Counts of male, female, adult, and child cyclists were
totaled for each hour and day. Means and standard deviations were calculated for the number of people observed
cycling before and after the bike lanes were installed.
We tested 3 hypotheses: 1) that the number of people
observed cycling would increase from baseline to followup, 2) that the proportion of cyclists riding with traffi
increased, and 3) that the proportion of people riding on
the street instead of the sidewalk increased. We used negative binomial regression to test the fist hypothesis and
binary logistic regression to test the second and third. The
outcome for fist hypothesis was the number of people
observed cycling. For the second and third, the outcome
was binary indicating appropriate versus inappropriate
behavior (eg, riding on the street or sidewalk and with or
against traffi). We considered a P-value of 0.05 or less
to be statistically signifiant. For all models time (pre/
post) was the predictor. The unit of analysis was day for
the fist hypothesis and individual cyclists for the second
and third hypotheses.