Birthweight (gms) BPDNProp Total BPD (Bronchopulmonary Dysplasia) by birth weight Proportion of BPD decreases with birthweight.
Q: Can we give a more precise functional form to the relationship between the probability of having BPD and birthweight? Ans: Yes, if exact birthweights rather than birthweight categories are recorded. By fitting a logistic regression model to the dichotomous response BPD.
BPD No BPD Dichotomous Outcomes
In linear regression, continuous Y: Topic 18: Logistic regression
Dichotomous Outcomes
Logistic regression
Plot of logistic
Maximum likelihood estimation of the parameters
Likelihood Ratio Test
Logistic regression of BPD on birthweight Logit estimates Number of obs = 223 W = (df=1) p value = Log likelihood = BPD | Coef. Std. Err. z P>|z| [95% Conf. Interval] birthwt | _cons |
Logistic regression of BPD on birthweight Logit estimates Number of obs = 223 W = (df=1) p value = Log likelihood = BPD | Odds Ratio Std. Err. z P>|z| [95% Conf. Interval] birthwt |
Birthweight (gms) Obsrvd BPD Fitted BPD N 0-950(750) (1150) (1550) Total223 Observed versus fitted