**4.4 Willingness to pay: yes or no—a binary logistic regression**

In order to generate an understanding of the importance of different factors influencing the willingness or non-willingness to pay, we conducted a binary logistic regression. Our hypotheses were that people with more pro-environmental behavior, pro-environmental attitudes, preferences for fossil-free power generation, and younger people are more willing to pay. Accordingly, attitudes toward the environment and energy technologies appeared to be important as impact factors, as well as gender and age. We also tested both, schooling and vocational education, but neither had significant impact on the dependent variable within the binary logistic regression model and were thus left out (**Table 3**). Our final model consists of six variables and considers 971 cases, representing roughly 97% of the sample. The remaining cases were excluded due to missing values regarding the dependent variable. A check for multicollinearity of the explaining variables showed no critical values. Using our binary logistic regression model, the rate of correctly predicted values rose from 51.3% to 66.9%. The pseudo-R squared (Nagelkerke) is 0.207, which is "acceptable" according to Backhaus et al. [50].

*The Hambach Forest in the German Debate on Climate Protection: Is There a Symbolic Value… DOI: http://dx.doi.org/10.5772/intechopen.101597*


*B: coefficient for the standard; S.E.: standard error for the coefficient around the constant; Wald: Wald Chi Square statistics; df: degree of freedom for the Wald Chi Square test; Exp (B): exponentiation of B coefficient, which is an odds ratio. Source: own.*

#### **Table 3.**

*Parameter estimate of the binary logistic regression.*

**Table 3** illustrates the influence of nearly each predictor variable, except gender, to the logistic model and the statistical significance (p < .05) of the Wald Chi Square test, which is obtained by squaring the ratio of the regression coefficient (B) to its standard error (S.E.). According to our analysis, the stated pro-environmental behavior, which normally correlates with strong pro-environmental attitudes, has the strongest impact on the WTP. This observation is not surprising, and this strong relationship has been shown by various studies (e.g., [51, 52]). The Odds Ratio (Exp(B)) indicates that, if the stated pro-environmental behavior increases by one unit, the probability to state a positive WTP increases by roughly 45%. The second largest impact factor is the attitude toward renewable energies (solar, wind, hydro, and biomass), which is also correlated significantly with stated pro-environmental behavior; an increase of one unit here implicates an increase of probability by 35%. This observation supports the assumption that not only values of the forest, but also a favor for the energy transition may have played a role in the decision to state a positive WTP. Further, an eco-centered conviction, measured by a statement regarding the perception of the vulnerability of the earth (Q7.8.3), has a clear positive impact on the WTP, although it is comparatively small. The acceptance of the use of lignite as an energy source (Q2.1e), instead, has a negative impact on the WTP; an increase by one unit of acceptance implies a decrease of roughly 14% in the probability to state a positive WTP. A negative impact on the probability to state a positive WTP can be observed for age and being male, although on a rather low level of significance. According to the literature, no general impact of gender on the WTP for environmental goods can be observed, since other factors such as attitudes, education, or income are generally more important [53]. However, a recent survey found that females probably are more pro-environmental in both Germany and the Netherlands [54]. By separating age from other factors, sometimes a negative impact can be observed, which means that older people are less willing to pay for environmental issues [53]. In these regards, our results are thus in line with previous findings.

#### **4.5 Willingness to pay—ordinary least squares (OLS) regression analysis**

In order to detect significant impact factors on the stated monetary values for the protection of the Hambach Forest, we conducted an OLS regression analysis. The

dependent variable, the stated amount of those willing to pay, was rather log-normal than normally distributed. Therefore, we logarithmized the dependent variable and applied a semi-log model. The following explanatory variables were considered:


However, only a very low rate of explained variation with an R<sup>2</sup> of .081 and a corrected R<sup>2</sup> of .057 could be reached using an OLS regression approach. The results are displayed in **Table 4**. No more than three variables with significant impact on the stated amount of the WTP were observed: gender (p = .01), household income (p = .001), and the degree of agreement with statement Q7.8c about the vulnerability of the earth (p = .05). As generally the case in CVs, household income has a positive effect on the WTP: the higher the income, the more easily people can afford to pay for environmental goods and services. Compared with the other variables, income had the strongest impact on the WTP, but only when considering the average household income. In other models, when per-capita income was considered the explaining variable, no significant impact was observable. Furthermore, in our case also being male had a positive impact on the WTP. It is important to note that males had a significantly higher household income in our survey. However, testing for collinearity did not reveal problematic values. Finally, the impact of perceived higher vulnerability of the earth on the WTP can be considered as an indicator for a higher estimation of existence values as well as fears of losses of environmental goods and services. Stronger agreement with this statement normally goes along with a more nature-centered point of view, which also could explain a higher WTP. The low rate of explained variation leads to the assumption that other factors, which we did not consider in our survey, may be of strong relevance for the stated amount. Imaginable are issues regarding personal budget constraints, a general estimation of woods and forests, or dissatisfaction with climate policies, as well as embedding effects resulting from our payment scenario. Alternatively, the group of respondents with a positive WTP is more homogeneous regarding attitudinal values compared with the same values over all respondents, as an analysis of variance showed.

*The Hambach Forest in the German Debate on Climate Protection: Is There a Symbolic Value… DOI: http://dx.doi.org/10.5772/intechopen.101597*


*Source: own.*

#### **Table 4.**

*Parameter estimate of the OLS regression.*
