Reliability and Predictive Validity of the Adaptive Health Behavior Inventory (AHBI): Adaptive Health Behavior Differences by Gender and Age
Summary
Objective: Describe research demonstrating that the Adaptive Health Behavior Inventory (AHBI) can detect and predict a range of adaptive health behavior differences between men and women and as adults age. Methodology: A cross-sectional analysis of AHBI response data collected from four surveys of adults at different times and geographies (Sample 1: 2001 national telephone-mail survey, n = 20,685) and three on-line surveys conducted in 2017 and 2018 (Sample 2: Baltimore–Washington DC, n = 2,002; Sample 3: Atlanta, Georgia, n = 2,000); Sample 4: Cincinnati, Ohio, n = 2,000). Analysis: Logistic regression assessed AHBI predictive effects identifying male versus female status controlling for age. Multiple regression assessed the AHBI predictive effects in identifying adaptive health behavior differences associated with aging controlling for sex. Results: The overall R-squared, chi square (χ2), and odds ratios for all four logistic regression analyses were statistically significant predicting gender: Sample 1, χ2(1, 21) = 183.7, p < .001, R-squared = .16; Sample 2, χ2(1, 21) = 42.4, p < .001, R-squared = .31; Sample 3, χ2(1, 21) = 28.7, p < .001, R-squared = .23; Sample 4, χ2(1, 21) = 28.7, p < .001, R-squared = .23. The mean R-squared for gender across all four samples represents a mean effect size of F-squared = .30. Sixty percent (60%) of AHBI measures demonstrated various levels of predictive reliability identifying gender differences. The overall R-squared, F, and β coefficients for all four multiple regression analyses were statistically significant predicting age: Sample 1, F(1, 21) = 183.7, p < .001, R-squared = .16; Sample 2, F(1, 21) = 42.4, p < .001, R-squared = .31; Sample 3, F(1, 21) = 28.7, p < .001, R-squared = .23; Sample 4, F(1, 21) = 38.5, p < .001, R-squared = .29. The mean R-squared of Samples 2 through 4 was .26, represents a mean effect size of F-squared = .35. Like with gender, sixty percent (60%) of AHBI measures demonstrated various levels of predictive reliability identifying lower or higher age. Discussion & Conclusions: Multiple measures within the AHBI demonstrate good and excellent predictive reliability in discriminating between women and men as well as younger versus older adults based on differences in adaptive health behavior.
Objective: Describe research demonstrating that the Adaptive Health Behavior Inventory (AHBI) can detect and predict a range of adaptive health behavior differences between men and women and as adults age. Methodology: A cross-sectional analysis of AHBI response data collected from four surveys of adults at different times and geographies (Sample 1: 2001 national telephone-mail survey, n = 20,685) and three on-line surveys conducted in 2017 and 2018 (Sample 2: Baltimore–Washington DC, n = 2,002; Sample 3: Atlanta, Georgia, n = 2,000); Sample 4: Cincinnati, Ohio, n = 2,000). Analysis: Logistic regression assessed AHBI predictive effects identifying male versus female status controlling for age. Multiple regression assessed the AHBI predictive effects in identifying adaptive health behavior differences associated with aging controlling for sex. Results: The overall R-squared, chi square (χ2), and odds ratios for all four logistic regression analyses were statistically significant predicting gender: Sample 1, χ2(1, 21) = 183.7, p < .001, R-squared = .16; Sample 2, χ2(1, 21) = 42.4, p < .001, R-squared = .31; Sample 3, χ2(1, 21) = 28.7, p < .001, R-squared = .23; Sample 4, χ2(1, 21) = 28.7, p < .001, R-squared = .23. The mean R-squared for gender across all four samples represents a mean effect size of F-squared = .30. Sixty percent (60%) of AHBI measures demonstrated various levels of predictive reliability identifying gender differences. The overall R-squared, F, and β coefficients for all four multiple regression analyses were statistically significant predicting age: Sample 1, F(1, 21) = 183.7, p < .001, R-squared = .16; Sample 2, F(1, 21) = 42.4, p < .001, R-squared = .31; Sample 3, F(1, 21) = 28.7, p < .001, R-squared = .23; Sample 4, F(1, 21) = 38.5, p < .001, R-squared = .29. The mean R-squared of Samples 2 through 4 was .26, represents a mean effect size of F-squared = .35. Like with gender, sixty percent (60%) of AHBI measures demonstrated various levels of predictive reliability identifying lower or higher age. Discussion & Conclusions: Multiple measures within the AHBI demonstrate good and excellent predictive reliability in discriminating between women and men as well as younger versus older adults based on differences in adaptive health behavior.