Weighted Logistic Regression Analysis of Effect of Early Childhood Stimulation on Violent Behavior

Residents Only, Odds Ratio (95% Confidence Interval)Total Sample, Odds Ratio (95% Confidence Interval)
Involved in physical fighta0.20 (0.05–0.78)1.02 (0.30–3.44)0.36 (0.12–1.06)0.76 (0.29–2.03)
Involved in violent crimeb0.35 (0.11–1.06)1.38 (0.46–4.16)0.33 (0.11–0.93)1.49 (0.51–4.33)
Weapon usec0.66 (0.27–1.63)1.31 (0.51–3.35)0.59 (0.25–1.38)1.38 (0.54–3.39)
  • Weighted logistic regression analyses: Variables entered were stimulation; supplementation housing; and maternal education on enrollment. The regression for violent crime is estimated on the subsample of mothers without secondary exams (91 of 105 for the entire sample and 79 of 91 for the resident sample) because none of the participants whose mother had secondary exams was involved in violent crime. Hence, the data is uninformative on how the interventions could have changed the probability of being involved in violent crime for these children.

  • a Involved in physical fight or started fight.

  • b Involved in any of the following: fight with weapon; hurt someone with weapon; carry gun in past month; threatened someone with gun; shot someone with gun; or gang member.

  • c Carried weapon in past month or ever threatened someone with a weapon.