Field
Last Published
|
Before
July 05, 2023 10:38 AM
|
After
July 05, 2023 10:43 AM
|
Field
Power calculation: Minimum Detectable Effect Size for Main Outcomes
|
Before
Our power calculations are based the results of the pilot study conducted in 2019. In the pilot we began
with a baseline of 61 candidates our partner had identified as eligible to fill 19 open spots in the mentoring
program. We randomized invitation priority among this group, and our partner extended invitations in
this random-priority order. After approximately 40% attrition between baseline and endline we
completed the study with a sample of 31 matched responses from girls in the randomized priority group.
Our pilot started with 51 candidates who completed the baseline survey. At the same scale it would take
six to seven years of enrollment to detect the desired effect size across all outcomes. The power of the
study could be scaled up either by increasing the number of treatment spots available and increasing the
size of the control group or by increasing the size of the control group itself, and we are in conversation
with our partner on the feasibility of expanding the sample in these ways, with targeted larger samples
reflected in the budget. For most outcomes these MDEs are conservatively low relative to the estimates
in the pilot, so it is likely that the study would reach statistical significance earlier than expected. Between
efforts to expand the sample and reaching less conservative target we are expecting that five years would
be enough data collection for all variables of interest.
|
After
Our power calculations are based the results of the pilot study conducted in 2019. In the pilot we began
with a baseline of 61 candidates our partner had identified as eligible to fill 19 open spots in the mentoring
program. We randomized invitation priority among this group, and our partner extended invitations in
this random-priority order. After approximately 40% attrition between baseline and endline we
completed the study with a sample of 31 matched responses from girls in the randomized priority group.
Our pilot started with 51 candidates who completed the baseline survey. At the same scale it would take
six to seven years of enrollment to detect the desired effect size across all outcomes. Between
efforts to expand the sample and reaching less conservative targets we are expecting that five years would be enough data collection for all variables of interest.
|
Field
Secondary Outcomes (End Points)
|
Before
A secondary outcome of interest is whether heterogeneous support from family members, including siblings participating in the same mentoring program, moderate the impact of the program on plans after high school graduation (including willingness to work, attain additional training or internships, further study, graduate from high school or stay at home.
.
|
After
A secondary outcome of interest is whether heterogeneous support from family members, including siblings participating in the same mentoring program, moderate the impact of the program on plans after high school graduation (including willingness to work, attain additional training or internships, further study, graduate from high school or stay at home).
.
|