Publication Date

2019

Abstract

Objectives: To demonstrate the use of machine-learning for reducing questionnaire response burden, we created multiple, shorter versions of the Mindfulness Inventory for Sport. We then tested the reliability and validity of scores derived from these shorter versions in athletic populations. Design: We used genetic algorithms to shorten the measure, and both cross-sectional and longitudinal data to test psychometric properties. Method: We collected data from 859 undergraduate exercise science students and 118 golfers. We used 75% of the student sample to shorten the measure, and the rest of the data to test the internal consistency, test-retest reliability, content validity, and factorial validity. For criterion validity, we explored relationships between the subscales and other measures of mindfulness, golf handicaps, and an objective measure of putting accuracy. Results: Genetic algorithms efficiently generated stable solutions to shortening the measure. Reliability decreased as the measure become shorter—especially between three and two items per subscale—but remained acceptable. Validity metrics for shorter versions were as good, and sometimes better, than the full questionnaire. Awareness and refocusing subscales demonstrated weak associations with golf handicap for long and short versions. Non-judgment showed no significant associations, and no subscales significantly predicted putting performance. Conclusions: Genetic algorithms provide efficient solutions to reducing questionnaire response burden for athletes.

School/Institute

Institute for Positive Psychology and Education

Document Type

Journal Article

Access Rights

ERA Access

Access may be restricted.

Share

COinS