Retooling the Traditional Approach to Studying the Belief–Attitude Relationship: Explaining Landowner Buy-In to Incentive Programs

Abstract

Perceived threat from traditional command-and-control implementation of the U.S. Endangered Species Act has resulted in the underprovision of endangered species habitat on private lands. Voluntary incentive programs aid recovery by reducing landowner costs and increasing benefits; however, program efficacy is directly related to participation. I used the reasoned action framework to examine landowner attitudes and beliefs about participating. Using data from a mail survey of Texas landowners, I employed two seldom-used methods (structure coefficients and commonality analysis) to enhance explanation from a regression of attitudes on behavioral beliefs. Two of eight measured beliefs (beneficial outcomes for an individual's land and for the target species) accounted for the majority of the variation in attitude. Landowner concerns about future regulations and technical assistance contributed least to attitude. Results provide program administrators with necessary information to create targeted messages to increase participation and thus enhance program efficacy. The use of structure coefficients and commonality analysis highlighted relationships with the greatest explanatory power, as well as those that would have been otherwise missed or marginalized. I recommend these methods be employed in all reasoned-action studies analyzed using multiple regression.

Keywords:

• commonality analysis
• endangered species
• incentive programs
• private lands
• reasoned action
• regression
• structure coefficients

Acknowledgments

This project was supported by the Department of Defense, Office of the Secretary of Defense and the Texas A&M Institute of Renewable Natural Resources. Special thanks to J. R. Conner, who provided review comments on an early draft. G. Luikart and C. Ratheal assisted with data collection, B. Collier assisted with statistical programming, and A. Snelgrove provided GIS support. I also thank J. Sell, R. N. Wilkins, and R. B. Ditton for project support.

Notes

^a Confidence intervals were generated using bootstrapping.

Note. See Table 1 for definitions of belief variables.

*p < .10. **p < 0.05. ***p < .01.

Note. R = .69, R ² = .48, adjusted R ² = .46. See Table 1 for definitions of belief variables. For commonality analysis, Unique is the R ² partition uniquely attributed to the predictor variable, and Common is the partition of the R ² the predictor variable holds in common with other predictor variables; Total is the R ² partition for which the predictor variable has some predictive role (unique variance + common variance).

*p < .05. **p < .01.

Nimon currently is finalizing an SPSS script for this purpose. Contact her for more information: [email protected].