Raising the Bar: Impacts and Implementation of the New Heights Program for Expectant and Parenting Teens in Washington, dc

Figure D.2. Pre- and post-expansion averages, by semester and parental 
status
 
Source: DCPS administrative data; DOH administrative data. 
Note: 
All regressions include age indicators, race and ethnicity indicators, and an indicator for being over-age 
when entering 9th grade.
An alternative approach to estimating impacts in this study could be to model the 
relationship between outcomes and time, both before and after New Heights expansion (instead 
of calculating simple averages). We illustrate this approach in Figure D.3, where we fit four lines 
corresponding to parents (blue) and nonparents (red), before and after New Heights expansion. 
Before New Heights expansion, the lines for parents and nonparents are approximately parallel 
D.5 

---

APPENDIX D: SUPPLEMENTAL ANALYSES
 
(that is, they have similar slopes), which suggests that nonparents are a reasonable comparison 
group for these outcomes.
We ultimately rejected this approach to estimating impacts because (1) there are not enough 
time points to credibly assess what functional form we should use to model the relationship 
between outcomes and time and (2) the presence of outliers can significantly distort the 
regression line. In short, we believe the simple difference-in-difference approach used in our 
main analysis is most appropriate in this context.
Figure D.3. Modeling pre- and post-expansion trends 
Source: 
DCPS administrative data; DOH administrative data. 
Note: 
All regressions include age indicators, race and ethnicity indicators, and an indicator for being over-age when 
entering 9th grade.
D.6 

---

APPENDIX D: SUPPLEMENTAL ANALYSES
 
Comparing outcomes of program participants and nonparticipants 
Although our primary analyses examine the impact of the offer of New Heights 
participation, we would logically expect the impact of the offer to be the result of program 
participation. However, this might not be the case if other policy changes or demographic shifts 
were affecting all parenting youth, including those who do not participate in New Heights. We 
therefore conducted a diagnostic descriptive analysis comparing the post-expansion outcomes of 
participants to parenting nonparticipants and to pre-expansion parents and nonparents. If New 
Heights genuinely has positive effects, we would expect to see better outcomes for New Heights 
participants than for parenting nonparticipants. We caution that this is just a diagnostic analysis, 
not an attempt to estimate the impact of New Heights participation. 
We calculated the average outcomes for nonparents and parents before New Heights 
expansion and the outcomes of nonparents, parents participating in New Heights, and parents not 
participating in New Heights after the expansion. We report these averages in Figures D.4–D.8. 
For every outcome, New Heights participants fare better in the post-expansion years than 
parenting females before the expansion and they fare better than parenting females who do not 
participate in New Heights after the expansion. These findings suggest that it is appropriate to 
attribute the impacts described in Table 2 of the main text to the New Heights program (rather 
than some other factor affecting all parenting females). Specifically: 
•