| Marginal effectsa | [95% conf. interval] |
---|
Wrist ranking |
 Likelihood of use | 0.2 | [− 0.2 0.6] |
 Perceived accuracy | 0.1 | [− 0.4 0.5] |
 Ease of use | − 1.1 | [− 2.1 0.0] |
 Comfort | 0.7 | [− 0.2 1.5] |
Arm ranking |
 Likelihood of use | 0.1 | [− 0.4 0.5] |
 Perceived accuracy | 0.2 | [− 0.4 0.8] |
 Ease of use | − 0.5 | [− 1.2 0.3] |
 Comfort | 0.5 | [0.1 1.5] |
Choice of wrist cuff | 0.2 | [− 0.1 0.5] |
Age | − 0.9 | [− 1.7 0.0] |
Education |
 Some college/college | 0.0 | [− 0.2 0.2] |
Income |
 35–59,999 | 0.0 | [− 0.2 0.2] |
 60,000+ | − 0.1 | [− 0.3 0.2] |
Gender |
 Men | 0.0 | [− 0.2 0.2] |
Circumference |
 Wrist | − 0.5 | [− 3.3 2.2] |
 Mid-upper arm | − 0.5 | [− 2.7 1.7] |
- Responses from baseline survey at Southcentral Foundation. Binary outcome logit model where willingness to change = 1 and ‘unwilling/willing but hesitant’ = 0. Estimated with robust standard errors
- aMarginal effects are interpreted for continuous regressors as elasticities at the mean where the dependent, outcome variables and independent variables change at a constant rate. The categorical variables are the marginal values taken as an approximate percentage effect of the variable in response to a discrete change from zero to one, while holding all other parameters constant. More accurate defined as the opposite of the chosen device. For example, for those who chose the wrist device, the more accurate device was presented as the arm