**Table 3.** Yasu River watershed [36,39].


Questionnaires were mailed in February and March 2016 to 34,691 households in 81 randomly selected postal codes within the study area. The questionnaire, written in Japanese, asked about SWB, relationships with others (social capital), nature (forest-related

activities), and other aspects of everyday life. A total of 3,220 questionnaires were returned with a 9.3% response rate.

The average age of the respondents was 65, in comparison to residents' average age among the six cities ranging from 40 to 46 as of 2015. Respondents were 35% female, whereas female residents in the six cities range from 48% to 51% as of 2015 [40].

Forest SWB was measured using one item intended to assess affective evaluations towards local forests. The SWB specifically related to local forests was measured by the responses to the following statement: "I feel happy when I see local mountains." The response options ranged from 1 = "I completely disagree" to 5 = "I strongly agree." The term "mountains," which is interchangeable with "mountain forests" in Japanese, was used as vernacular to question the feelings regarding forests, which are located mainly in mountainous areas in Japan [41]. The distribution of forest SWB responses is presented in Figure 2 with the average score in brackets. Responses were widely distributed, indicating that there was sufficient individual variation.

**Figure 2.** Forest subjective well-being (SWB) in the entire watershed study (feelings regarding local forests) [36]. The authors gave permission to use this chart.

Ordinary least squares (OLS) regressions were performed in STATA to identify significant associations (*p* < 0.10) between forest SWB, demographic factors, and (i) natural, (ii) man-made, and (iii) social capitals, with positive and negative coefficients indicating the direction of these associations (Table 4). In the written explanation of the results to follow, the numbers for each variable in Table 4 will be used in text to refer to their corresponding variables.

**Table 4.** Summary results of regression analyses of the entire watershed study: explanatory variables with positive and negative statistically significant coefficients (*p* < 0.10) \* [36].


\* Explanatory variables representing built or manufactured capital such as hospitals are omitted from this table for simplicity.

Considering the groups of variables, certain demographic variables have positive or negative correlations with the indicators.


Next, we considered variables relating to respondents' behavior regarding forests.


Unexpectedly, there was no significant associations between the physical presence of forests and forest SWB.

• The forest ratios of the respective postal areas where respondents resided did not correlate with forest SWB.

The adjusted R<sup>2</sup> value for the models with dependent variable 1 (forest SWB) was 0.156. The F-statistic *p*-value of the corresponding OLS was less than 0.0001. The adjusted R <sup>2</sup> values for models with a general SWB were larger than those for the forest SWB (0.360).
