**3. Materials and Methods**

The questionnaire consisted of three parts. The first part was related to sociodemographic characteristics of respondents: gender, age, education, and marital status. The second part of the questionnaire was referring to identifying the phase in behavior change. It was performed by adapting formulations from previous research [24] to the topic of this research. Hereby, the respondents were offered to choose one of the following options: "So far, I've never thought about staying at a rural green hotel when traveling," "I have considered staying at a rural green hotel when traveling. However, I haven't put this plan into practice yet," "For my last travel I have stayed at a rural green hotel. It is my firm intention to do this in the future," or "For me it is a given to stay at a rural green hotel when traveling." According to the choice of one of the selected options, respondents were segmented into four phases of behavior change: pre-decision, pre-action, action, and post-action phase, respectively. The third part of the questionnaire was related to measuring variables from TPB (on a five-point Likert scale). That was performed by adapting items from previous research [32].

Personal communication with respondents from the Republic of Serbia was performed. The convenience sampling method was implemented. In order to try to interview the average user of tourist services, the respondents were approached at several of the largest Serbian towns near famous shopping centers during all seven days of one week. The precondition for a person to become a respondent was that he/she was using hotel services during 12 months before the interview. The questionnaire was either read to the respondent and he/she provided answers to the interviewer, or the respondent personally filled in the questionnaire in the presence of the interviewer. The questionnaires with no answer regarding phase in behavior change, or on more than one item referring to latent variables from TPB (43 in total), were excluded. In the case of missing only one item for some of the variables from TPB, it was filled in by a mean of other items belonging to the same latent variable. Data were processed in 2019. It included analysis of 289 answers.

The sample consisted of 48.8% males and 51.2% females. Hereby, the average respondent was 37.05 years old (standard deviation 10.21). When it comes to education, 45.3% of the respondents had finished secondary school, 7.6% of the respondents were students, and 47.1% had finished college or faculty. As for marital status, 63.3% of the respondents were married and 36.7% were single. Furthermore, 23.5% of the respondents were in the first, 41.9% in the second, 27.7% in the third, and 6.9% in the fourth phase of behavior change.

When analyzing the influences of TPB elements on intention to visit green rural hotels in different phases, multigroup structural equation modelling was performed. Hereby, the respondents in one of the first three phases were taken into account—as in [24]. Prior to performing multigroup SEM, the questionnaire was tested.

#### **4. Results**

#### *4.1. Testing the Questionnaire*

As all four constructs from TPB are reflective, individual indicator reliability, internal consistency reliability, convergent validity, and discriminant validity were examined [41]. For analyzing indicator reliability, standardized loading for each indicator was checked. As expected, they were higher than 0.7 (Table 1).


### **Table 1.** Quality criteria of the reflective constructs.

Satisfactory levels were obtained in the case of internal consistency reliability and convergent validity, as well. Hereby, the values of CR (composite reliability) and AVE (average variance extracted) were above 0.7 and 0.5, respectively [42,43].

For assessing discriminant validity, the Fornell–Larcker criterion was applied (Table 2).


**Table 2.** Discriminant validity assessment: Fornell–Larcker criterion.

Hereby, each construct's square root of AVE is higher than its correlations with other constructs [41], which confirms discriminant validity.

Having in mind that all variance inflation factor (VIF) values for all latent variables are lower than 3.3, "the model can be considered free of common method bias" [44] (p. 7).

#### *4.2. Testing the Hypotheses*

The effects of independent variables on Intention to visit green rural hotel have been analyzed by using PLS-SEM path coefficients. Table 3 presents their values for the entire model and for each of three phases as well. Furthermore, the R2 value equaled 0.311.


**Table 3.** Path coefficients.

At the level of the entire model, significant positive effects have been recorded for all three independent constructs, attitudes, subjective norms, and perceived behavioral control (0.327, 0.189, and 0.259, respectively). When it comes to phases, attitudes construct had a significant positive effect only in the second phase (0.643), subjective norms in the first (0.787) and perceived behavioral control in the third phase (0.891). Differences in path coefficients between phases have been tested by the implementation of Multi-group Analysis (MGA)—Table 4.


Perceived behavioral control 0.109 0.793 \* 0.902 \*

**Table 4.** Multi-group analysis (MGA)—coefficients differences.

Hereby, in relation to attitudes construct, the path coefficient in phase 2 is significantly higher than coefficients in phases 1 and 3. In the case of subjective norms, the coefficient in phase 1 is significantly higher than coefficients in phases 2 and 3. Finally, in relation to perceived behavioral control, the path coefficient in phase 3 is significantly higher than coefficients in phases 1 and 2.

#### **5. Discussion and Conclusions**

\* p < 0.05 or p > 0.95

From the research conducted within this paper, important implications can be derived for both theory and practice. When it comes to the theoretical contribution, it should be stressed that, according to the authors' knowledge, this research is the first dynamic approach to TPB in the context of green rural hotels. That presents the key contribution of this paper. The special significance of the topic is that it belongs to one of the two identified methodological issues regarding the TPB in recent studies—dynamic approach and asymmetrical modelling.

The results show that when the model is performed for all the respondents, the influence of each of the independent variables (attitudes, subjective norms, perceived behavioral control) on dependent variables is statistically significant. That is in accordance with almost all previous researches that implement SEM in general. Moderation analysis is rarely present [20,36] or fails to provide significant differences in those researches [20]. However, when moderation analysis is performed within this research, based on the phase in behavior change (pre-decision, pre-action, action), it can be concluded that the significant influence of all the variables is not present in each of the phases. In the first phase, there is a significant influence of only subjective norms on intention to visit green rural hotels; in the second phase, solely attitudes significantly affect the dependent variable; while in the third phase, perceived behavioral control is the only significant predictor of green rural hotels choice.

In the previous context can be discussed the confirmation of hypotheses. It should be noted that the first hypothesis is partially confirmed since intention to visit a green rural hotel is in the first phase of tourists' behavioral change (pre-decision), influenced by subjective norms, but not by perceived behavioral control. A similar situation can be observed regarding the second hypothesis, as well. Hereby, intention to visit a green rural hotel is in the second phase of tourists' behavioral change (pre-action), influenced by attitudes, but it is not affected by perceived behavioral control. The third hypothesis is confirmed since intention to visit a green rural hotel is in the third phase of tourists' behavioral change (action), influenced by perceived behavioral control. The fourth hypothesis is partially confirmed. When comparing the third phase of behavior change to previous phases, the influence of perceived behavioral control on intention to visit a green rural hotel is the strongest and statistically significant. However, in previous phases, influences of that variable are neither statistically significant, nor is there an increase in the second phase in comparison to the first.

The difference from the expectations regarding influence of independent variables is noticed in previous research as well [24]. From the results of this research, it can be concluded that, when a person has never thought to stay at a rural green hotel when traveling, the greatest influence on intention to visit such a hotel is the perception of opinion of people who are considered important. Furthermore, for people who have considered staying at a rural green hotel when traveling, but have not put that plan into practice yet, the greatest influence is the one of their own attitudes. Finally, for respondents that are already in the phase of action, the only importance is the perception of their own ability to perform such behavior. It was hypothesized that such influence would be important in previous phases as well, but the results did not support it. Two explanations can be given regarding such results. Firstly, because of the underdeveloped offer of such hotels in domestic conditions, that factor might become important to respondents only when they face the actual visit to them. Within the research, the questions were formulated in general, so it cannot be examined whether respondents in the action phase are referring to the domestic offer of green rural hotels or from such hotels abroad. Furthermore, the decision regarding visiting hotels is usually made much faster than, for example, the decision to buy a hybrid vehicle or to move into an energy-efficient home. That can be an additional explanation why the perception of own ability to perform a behavior does not become important before the action phase.

The practical recommendations can be given for different segments obtained in this research based on the phase in behavior change. All these recommendations can be of special importance to sustainability; as it is proven within the paper, the significance of rural tourism and green hotels in that context. For people who have never thought to stay at a rural green hotel when traveling, influencing persons can be relied on. Additional research is necessary to discover persons who might be considered as relevant for a great part of respondents from that segment and, if using massive communication, it might be chosen to cooperate with some of the celebrities. Hereby, attention should be dedicated to those people who like to travel and act socially responsible. They can be engaged in

promotional campaigns, in which the emphasis should be on rural areas and their natural beauties. For this purpose, social networks, especially Instagram, could be of great help. For the segment of respondents who have considered staying at a rural green hotel when traveling, but have not put that plan into practice yet, it is of the greatest importance to influence their positive beliefs about performing such actions as well as their evaluations. A campaign containing information that would help in accomplishing that goal would be recommendable. Thus, it may include some information related to advantages of rural areas, such as natural and healthy environment, less noise, clean air, and domestic food. Besides presenting the countryside as a place for rest and relaxation, attention should be dedicated to the ethical aspect as well. Hereby, potential tourists should be informed about the sustainability concept, i.e., environmental protection and all benefits for the social community in rural areas. When considering the segment of respondents in the action phase, it is important to increase their perception of their ability to stay at green rural hotels. In addition to their consideration of the resources and time needed to visit rural green hotels, support should also be provided to the supply side of the market. Hereby, it would be useful to provide them with information about offers of such hotels in domestic conditions, about the best routes to approach them, and the necessary time and costs when performing such actions. For example, popular TV shows presenting routes to popular destinations for summer vacations could also be used to transmit previously listed information.

Future researches could use larger and more representative samples. Additional moderations considering whether the destination is domestic or foreign can be included. Rural tourism offer can be considered without focusing only on green hotels. Finally, asymmetrical modelling could be implemented on the data of this research as well, in order to obtain an even deeper understanding of TPB functioning. Beside elements of TPB, that analysis could also integrate demographic data.

**Author Contributions:** All of the authors formulated goals of the research and interpreted available literature; conducting and analyzing research was performed by N.M. and N.D., while implications were developed by A.G.

**Funding:** This research received no external funding

**Conflicts of Interest:** The authors declare no conflict of interest.
