Calibration is a fundamental process in fsQCA (Dușa, 2019). Our study adopts the direct method of calibration (Rihoux & Ragin, 2008), which requires us to set three main quantitative anchors to constitute a fuzzy set. These three quantitative anchors are the threshold for full membership, the crossover point, and the threshold for non-membership. In our study, as both antecedent condition―Hofstede’s cultural dimension and outcome condition―environmental performance index range from 0 to 100, we set 50 as the crossover point. Meanwhile, 75 (median of 50 and 100) and 25 (median of 0 and 50) are set to “full in” and “fully out”. However, to avoid the case ambiguity caused by the use of the 50 points as crossover point, we use the score equal to 50.001 as the crossover point in calibration process. Table 2 displays the calibration thresholds for all antecedent conditions and the outcome condition as well as the descriptive statistics.
[Insert Table 2 about here]
4.2 Analysis of necessity
Before creating the truth table, a crucial step of fsQCA, it is indispensable to carry out an analysis of the necessity for each condition. To identify whether any of the six national culture dimensions are necessary for high or low EPI, we conduct necessity analysis to tests whether the antecedent condition is always present (absent) where the outcome condition is present (absent). Schneider and Wagemann (2012) argue that if the consistency of an antecedent condition exceeds the threshold of 0.9, it should be regarded as a necessary condition.
Table 3 shows the result of the analysis of the necessary conditions. As reported in the table, each consistency does not exceed the threshold of 0.9 where the outcome condition is high EPI, so there is no cultural dimension recognized as a necessary condition for high EPI. Conversely, the consistency of power distance (PDI) and collectivism (～IDV) exceeds 0.9, which means high power distance and collectivism are identified as necessary conditions for low EPI.
[Insert Table 3 about here]
4.3 Analysis of sufficiency
After calibrating the raw data and conducting the analysis of necessity, we establish the truth table to assign each case to one of these truth table rows on the basis of Boolean algebra (Schneider & Wagemann, 2010). In the next step, we simplify these configurations by logical minimization and reduce the initial truth table by specifying the frequency value and consistency threshold. We analyze sufficiency following the established fsQCA default rule by adopting a frequency value>1, a consistency threshold>0.75, and a proportional reduction in inconsistency (PRI)>0.7. We assume that the presence and the absence of each condition would affect the outcome on account of the lack of concrete theory or literature.
The qualitative comparative analysis of fuzzy sets usually generates three types of solutions: complex solution, parsimonious solution, and intermediate solution. A complex solution does not incorporate any logical reminders, and conversely, a parsimonious solution incorporates all of the logical reminders, while the intermediate solution merely involves logical reminders in accordance with theory or literature. As a result, the intermediate solution is closest to theoretical reality (Rihoux & Ragin, 2008). Consequently, we following the previous studies, focus on the intermediate solution for the interpretation of the final combination. A more comprehensive and detailed view of the findings can be obtained by combining the parsimonious solution and the intermediate solution. Causal conditions can be divided into the core condition and the peripheral condition. The former, existing a key meaning to the interpretation of the causality, coexist in the intermediate solution and the parsimonious solution, comparatively; the latter, no existing a key influence to the interpretation of the causality, merely appears in the intermediate solution (Fiss, 2011). Following the study of Ragin (2008), we also use the black circles (“●”) to represent the presence of a condition, and white circles (“○”) to represent its absence. Moreover, large circles indicate core conditions, and relatively small circles refer to peripheral conditions. Blank spaces in a solution indicate a “don’t care” situation in which this causal condition may be either present or absent.
The analysis of sufficiency generated five solutions, four configurations leading to high EPI and one configuration leading to low EPI. The findings are shown in detail in Table 4.
4.3.1 Cultural profiles – Antecedents of high environmental performance
The result shows that four casual recipes are associated with high EPI. The overall solution has a consistency level of 0.999, revealing that the degree to which the configurations led to high EPI is high. On the other hand, the coverage of the overall solution is 0.469, which indicates that almost 46% of the cases with a high level of EPI showed these four combinations of causal conditions. In addition, Table 4 also shows one casual recipe associated with low EPI. The solution consistency is 0.976 and coverage is 0.196. All five configurations demonstrate a high degree of consistency, and further, the coverage of each configuration provides evidence of its relative empirical importance. Therefore, the explanatory solution of high EPI consists of four different but equifinal configurations which are sufficient for leading to high EPI. The presence of solution 1 (1a and 1b) and 2 (2a and 2b) imply the existence of second-order in this study (Fiss, 2011). The following section describes each configuration specifically.
Solution 1a and Solution 1b, as a set of neutral permutations, have the same core conditions, namely, low power distance and individualism. Specifically, in solution 1a, indulgence appears as the peripheral condition. Masculinity, uncertainty avoidance, and long-term orientation appear as conditions of indifference; in other words, both masculinity and femininity, high and low uncertainty avoidance, as well as long-term and short-term orientation, may be integrative dimensions that are related to high EPI. This solution includes the USA, Denmark, Australia, the UK, etc. In solution 1b, uncertainty avoidance and long-term orientation both appear as peripheral conditions. Masculinity and indulgence are conditions of indifference. This solution includes Hungary, Estonia, Latvia, Germany, etc. Solution 1(1a and 1b) is essentially in accordance with the point of view about previous culture studies, which emphasize the importance of low power distance and individualism in generating high environmental performance. Low power distance means distribution of power, in which individuals are more likely to take part in the decision-making process and are more engaged in world issues, such as the environment; Individualism means autonomy, independence, and freedom. Emphasizing low power distance and individualism as its core conditions, solution 1 stresses their equal importance in generating the combined effect on high EPI.
Solution 2a and Solution 2b have the same core conditions, namely, individualism and indulgence. Specifically, in solution 2a, high certainty avoidance appears as the peripheral condition. Power distance, masculinity, and long-term orientation appear as conditions of indifference. This solution includes Finland, Malta, Iceland, Norway, etc. In solution 2b, masculinity, uncertainty avoidance, and long-term orientation appear as the peripheral condition. Power distance appears as a condition of indifference. This solution includes Austria, Luxembourg, Belgium, etc. Solution 2(2a and 2b) partially concurs with the point of view from previous cultural research which investigated the effects of independent cultural dimensions on EPI. Namely, this solution emphasizes the great importance of individualism and indulgence. In societies characterized by individualism and indulgence, individuals are more encouraged to make the environmental investment with a high degree of personal initiative and fewer constraints.
4.3.2 Cultural profiles – Antecedents of low environmental performance
The result also shows that one causal recipe is associated with low environmental performance. The level of consistency is 0.976. It points out the fact that the combination of high power distance, collectivism, masculinity, low uncertainty avoidance, short-term orientation, and restriction brings about a low level of environmental performance. This solution includes the Philippines, Malaysia, Mozambique, and Dominican, etc. This result concurs with the point of view from previous cultural studies, which identified high power distance, collectivism, and short-term orientation as inhibitors of environmental performance. As the only core condition leading to low environmental performance, restriction is considered to having an essential meaning.
[Insert Table 4 about here]
4.4 Robustness tests
The robustness test of QCA needs to stay true to fundamental principles and nature of set-theoretic methods instead of copying the robustness test of qualitative analysis. Following the recommendations of Schneider and Wagemann (2012), we conduct three robustness checks for our sufficiency analysis. Firstly, we modify the consistency threshold from the former value of 0.75 to 0.8 adopted in the fsQCA procedure. The solutions are completely unchanged. Secondly, we rerun our sufficiency analysis with a higher proportional reduction in the inconsistency of PRI≥0.75. The solutions are unchanged and parameters of fit remain similar. Finally, following the suggestion of Fiss (2011), we recalibrate the data of the outcome by setting the fully-in and fully-out cutoff points at the 95th percentile (80.585) and the 5th percentile (32.200), with the crossover point set to the median (57.650). However, to avoid the case ambiguity problem, we increase the crossover point by 0.001 and use the score equal to 57.651. Solution 1a, Solution 1b, and Solution 2b remain completely unchanged, while Solution 2a is absent from the sufficiency analysis. All of these modifications confirm the robustness of our original results.