Table 1 presents the basic demographic characteristics of the respondents. Each scale was assessed for reliability and validity in the first step of the analysis. Table 2 gives the CFA results and includes the scales and the items. The Cronbach’s α coefficient values for the PRO, PES, PUS, and PNE scales were 0.687, 0.789, 0.628, and 0.702, respectively, showing them to be sufficiently reliable and valid.
Table 1
Characteristic features of the respondents (N = 420)
Variables
|
Frequency (%)
|
Variables
|
Frequency (%)
|
Gender
Male
Female
|
298 (71)
122 (29)
|
Monthly income (US$)
500 and below
|
292 (69,5)
|
Age (years)
15–34
35–49
> 50
|
138 (32,9)
129 (30,7)
153 (36,4)
|
501–1000
1001–2000
More than 2001
Occupation
|
70 (16,7)
31 (7,4)
27 (6,4)
|
Educational level
Middle school and below
High school
Graduate school
|
296 (70,5)
98 (23,3)
26 (6,2)
|
Farmer
Housewife
Self-employed
Other (student, worker, etc.)
|
249 (59,3)
83 (19,8)
64 (15,2)
24 (5,7)
|
Table 2
The results of the confirmatory factor analysis with scales and items.
Scales and items
|
Factor Load
|
t-value
|
Cronbach’s α
|
Perceived Revenue Opportunities (PRO) Scale
|
|
|
0.687
|
PRO1: Tourists coming to the region provide income opportunities.
|
0.765
|
12.502
|
|
PRO2: I generate income by providing guidance services to tourists coming to the region.
|
0.828
|
12.636
|
|
PRO3: I earn income by selling local products to tourists coming to the region.
|
0.665
|
-
|
|
Perceived Ecotourism Satisfaction (PES) Scale
|
|
|
0.789
|
PES1: I am satisfied with the tourists coming to the region.
|
0.606
|
-
|
|
PES2: I want more tourists come to the region.
|
0.826
|
5.960
|
|
Perceived Insufficient Services (PUS) Scale
|
|
|
0.628
|
PUS1: Infrastructure facilities for tourists are insufficient.
|
0.775
|
3.866
|
|
PUS2: I feel inadequate in communicating with the tourists coming to the area.
|
0.553
|
-
|
|
Perceived Negative Effects (PNE) Scale
|
|
|
0.702
|
PNE1: Tourists coming to the region adversely affect the lifestyle of the local people.
PNE2: Tourists coming to the region cause environmental pollution.
PNE3: Tourists coming to the area harm the traditions of the local people.
PNE4: Sometimes there is a dispute with the tourists coming to the region.
|
0.404
0.494
0.801
0.678
|
-
6.564
6.692
6.798
|
|
The path diagram presents the relationships between PES and PRO, where indicator variables are shown as rectangles and latent variables as ovals. A regression coefficient is indicated by each arrow from one latent or observed variable to another (Fig. 2). Each oval with an arrow to a latent or observed variable indicates the error term. Double-pointed arrows pointing between error terms indicate the covariances (Gültekin 2022). In the model, there were three observed, endogenous variables (PRO1-PRO3) and two observed variables (PES1-PES2), i.e., five observed variables. The eight unobserved variables included the endogenous variable (PES) and the unobserved, exogenous variables (PES and e1-e6). Thus, the total number of variables in the model was thirteen. The GOF values of the data were acceptable in the details of the hypothesized path model (DF = 4, CMIN = 17.102, p < 0.002, RMSEA = 0.088, RMR = 0.028, CMIN/DF = 4.275, NFI = 0.971, GFI = 0.984, TLI = 0.944, CFI = 0.977, and IFI = 0.978).
The data (β = 0.39, p < 0.002) supported Hypothesis 1 (“PRO is positively associated with PES”), i.e., that the PRO of the forest villagers significantly affected their PES. The PRO scale items state that the satisfaction level of rural people is highly correlated with their perception of “generating income by selling local products, earning money by providing guidance services, and providing income opportunities”. The responses of the forest villagers indicated that they were not adequately satisfied with the ecotourism opportunities in the region (PES1 mean = 3.68; PES2 mean = 3.95).
After the first stage of the SEM model, the complete multidimensional model and indirect effects of PNE and PUS were set up to achieve the mediator effects and the relationships between PRO, PES, PNE and PUS. The complete structural model can be seen in Fig. 3. There were a total of 29 variables in the model. The observed, endogenous variables in the model were PRO1-PRO3 (3 observed variables), PNE1-PNE4 (4 observed variables), PUS1-PUS3 (2 observed variables), and PES1-PES2 (2 observed variables). There were a total of 11 observed variables. The unobserved endogenous variables were PUS, PNE, and PES and the unobserved exogenous variables were PRO and e1-e14 (a total of 15 unobserved variables). The values of the full structured model for the model fit criteria were found as: DF = 38, CMIN = 68.350, p < 0.002, CMIN/DF = 1.799, RMR = 0.045, RMSEA = 0.044, GFI = 0.972, CFI = 0.971, NFI = 0.937, TLI = 0.958, and IFI = 0.971. These indices were used in the determination of the efficiency of the SEM models and indicated that the hypothesized path model fit was acceptable (Kline 2015). When the PUS and PNE were considered as mediators and included in the structural models, it confirmed that the relationship between the PRO and PES differed in the regression coefficient. Moreover, with the suppression effect, the regression coefficient was reduced from 0.39 to 0.32 (Agler and De Boeck 2017), i.e., when the complete model was regarded as being significant for the PRO and PES of the forest villagers, the PUS and PNE were highly effective. Table 3 presents the complete multidimensional model and indirect effects of PUS and PNE, and the total effects between PES (path c) and PRO. The estimates (bias-corrected 95% CI) of the indirect effects calculated via the bootstrap analysis using 5,000 replications revealed the significant mediator effects of the relationship between PES and PRO and PUS and PNE (β = 0.073, lower bound = 0.007, upper bound = 0.170). Significant correlations were also found between PES and PRO and PUS (paths a, b, d, and e) and PNE. When the responses of the forest villagers to the expressions on the PRO scale were examined, they would positively affect their PUS included: “lack of infrastructure and communication problems with the tourists. Nevertheless, the PNE scale drew some negative perceptions for forest villagers such as adversely affecting the lifestyle of the local people, environmental pollution concerns, and sometimes having conflicts”. The PRO-PUS-PNE-PES model indicated that forest villagers were quite attached to ecotourism components, even though there were some negative perceptions on their PRO and PES. For example, the PES scale suggested that “Rural people want to see more tourists in the region and they are satisfied with tourists coming to the region”. These expressions are particularly effective in understanding the PES level of the forest villagers.
Table 3
Results of complete mediation models.
|
Result variables
|
PES
|
PNE
|
PUS
|
β
|
|
SE
|
β
|
|
SE
|
β
|
|
SE
|
PRO (path c)
|
0,244**
|
|
0,064
|
|
|
|
|
|
|
R2
|
|
0,317
|
|
|
|
|
|
|
|
PRO (path a)
|
|
|
|
-0,231**
|
|
0,051
|
|
|
|
R2
|
|
|
|
|
0,126
|
|
|
|
|
PRO (path d)
|
|
|
|
|
|
|
0,229**
|
|
0,071
|
R2
|
|
|
|
|
|
|
|
0,079
|
|
PNE (path b)
PUS (path e)
|
-0,104*
0,213**
|
|
0,082
0,075
|
|
|
|
|
|
|
PRO (path c')
|
0,244**
|
|
0,064
|
|
|
|
|
|
|
R2
|
|
0,222**
|
|
|
|
|
|
|
|
Indirect effect
|
0,073** (0,007 − 0,170)
|
|
|
|
|
*p < 0,05, **p < 0,001, SE: Standard error. |
According to the PNE model “H2: Perceived Revenue Opportunities are negatively associated with Perceived Negative Effects” is supported by the data (β = 0.529, p < 0.001). Hypothesis 2 predicted the negative direct effect of PRO on PNE. This can be explained as forest villagers’ PRO reduces their PNE. The following hypotheses were also supported according to the PNE and PUS models: “H3: Perceived Negative Effects are negatively associated with Perceived Ecotourism Satisfaction”, “H4: Perceived Insufficient Services play a mediating role on PES in the effect of Perceived Revenue Opportunities. H5: Perceived Revenue Opportunities negatively affect Perceived Insufficient Services. H6: Perceived Insufficient Services negatively affect Perceived Ecotourism Satisfaction. H7: Perceived Insufficient Services negatively mediate the relationship between Perceived Revenue Opportunities and Perceived Ecotourism Satisfaction.”. In other words, at 95% CI, the indirect effect estimation was 0.073 and 0.007–0.170 between PRO and PES. This effect was significant. Hypothesis 3 predicted a negative direct effect of PNE on PES. Hypothesis 4 estimated a positive indirect effect of PUS on PRO and PES. Hypothesis 4 estimated a negative indirect effect of PNE on PES. However, Hypothesis 7 estimated a negative indirect effect of PUS on PES. These indirect effects implied that forest villagers exhibited both negative and positive dimensions, even if their PES level was positive.
Table 3 presents the complete multidimensional model and the indirect effects of PUS and PNE. The indirect effects and mediation model of the original sample were computed as well as the estimated parameters of the β bootstrap samples. For the upper and lower bounds, 5000 samples from the original sample size were used to achieve improved results at a 95% confidence interval (CI).
The total effects between PRO and PES can be seen in Table 3 (path c). The relationship between PRO and PES showed that the PNE and PUS had a mediating effect that was statistically significant (β = 0.073, lower bound = 0.007, upper bound = 0.170). Significant relationships were also found between PRO and PNE, PRO and PUS, PNE and PES, and PUS and PES (path a, b, d and e).