Understanding Constraints and Facilitators to Salmon Angling Participation: Insights from Structural Equation Modeling

ABSTRACT

Atlantic salmon sport fisheries have declined remarkably in many countries and participation seems to correlate with salmon abundance. We investigated angling participation in the Atlantic salmon sport fishery in Norwegian rivers by incorporating facilitators for participation in a constraint–negotiation model. We conducted an Internet survey of Norwegian anglers yielding 3,635 responses (40% response rate). The structural model confirmed our hypotheses and supported the conceptual constraints-effects-mitigation model of leisure constraint negotiation. Of the constraints and facilitators investigated, the structural constraints and facilitators subcategory “quality of fishing” exerted the largest influence on angling participation. The influence of constraints and facilitators was mitigated by use of corresponding negotiation strategies where “skills, knowledge, and money,” and different substitution strategies were important. To increase participation, we suggest increasing salmon abundance, offering longer fishing stretches per angler, and providing better information about where to book salmon angling.

KEYWORDS:

• Constraint negotiation
• facilitators
• fishing tourism
• motivations
• structural equation modeling

Acknowledgments

The article was significantly improved by comments from two reviewers and the Co-Editor, which we sincerely appreciate.

Funding

This study is part of the project SALMONCHANGE and was supported by the Research Council of Norway under grant number 208056, and the Norwegian Environment Agency under grant number 2013/1686-21052013.

Notes

1. Given this study uses a constraint–facilitator continuum (from negative to positive) H8 assumes a positive effect.

2. 80—86% of anglers paying the fee any of the years 2012–2014 did so online.

3. Bandura (1997) also includes a fourth type of self-efficacy: physiological and affective states. We had included this in the questionnaire draft. Comments from pretesting about the wording and strangeness of this type as expressed in the Norwegian language made us eliminate it and use our alternative measure consisting of three sub-categories of self-efficacy.

4. We additionally estimated our model using the non-normality robust standard errors (Satorra-Bentler) in Stata. The results did not differ substantially from those obtained from the standard maximum likelihood estimation. Thus, we have chosen to report the results from the latter estimation.

5. The amount of missing data on the observed variables varied between 1% and 22%. PMS was used as it is shown to be a good representation of the original data when the missing data are less than about 20% (Downey & King, 1998).

6. Given that not all readers are accustomed to Raykov’s factor rho coefficients, we also provide Cronbach’s Alpha coefficients in Table 1.

7. There is a clear warning in the SEM literature against accepting/rejecting a model solely based on model fit thresholds. Thus, we first interpreted the parameter estimates according to our theoretical assumptions. Once the parameter estimates made sense, we examined the fit measures. The model fit measures were shown to be acceptable.

8. Note that the range interval is truncated because anglers younger than 16 years (from 2013 on 18 years) do not pay the fee, and therefore are not present in the register. In the survey 80+ years was set as the upper age alternative.

9. A beat is defined as a length of river or bank, let or fished as a unit by angling (McLay & Gordon-Rogers, 1997).

Additional information

Funding

This study is part of the project SALMONCHANGE and was supported by the Research Council of Norway under grant number 208056, and the Norwegian Environment Agency under grant number 2013/1686-21052013.