Figure - PMC

Skip to main content

View full-text article in PMC

. Author manuscript; available in PMC: 2010 Apr 1.

Published in final edited form as: Soc Sci Med. 2009 Mar 11;68(8):1439–1447. doi: 10.1016/j.socscimed.2009.02.003

Nonlinear regression models (thick black lines) of the effect of educational attainment (see Table 1 for coding details) on current smoking (A), and ln(cigarettes/day) (B) from Equations S1a–S1b overlaid on the parallel nonparametric models (thin black lines) with pointwise 95% confidence intervals (thin grey lines) from Equations 1a–1b. A fully cubic specification models educational attainment’s effect on current smoker status, producing the fit in Panel A. A linear threshold of ln(cigarettes/day) reflects the initial unresponsiveness of consumption to educational attainment followed by a strong negative effect at four years of high school education or more. Equation 2b estimated that the change in effect of educational attainment on ln(cigarettes/day) status was at twelfth grade, but no diploma, and higher (38, SE 0.71).

Nonlinear regression models (thick black lines) of the effect of educational attainment (see Table 1 for coding details) on current smoking (A), and ln(cigarettes/day) (B) from Equations S1a–S1b overlaid on the parallel nonparametric models (thin black lines) with pointwise 95% confidence intervals (thin grey lines) from Equations 1a–1b. A fully cubic specification models educational attainment’s effect on current smoker status, producing the fit in Panel A. A linear threshold of ln(cigarettes/day) reflects the initial unresponsiveness of consumption to educational attainment followed by a strong negative effect at four years of high school education or more. Equation 2b estimated that the change in effect of educational attainment on ln(cigarettes/day) status was at twelfth grade, but no diploma, and higher (38, SE 0.71).