Table 3. Odds ratios of bike use associated with education level and individual income in single and multiple predictor models after controlling for different sets of covariates.

Socio-economic position	Model 0		Model 1
	OR (CI 95%)	p-value	OR (CI 95%)	p-value
Education level single predictor models
Education level		0.001		<0.001
Less than HS	Referent		Referent
HS/professional	0.95 (0.87, 1.03)	0.209	0.88 (0.80, 0.96)	0.004
College or higher	0.78 (0.69, 0.89)	<0.001	0.70 (0.61, 0.80)	<0.001
Age			0.99 (0.98, 0.99)	<0.001
Gender (woman)			0.38 (0.35, 0.41)	<0.001
Income level single predictor models
Income level*		0.009		<0.001
Low	Referent		Referent
Medium-low	1.22 (1.04, 1.43)	0.014	0.98 (0.83, 1.15)	0.773
Medium-high	1.14(0.97, 1.35)	0.104	0.80 (0.68, 0.95)	0.011
High	0.96 (0.79, 1.16)	0.680	0.65 (0.53, 0.79)	<0.001
Age			0.99 (0.98, 0.99)	<0.001
Gender (woman)			0.37 (0.32, 0.41)	<0.001

HS, high school.

^*For the surveys from 2008 to 2012 and 2014 the values were: low equivalent to ≤$200, medium-low equivalent to $201–$400, medium-high equivalent to $401–$800, and high equivalent to ≥$801. For the survey from 2013 the values were: low equivalent to ≤$100, medium-low equivalent to $101–$400, medium-high equivalent to $401–$800, and high equivalent to ≥$801. And for the surveys from 2015 to 2018 the values were: low equivalent to ≤$400, medium-low equivalent to $401–$800, medium-high equivalent to $801–$1,600, and high equivalent to ≥$1,601.

Model 0 definition: model without covariates; city as random intercept and year as random slope.

Model 1 definition: model after controlling for age and gender; city as random intercept and year as random slope.

Sample size education level single predictor models: Model 0 (n = 77,765); Model 1 (n = 77,765).

Sample size income level single predictor models: Model 0 (n = 40,545); Model 1 (n = 40,545).