Sağlığı Koruma Davranışı Ölçeği (SKDÖ)

The CVI of the scale items ranged between 0.84 and 1.00, indicating that all items were deemed appropriate by experts. To examine the construct validity of the Turkish version of the HPBS, EFA was conducted using Principal Axis Factoring with Varimax rotation. The KMO measure verified sample adequacy (KMO = 0.776), and Bartlett's test of sphericity was significant (χ2 (496) = 7702.990, p < 0.001). The factor structure was fixed according to the original instrument, making this a semi-confirmatory EFA. According to the results of the exploratory factor analysis, the developed scale explained 54.82% of the total variability. Convergent validity was supported by substantial factor loadings, with all items loading ≥ 0.40 on their intended factors and communalities ranging from 0.32 to 0.71, indicating that items strongly converged on their respective factors. Discriminant validity was supported by minimal cross-loadings, as all items loaded higher on their intended factors than on any other factor, and the five-factor structure corresponded to theoretically distinct dimensions of health protection behavior. The analysis revealed five factors consistent with the original scale: General Behavior, Interpersonal Relationship, Health Care, Self-Knowledge, and Nutrition. These five factors explained 54.82% of the total variance. According to Spearman correlation analysis, HPBS was compared with HLBS II and the WHOQOL-BREF scale test in terms of criterion validity. A positive, strongly significant correlation was found between HPBS and HLBS II (Rho = 0.506, p < 0.001) and a positive, moderately significant correlation was found between HPBS and WHOQOL-BREF (Rho = 0.644, p < 0.001). The validity of known groups in the study is based on linear regression analysis. In the linear regression model evaluating the relationship between selected variables and HPBS scores, the following variables were included: age, gender, marital status, educational status, occupation, income, chronic disease, social health perception, general health perception, smoking, alcohol use, healthy eating perception, sleep patterns, exercise, bodynmass index, daily screen time, time allocated for oneself, HLBS II, and WHOQOL-BREF scores. These variables in the model explained 50% of the variance in the HPBS scale (R2 = 0.709, Adjusted R2 = 0.503, F = 52.876, p < 0.001). In the backward method linear regression analysis, those with good social health perception had a lower risk of smoking compared to those with poor social health perception (β = −3.513, 95% CI: −5.54; −1.47). Nonsmokers had higher HPBS scores than smokers (β = 2.653, 95% CI: 0.21, 5.09). Participants with a good perception of healthy eating also had higher HPBS scores compared to those with a poor perception (β = −2.368, 95% CI: −4.20, −0.53). Furthermore, HPBS scores increased significantly with higher HLBS II (β = 0.199, 95% CI: 0.14, 0.25) and WHOQOL-BREF scores (β = 0.408, 95% CI: 0.31, 0.49). These findings further supported construct validity by demonstrating the ability of the HPBS to discriminate between theoretically distinct groups, consistent with expectations. Note: CFA results are not presented in this manuscript to avoid potential inflation of model fit indices due to using a single dataset for both derivation and validation. Future studies should conduct CFA using independent samples to further confirm convergent and discriminant validity.

The CVI of the scale items ranged between 0.84 and 1.00, indicating that all items were deemed appropriate by experts. To examine the construct validity of the Turkish version of the HPBS, EFA was conducted using Principal Axis Factoring with Varimax rotation. The KMO measure verified sample adequacy (KMO = 0.776), and Bartlett's test of sphericity was significant (χ2 (496) = 7702.990, p < 0.001). The factor structure was fixed according to the original instrument, making this a semi-confirmatory EFA. According to the results of the exploratory factor analysis, the developed scale explained 54.82% of the total variability. Convergent validity was supported by substantial factor loadings, with all items loading ≥ 0.40 on their intended factors and communalities ranging from 0.32 to 0.71, indicating that items strongly converged on their respective factors. Discriminant validity was supported by minimal cross-loadings, as all items loaded higher on their intended factors than on any other factor, and the five-factor structure corresponded to theoretically distinct dimensions of health protection behavior. The analysis revealed five factors consistent with the original scale: General Behavior, Interpersonal Relationship, Health Care, Self-Knowledge, and Nutrition. These five factors explained 54.82% of the total variance. According to Spearman correlation analysis, HPBS was compared with HLBS II and the WHOQOL-BREF scale test in terms of criterion validity. A positive, strongly significant correlation was found between HPBS and HLBS II (Rho = 0.506, p < 0.001) and a positive, moderately significant correlation was found between HPBS and WHOQOL-BREF (Rho = 0.644, p < 0.001). The validity of known groups in the study is based on linear regression analysis. In the linear regression model evaluating the relationship between selected variables and HPBS scores, the following variables were included: age, gender, marital status, educational status, occupation, income, chronic disease, social health perception, general health perception, smoking, alcohol use, healthy eating perception, sleep patterns, exercise, bodynmass index, daily screen time, time allocated for oneself, HLBS II, and WHOQOL-BREF scores. These variables in the model explained 50% of the variance in the HPBS scale (R2 = 0.709, Adjusted R2 = 0.503, F = 52.876, p < 0.001). In the backward method linear regression analysis, those with good social health perception had a lower risk of smoking compared to those with poor social health perception (β = −3.513, 95% CI: −5.54; −1.47). Nonsmokers had higher HPBS scores than smokers (β = 2.653, 95% CI: 0.21, 5.09). Participants with a good perception of healthy eating also had higher HPBS scores compared to those with a poor perception (β = −2.368, 95% CI: −4.20, −0.53). Furthermore, HPBS scores increased significantly with higher HLBS II (β = 0.199, 95% CI: 0.14, 0.25) and WHOQOL-BREF scores (β = 0.408, 95% CI: 0.31, 0.49). These findings further supported construct validity by demonstrating the ability of the HPBS to discriminate between theoretically distinct groups, consistent with expectations. Note: CFA results are not presented in this manuscript to avoid potential inflation of model fit indices due to using a single dataset for both derivation and validation. Future studies should conduct CFA using independent samples to further confirm convergent and discriminant validity.

The CVI of the scale items ranged between 0.84 and 1.00, indicating that all items were deemed appropriate by experts. To examine the construct validity of the Turkish version of the HPBS, EFA was conducted using Principal Axis Factoring with Varimax rotation. The KMO measure verified sample adequacy (KMO = 0.776), and Bartlett's test of sphericity was significant (χ2 (496) = 7702.990, p < 0.001). The factor structure was fixed according to the original instrument, making this a semi-confirmatory EFA. According to the results of the exploratory factor analysis, the developed scale explained 54.82% of the total variability. Convergent validity was supported by substantial factor loadings, with all items loading ≥ 0.40 on their intended factors and communalities ranging from 0.32 to 0.71, indicating that items strongly converged on their respective factors. Discriminant validity was supported by minimal cross-loadings, as all items loaded higher on their intended factors than on any other factor, and the five-factor structure corresponded to theoretically distinct dimensions of health protection behavior. The analysis revealed five factors consistent with the original scale: General Behavior, Interpersonal Relationship, Health Care, Self-Knowledge, and Nutrition. These five factors explained 54.82% of the total variance. According to Spearman correlation analysis, HPBS was compared with HLBS II and the WHOQOL-BREF scale test in terms of criterion validity. A positive, strongly significant correlation was found between HPBS and HLBS II (Rho = 0.506, p < 0.001) and a positive, moderately significant correlation was found between HPBS and WHOQOL-BREF (Rho = 0.644, p < 0.001). The validity of known groups in the study is based on linear regression analysis. In the linear regression model evaluating the relationship between selected variables and HPBS scores, the following variables were included: age, gender, marital status, educational status, occupation, income, chronic disease, social health perception, general health perception, smoking, alcohol use, healthy eating perception, sleep patterns, exercise, bodynmass index, daily screen time, time allocated for oneself, HLBS II, and WHOQOL-BREF scores. These variables in the model explained 50% of the variance in the HPBS scale (R2 = 0.709, Adjusted R2 = 0.503, F = 52.876, p < 0.001). In the backward method linear regression analysis, those with good social health perception had a lower risk of smoking compared to those with poor social health perception (β = −3.513, 95% CI: −5.54; −1.47). Nonsmokers had higher HPBS scores than smokers (β = 2.653, 95% CI: 0.21, 5.09). Participants with a good perception of healthy eating also had higher HPBS scores compared to those with a poor perception (β = −2.368, 95% CI: −4.20, −0.53). Furthermore, HPBS scores increased significantly with higher HLBS II (β = 0.199, 95% CI: 0.14, 0.25) and WHOQOL-BREF scores (β = 0.408, 95% CI: 0.31, 0.49). These findings further supported construct validity by demonstrating the ability of the HPBS to discriminate between theoretically distinct groups, consistent with expectations. Note: CFA results are not presented in this manuscript to avoid potential inflation of model fit indices due to using a single dataset for both derivation and validation. Future studies should conduct CFA using independent samples to further confirm convergent and discriminant validity.