Notes: Robust standard errors clustered by family ID. Analyses include all controls from Table 1. Column 1 includes all children from non-teen-childbearing families as controls, with control students’ t = 0 set at their seventh observation (mean age = 13.6). Column 2 includes matches to younger siblings from non-teen-childbearing families to the teen aunts/uncles based on the first-observed characteristics, including first-observed test scores. Column 3 replaces first-observed test scores with scores from two years before birth. Column 4 replaces one observed test score with three-year test score fiends (t = −2, −3, and −4). Column 5 adds three-year family average test score trends. Column 6 adds the neighborhood requirement. Scores at t = 0, repeats grade, drops out, juvenile justice, and college-going outcomes include one weighted observation from the siblings and their controls. Column 1 test scores at t = 0 include multiple observations per individual control. Fixed-effects models include all available observations from t = −5 to t = +2 from the siblings and their noted controls, excluding t = −1. Test scores reported as nationally norm-referenced percentiles (1–100). The mean for the preferred IND+FAM control group scores at the percentile 43.6 in t = 0, with 27.1 percent repeating a grade in any year following the birth, 15.5 percent dropping out following the birth, 7.8 percent juvenile justice system exposure following the birth, 56.6 percent college attendance, 26.2 percent obtaining any degree or certificate, and 17.0 percent obtaining a four-year degree.