Does recreation specialization affect birders’ travel intention?

ABSTRACT

Using data from a survey of Italian birdwatchers, we examined whether recreation specialization affects birders’ travel intention through a two-dimensional framework based on the “behavior” and “skills, knowledge, and commitment” constructs. The model was estimated through a partial least squares structural equation “spread” model. We implemented a second-stage analysis, using a seemingly unrelated regression (SURE) model to identify which birder characteristics, attitudes, and preferences significantly affected the path scores. The findings demonstrated a significant and positive relationship between recreation specialization and birders’ travel intention, and offer evidence that birders’ behavior and skills, knowledge, and commitment were statistically significant lower hierarchical order constructs of recreation specialization. The intensity of these connections varied according to the birder’s profile, the source of information used to choose the destination site and the reasons behind the choice of site for birdwatching.

KEYWORDS:

• Birdwatching
• recreation specialization
• travel intention
• PLS-SEM
• SURE model

Acknowledgments

The authors wish to thank Roberto Lardelli of Ornitho and Luciano Ruggeri of EBN Italia for their support in the administration of the survey. Further, they would like to thank also the editor and the anonymous reviewers for their helpful comments and valuable suggestions.

Notes

1. Moderation emerges when the effect of one variable on another variable depends on the value of a third variable. In contrast, mediation occurs when the effect of one variable on another variable is mediated by the presence of one or more variables.

2. PLS-SEM, as well as the most popular CB-SEM, is a second-generation statistical model to analyze complex relationships among constructs. In Appendix B of the online supplementary materials, we illustrate the main differences between the two SEM models and offer a short presentation of PSL-SEM.

3. This dilemma results from the trade-off between the use of measures that cover the majority of variation in a trait (domain-level measurement) and the use of measures that assess a few specific traits (facet-level measurement) more precisely (Hair et al., 2018).

4. This phenomenon describes the influence that two scales called with different names measure different constructs (Hair et al., 2018).

5. This two-dimensional model was supported by a preliminary factorial analysis. We unified skills and knowledge and commitment, and analyzed both as a unique construct.

6. The minimum sample size in PLS-SEM should be equal to 10 times the largest number of formative indicators used to measure a single construct, or 10 times the largest number of structural paths directed at a particular construct in the structural model.

7. In Appendix C of the online supplementary materials, we present the criteria and statistics used to evaluate the performance of the PLS-SEM.