2c. On the following page is Table 3, explain what the first colunm and second column are showing. (This is information we discussed in the Asynehtunous Lecture: W) 2d. (Still referring to Table 3) The third column is the ZSLS (two staged least squares), it gives us the coefficient estimates with the correct standard errors for the N regression. \"mat do the estimates tell us about the effect of SEED on reading and math test scores? 2e. Next refer to Table 4 also on the next page. These are the ZSLS results broken up by demographic characteristics. Explain what Table 4 says about the effects by different demographic subsamples. 2f. What do these results tell us about the effects of charter schools on short-term outcomes? What are reasons that we canit extrapolate these results as estimates of short-term outcomes for all charter schools? 2g. What are the key takeaways from these results (i.e. how would you describe these results to a policymaker interested in improving academic achievement?) Table 3: Lottery Results First Reduced 2SLS Stage Form Outcome Controls (1) (2) (3) Reading Basic 0.934 0.236 0.252 (0.075) (0.100) (0.106) Baseline Scores 0.922 0.168 0.182 (0.076) (0.081) (0.088) Baseline Scores 0.949 0.188 0.198 and Demographics (0.078) (0.087) (0.093) 303 Math Basic 0.933 0.384 0.412 (0.075) (0.116) (0.122) Baseline Scores 0.921 0.283 0.307 (0.076) (0.083) (0.090) Baseline Scores 0.949 0.218 0.230 and Demographics (0.078) (0.081) (0.084) 301 NOTES: This table reports estimates of the effect of attending SEED on achievement. The sample is students who applied to the SEED School of Washington, D.C., in 2007 and 2008. Columns (1)-(3) report the first stage, reduced form, and 2SLS coefficients from instrumenting years in SEED us- ing an indicator for having won the SEED lottery. This indicator is equal to one if the applicant was offered admission either immediately or off the waitlist. Applicants with sibling priority or who applied late and were not included in the original lottery are excluded. All regressions include an indicator variable for sex, since separate lotteries were conducted for males and females. All regressions combine the 2007 and 2008 cohorts and include dummies for grade of test and applicant year as well as a contempo- rancous sibling dummy and the interaction of the contemporancous sibling dummy with applicant year. Estimates are also reported for regressions including controls for baseline test scores in reading and math as well as demographic controls for free lunch eligibility, special education status, and English Language Learner status. Because every applicant in the lottery sample is black, race controls are not included. Robust standard errors (clustered at the student level) are reported in parentheses. Numbers of observations are reported directly below each set of estimates.observations are reported directly below each set of estimates. Table 4: Lottery Results by Subsample Full Free Non- Non- Sample Male Female p-value Lunch Free Special p-value Education Special p-value Lunch Education Outcome (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) Reading 0.198 -0.138 0.382 0.014 0.254 0.023 0.109 0.105 0.224 0.640 (0.093) (0.145) (0.155) (0.123) (0.075) (0.235) (0.100) 303 94 126 184 58 38 250 Math 0.230 0.037 0.265 0.280 0.196 0.125 0.636 0.102 0.288 0.555 (0.084) (0.156) (0.142) (0.104) (0.108) (0.302) (0.090) 301 94 125 183 58 38 249 NOTES: This table reports estimates of the effect of attending SEED on achievement for subsets of the lottery sample, Columns (1)-(3), (5)-(6), and (8)-(9) report 2SLS coefficients from instrumenting years in SEED using an indicator for having won the SEED lottery. This indicator is equal to one if the applicant was offered admission either immediately or off the waitlist. Columns (4), (7), and (10) report p-values for the F-test for the hypothesis that the coefficients in the preceding two columns are equal. Applicants with sibling priority or who applied late and were not included in the original lottery are excluded. All regressions include an indicator variable for sex, since separate lotteries were conducted for males and females. Because every single male student in the 2008 lottery was offered admission to SEED, the regressions for males and females in Columns (2) and (3) only include the 2007 cohort and include grade of test dummies and a contemporaneous sibling dummy. All other regressions combine the 2007 and 2008 cohorts and include dummies for grade of test and applicant year as well as a contemporancous sibling dummy and the interaction of the contemporancous sibling dummy with applicant year. All regressions include controls for baseline test scores in reading and math as well as demographic controls for race, free lunch eligibility, special education status, and English Language Learner status. Because every applicant in the lottery sample is black, race controls are not included. Robust standard errors (clustered at the student level) are reported in parentheses. Numbers of observations are reported directly below standard errors