that the acceptable thresholds for the
,
, and
coefficients are above 0.8, 0.7, and 0.6, respectively, and thus, these values are used as the cutoff criteria herein.
Table 3 and
Table 4 show the results of the reliability analyses. It is apparent from
Table 3 that all the raw
values of the items included in the SRAS were below the overall Cronbach’s
coefficient shown in
Table 4. This suggests that deleting any of the items in
Table 2 will not improve the overall reliability of the scale. Furthermore, the estimated
,
, and
coefficients shown in
Table 4 are all above the cutoff criteria, indicating that the SRAS items used in this study (see
Table 2) are statistically valid.
Next, we explain the details of the environmental awareness indices used in the study (see
Table 1). Many studies measuring an individual’s environmental awareness level use a Likert-type scale [
5,
10,
24]. Thus, this study also asked the survey respondents about their interests in environmental issues using this type of scale. In this study, the environmental awareness indices are ranked on a five-point scale (1 = “have no interest at all” to 5 = “have a strong interest”). It is known that an individual’s attitude regarding environmental policy differs depending on whether the environmental issue described in the survey is about global or local environmental problems [
25]. Thus, in this study, we obtained data for an individual’s environmental awareness level toward both global and local environmental problems. Equation (5) captures how global and local environmental awareness influence the SRAS, Equation (6) examines the effects of the SRAS on the global environmental awareness index, and Equation (7) identifies the relationship between the local environmental awareness index and the SRAS. Using Equations (5)–(7), the magnitude of the effects between the SRAS and the environmental awareness indices are investigated by estimating the average marginal effects of the changes in these indices.
To determine the appropriateness of using environmental awareness indices for predicting an individual’s altruism level, we tested the effects of demographic variables (age, educational achievement, income levels, gender, having children), the degree of political and social awareness, and attitudes toward donation on the environmental awareness indices and the SRAS. This was done by comparing the estimation results of the models presented in Equations (5)–(7). The variables to examine their effects on the environmental awareness indices and the SRAS are included as in Equations (5)–(7). The multicollinearity among all the independent variables was assessed by the variance inflation factor (VIF). The VIF ranged between 1.20 and 2.52, suggesting that all the models of this study have low multicollinearity among the covariates.
The reason for including the demographic variables is because previous studies suggest that these variables are likely to influence an individual’s altruistic behavior. For example, Chou [
26] finds that age has a positive effect on altruism, Westlake et al. [
27] reveal that educational achievement has a positive impact on altruism, and Hoffman [
28] suggests that altruism increases with income. Furthermore, Piper and Schnepf [
29] and Branas-Garaza et al. [
30] indicate that women are more altruistic than men. Although the effects of having children on altruism vary under different circumstances, the issue has long been discussed in the context of parental altruism [
31,
32,
33].
The social interest variable considers how an individual’s degree of political and social awareness is related to altruism and environmental awareness levels. This variable is included in our models since previous studies suggest that altruism and environmental awareness are strongly related to prosocial behavior and political orientation [
34,
35,
36,
37].
Finally, we included pure donation and gifted donation (see
Table 1) in the models to consider the effects of donation attitudes on individuals’ altruism and environmental awareness levels. Japan is ranked as one of the lowest developed nations in the World Giving Index [
38], and Japanese citizens are known to be less active and generous with respect to donation. However, more Japanese people are becoming interested in donation due to the program called the Furusato Nozei (hometown tax donation system), which was launched in 2008 [
39]. However, the Furusato Nozei program is a donation system based on reciprocal gifts. Whenever the donors donate to the municipal governments, these local governments send back special gifts to the donors in exchange for the money donated. Hence, the type of donation popular in Japan is not a donation in a pure sense, and when administering a survey on Japanese respondents to assess their donation attitudes, it is important to separate whether they are willing to donate with or without receiving gifts in return. Variables to assess both pure donation and gifted donation are included in the models to control for these effects.