When Opportunity Knocks, Who Answers?: New Evidence on College Achievement Awards

Joshua Angrist; Philip Oreopoulos; Tyler Williams

doi:10.3368/jhr.49.3.572

Article Figures & Data

Figures

Tables

Download figure
Open in new tab
Download powerpoint
Figure 1
Densities of Full-Year Program Earnings
Note: The figure plots the smoothed kernel densities of OK program earnings for the 2008–9 school year. Control earnings are hypothetical; treated earnings are actual. Full-year courses are double-weighted in the earnings calculation. The sample used to make this figure includes students with grades in fall and spring.
Download figure
Open in new tab
Download powerpoint
Figure 2
Full-Year Effects on Number of Courses with Grades Above Award Thresholds
Note: The figure shows treatment effects on the number of courses in which students earned a grade at or above a given threshold, where the thresholds are plotted on the x axis. Control densities are kernel density plots of grades at the course level using a normal kernel, taking only grades between 60 and 80 percent (inclusive). Treatment effects were estimated using the same models as for Table 3.

Tables

Figures

View popup

Table 1

Descriptive Statistics and Covariate Balance by Gender

	Women				Men
	First Years		Second Years		First Years		Second Years		All
	Control Mean 1	Treatment Difference 2	Control Mean 3	Treatment Difference 4	Control Mean 5	Treatment Difference 6	Control Mean 7	Treatment Difference 8	Control Mean 9	Treatment Difference 10
Age	18.2 [0.608]	−0.105 (0.056)^{^*}	19.2 [0.514]	0.011 (0.056)	18.4 [0.815]	0.014 (0.104)	19.2 [0.460]	0.069 (0.070)	18.7 [0.757]	−0.012 (0.036)
High school grade average	82.8 [6.56]	0.145 (0.238)	82.4 [6.19]	0.302 (0.217)	82.3 [6.44]	−0.344 (0.310)	82.1 [6.73]	−0.387 (0.338)	82.5 [6.44]	−0.024 (0.134)
First language is English	0.404 [0.491]	0.057 (0.056)	0.426 [0.495]	−0.046 (0.057)	0.479 [0.501]	−0.060 (0.065)	0.333 [0.474]	0.097 (0.069)	0.416 [0.493]	0.009 (0.031)
Mother a college graduate	0.395 [0.490]	0.065 (0.056)	0.477 [0.500]	−0.016 (0.058)	0.479 [0.501]	0.050 (0.065)	0.424 [0.497]	−0.034 (0.070)	0.439 [0.496]	0.020 (0.031)
Father a college graduate	0.479 [0.500]	0.051 (0.057)	0.581 [0.494]	0.009 (0.058)	0.603 [0.491]	0.047 (0.063)	0.475 [0.502]	0.105 (0.071)	0.532 [0.499]	0.049 (0.031)
Correctly answered harder question on scholarship formula	0.616 [0.487]	0.022 (0.053)	0.690 [0.464]	−0.010 (0.054)	0.719 [0.451]	−0.080 (0.061)	0.697 [0.462]	0.002 (0.065)	0.666 [0.472]	−0.014 (0.029)
Controls who would have earned some scholarship money	0.883 [0.322]		0.968 [0.177]		0.908 [0.289]		0.978 [0.148]		0.923 [0.266]
Hypothetical earnings for controls	1,240 [1,220]		1,390 [1,090]		1,430 [1,230]		1,400 [1,270]		1,330 [1,190]
Observations		449		377		246		199		1,271
F test for joint significance		1.11 {0.355}		0.453 {0.843}		0.858 {0.525}		1.43 {0.198}		0.515 {0.797}

Notes: “Control Mean” columns report averages and standard deviations for variables in the left-most column, within the relevant gender-year subgroup. ‘Treatment Difference” columns report coefficients from regressions of each variable in the left-most column on a treatment dummy, with sampling strata controls (gender, year in school, and high school grade quartile). The last row presents within-column F tests of joint significance of all treatment differences. Control group standard deviations are in square brackets, robust standard errors are in parentheses, and p values for F tests are in curly braces. Some respondents did not answer the parents’ education questions. They are coded as a separate category (“missing”) and are not coded as high school or college graduates.
↵* significant at 10 percent;
** significant at 5 percent;
*** significant at 1 percent

View popup

Table 2

Fraction of Treated Students Making Pro gram-Related Contact by Gender and Year

Contact Type	Women			Men			All
Contact Type	First-Years 1	Second-Years 2	All 3	First-Years 4	Second-Years 5	All 6	First-Years 7	Second-Years 8	All 9
Emailed advisor (Fall)	0.450	0.390	0.420	0.410	0.270	0.340	0.430	0.330	0.380
Emailed advisor (Spring)	0.520	0.440	0.480	0.660	0.380	0.520	0.590	0.410	0.500
Emailed advisor (Fall or Spring)	0.790	0.700	0.745	0.750	0.560	0.655	0.770	0.630	0.700
Checked scholarship earnings online	0.760	0.780	0.770	0.650	0.710	0.680	0.705	0.745	0.725
Emailed the program website	0.270	0.320	0.295	0.250	0.300	0.275	0.260	0.310	0.285
Any contact	0.900	0.870	0.885	0.840	0.840	0.840	0.870	0.855	0.863
Observations	100	100	200	100	100	200	200	200	400

Notes: This table shows the proportion making the indicated form of program-related contact.

View popup

Table 3

Effects on (Hypothetical) Program Earnings

	Women			Men			All
	First-Years 1	Second-Years 2	All 3	First-Years 4	Second-Years 5	All 6	First-Years 7	Second-Years 8	All 9
Panel A. Fall
Control mean	645 [657]	695 [589]	667 [628]	770 [670]	774 [642]	760 [658]	682 [663]	707 [602]	693 [637]
Treatment effect	−18.8 (53.1)	99.7 (60.9)	39.9 (39.9)	33.9 (69.8)	49.2 (73.1)	11.9 (51.3)	−5.73 (41.9)	72.0 (45.9)	28.0 (31.1)
N	444	374	818	246	195	441	690	569	1,259
Panel B. Spring
Control mean	589 [608]	711 [598]	640 [606]	644 [600]	655 [683]	649 [633]	605 [606]	696 [622]	642 [614]
Treatment effect	−57.6 (49.4)	24.7 (66.4)	−19.1 (39.6)	−20.0 (59.5)	170 (80.7)^{^**}	35.5 (49.4)	−52.5 (37.6)	77.3 (51.0)	4.47 (30.8)
N	441	340	781	242	183	425	683	523	1,206
Panel C. Full Year
Control mean	1,240 [1,220]	1,390 [1,090]	1,300 [1,170]	1,430 [1,230]	1,400 [1,270]	1,420 [1,240]	1,290 [1,230]	1,390 [1,140]	1,330 [1,190]
Treatment effect	−80.2 (95.3)	165 (121)	33.0 (74.1)	7.01 (121)	255 (144)^{^*}	54.8 (95.2)	−64.3 (74.3)	180 (91.3)^{^**}	41.1 (58.2)
N	441	339	780	242	181	423	683	520	1,203

Notes: “Control Mean” rows list averages and standard deviations of program earnings, within the relevant gender-year subgroup. “Treatment Effect” rows report coefficients from regressions of program earnings on a treatment dummy, with sampling strata controls (gender, year in school, and high school grade quartile) and controls for high school grade average, whether students’ first language is English, parents’ education, and whether students answered questions on program rules correctly Control earnings are hypothetical; treated earnings are actual. Full-year courses are double-weighted in the earnings calculation. The sample used for the full-year estimates includes students with grades in Fall and Spring. The Fall analysis omits full-year courses. If we restrict the Fall and Spring samples to be the same as the full-year sample, then the effects for the full-year are the sum of the Fall and Spring effects. (This is also true in later tables.) Robust standard errors are in parentheses; standard deviations are in square brackets.
↵* significant at 10 percent;
↵** significant at 5 percent;
*** significant at 1 percent

View popup

Table 4a

Effects on Average Grades

	Women			Men			All
	First-Years 1	Second-Years 2	All 3	First-Years 4	Second-Years 5	All 6	First-Years 7	Second-Years 8	All 9
Panel A. Fall
Control mean	68.1 [11.6]	71.0 [8.40]	69.4 [10.4]	70.7 [10.9]	72.4 [8.39]	71.4 [10.0]	68.9 [11.4]	71.4 [8.41]	70.0 [10.3]
Treatment effect	0.424 (0.945)	0.420 (0.947)	0.461 (0.662)	0.452 (1.18)	−0.520 (1.07)	−0.496 (0.827)	0.236 (0.740)	0.064 (0.694)	0.076 (0.515)
N	444	374	818	246	195	441	690	569	1,259
Panel B. Spring
Control mean	67.4 [11.3]	71.2 [9.02]	68.9 [10.5]	68.8 [11.2]	70.0 [10.6]	69.3 [10.9]	67.8 [11.2]	70.8 [9.46]	69.0 [10.6]
Treatment effect	−0.814 (1.16)	−0.118 (1.13)	−0.471 (0.801)	−0.971 (1.56)	2.54 (1.41)^{^*}	0.106 (1.03)	−0.966 (0.901)	0.727 (0.901)	−0.225 (0.634)
N	441	340	781	242	183	425	683	523	1,206
Panel C. Full Year
Control mean	67.9 [10.7]	71.1 [7.77]	69.2 [9.69]	69.9 [10.3]	71.5 [8.59]	70.5 [9.70]	68.4 [10.6]	71.2 [7.99]	69.6 [9.70]
Treatment effect	−0.323 (0.958)	0.470 (0.932)	0.076 (0.662)	−0.233 (1.21)	1.17 (1.09)	−0.146 (0.840)	−0.458 (0.745)	0.614 (0.719)	−0.025 (0.522)
N	441	339	780	242	181	423	683	520	1,203

Notes: “Control Mean” rows list averages and standard deviations of average grades, within the relevant gender-year subgroup. “Treatment Effect” rows report coefficients from regressions of average grades on a treatment dummy, with sampling strata controls (year in school, and high school grade quartile) and controls for high school grade average, whether students’ first language is English, parents’ education, and whether students answered questions on program rules correctly. Average grades are on a 100 point scale. Full-year courses are double-weighted in the average grade calculation. The sample used for the full-year estimates includes students with grades in Fall and Spring. The Fall analysis omits full-year courses. Robust standard errors are in parentheses; standard deviations are in square brackets.
↵* significant at 10 percent;
** significant at 5 percent;
*** significant at 1 percent

View popup

Table 4b

Effects on GPA

	Women			Men			All
	First-Years 1	Second-Years 2	All 3	First-Years 4	Second-Years 5	All 6	First-Years 7	Second-Years 8	All 9
Panel A. Fall
Control mean	2.39 [0.982]	2.64 [0.765]	2.50 [0.900]	2.61 [0.920]	2.75 [0.743]	2.66 [0.856]	2.46 [0.968]	2.67 [0.760]	2.55 [0.890]
Treatment effect	0.021 (0.079)	0.046 (0.081)	0.038 (0.056)	0.046 (0.103)	−0.039 (0.098)	−0.034 (0.073)	0.014 (0.063)	0.015 (0.061)	0.009 (0.044)
N	444	374	818	246	195	441	690	569	1,259
Panel B. Spring
Control mean	2.34 [0.916]	2.64 [0.783]	2.47 [0.875]	2.47 [0.935]	2.54 [0.880]	2.50 [0.912]	2.38 [0.922]	2.61 [0.810]	2.48 [0.885]
Treatment effect	−0.049 (0.081)	0.018 (0.090)	−0.016 (0.059)	−0.003 (0.106)	0.266 (0.119)^{^**}	0.071 (0.079)	−0.037 (0.064)	0.102 (0.073)	0.022 (0.048)
N	441	340	781	242	183	425	683	523	1,206
Panel C. Full Year
Control mean	2.37 [0.895]	2.64 [0.689]	2.49 [0.825]	2.55 [0.870]	2.67 [0.739]	2.59 [0.822]	2.42 [0.890]	2.65 [0.702]	2.52 [0.825]
Treatment effect	−0.021 (0.073)	0.055 (0.079)	0.018 (0.053)	0.019 (0.096)	0.126 (0.097)	0.021 (0.070)	−0.019 (0.058)	0.075 (0.061)	0.019 (0.042)
N	441	339	780	242	181	423	683	520	1,203

Notes: “Control Mean” rows list averages and standard deviations of GPA, within the relevant gender-year subgroup. “Treatment Effect” rows report coefficients from regressions of GPA on a treatment dummy, with sampling strata controls (year in school, and high school grade quartile) and controls for high school grade average, whether students’ first language is English, parents’ education, and whether students answered questions on program rules correctly GPA is on a four-point scale. The sample used for the full-year estimates includes students with grades in Fall and Spring. The Fall analysis omits full-year courses. Robust standard errors are in parentheses; standard deviations are in square brackets.
* significant at 10 percent;
↵** significant at 5 percent;
*** significant at 1 percent

View popup

Table 5

Effects on Components of the OK Scholarship Formula

	Women			Men			All
	First-Years 1	Second-Years 2	All 3	First-Years 4	Second-Years 5	All 6	First-Years 7	Second-Years 8	All 9
Panel A. Number of Courses with Grade of At Least 70 Percent
Control mean	4.58 [3.35]	5.22 [2.84]	4.85 [3.16]	5.18 [3.17]	5.01 [2.96]	5.11 [3.08]	4.75 [3.30]	5.16 [2.87]	4.92 [3.14]
Treatment effect	−0.034 (0.260)	0.422 (0.335)	0.185 (0.205)	0.128 (0.356)	0.954 (0.405)^{^**}	0.338 (0.268)	−0.010 (0.208)	0.572 (0.252)^{^**}	0.239 (0.161)
N	441	339	780	242	181	423	683	520	1,203
Panel B. Total Grade Percentage Points Over 70 Percent
Control mean	38.9 [46.2]	43.3 [42.1]	40.8 [44.5]	45.5 [47.4]	45.0 [50.4]	45.3 [48.5]	40.9 [46.6]	43.8 [44.4]	42.1 [45.7]
Treatment effect	−3.84 (3.76)	6.16 (4.64)	0.726 (2.88)	−0.290 (4.57)	7.98 (5.49)	1.05 (3.62)	−3.17 (2.87)	6.15 (3.52)^{^*}	0.861 (2.25)
N	441	339	780	242	181	423	683	520	1,203

Notes: The dependent variable in Panel A is the total number of courses in which the student received a grade at 70 percent or higher over both semesters. In Panel B, the dependent variable is the sum of the percentage points by which the student’s grades exceeded 70 percent. “Control Mean” rows list averages and standard deviations, within the relevant gender-year subgroup. “Treatment Effect” rows report coefficients from regressions on a treatment dummy, with sampling strata controls (gender, year in school, and high school grade quartile) and controls for high school grade average, whether students’ first language is English, parents’ education, and whether students answered questions on program rules correctly. Full-year courses are double-weighted in the calculation of both dependent variables. The sample used to make this table includes students with grades in Fall and Spring. Robust standard errors are in parentheses; standard deviations are in square brackets.
↵* significant at 10 percent;
↵** significant at 5 percent;
*** significant at 1 percent

View popup

Table 6

Full-Year Effects (Students Who Calculated Awards Correctly)

	Women			Men			All
	First-Years 1	Second-Years 2	All 3	First-Years 4	Second-Years 5	All 6	First-Years 7	Second-Years 8	All 9
(Hypothetical) program earnings	−218 (130)^{^*}	219 (155)	−9.32 (101)	102 (144)	370 (172)^{^**}	160 (111)	−80.4 (97.2)	245 (114)^{^**}	63.7 (74.8)
Average grades	−1.23 (1.10)	0.999 (1.12)	−0.161 (0.779)	0.839 (1.51)	1.73 (1.31)	0.754 (1.00)	−0.351 (0.913)	1.03 (0.879)	0.219 (0.634)
GPA	−0.107 (0.088)	0.112 (0.095)	−0.002 (0.064)	0.123 (0.118)	0.167 (0.117)	0.103 (0.083)	−0.008 (0.072)	0.117 (0.074)	0.044 (0.052)
Number of courses with grade of at least 70 percent	−0.339 (0.333)	0.715 (0.410)^{^*}	0.165 (0.264)	0.429 (0.431)	1.19 (0.497)^{^**}	0.637 (0.323)^{^**}	−0.008 (0.265)	0.813 (0.309)^{^***}	0.353 (0.203)^{^*}
Total grade percentage points over 70 percent	−9.21 (5.25)^{^*}	7.38 (5.98)	−1.29 (3.96)	2.97 (5.37)	12.6 (6.49)^{^*}	4.82 (4.19)	−3.98 (3.81)	8.19 (4.37)^{^*}	1.42 (2.91)
N	441	339	780	242	181	423	683	520	1,203

Notes: “Number of courses with grade of at least 70 percent” is the total number of courses in which the student received a grade at 70 percent or higher. “Total grade percentage points over 70 percent” is the sum of the percentage points by which the student’s grades exceeded 70 percent. Each row reports coefficients from regressions of the indicated variable on a treatment dummy, with sampling strata controls (gender, year in school, and high school grade quartile) and controls for high school grade average, whether students’ first language is English, parents’ education, and whether students answered questions on program rules correctly. Full-year courses are double-weighted in the calculation of the dependent variables. The sample used to make this table includes students with grades in Fall and Spring. Robust standard errors are in parentheses.
↵* significant at 10 percent;
↵** significant at 5 percent;
↵*** significant at 1 percent

View popup

Table 7

Effects in Fall 2009

	Women			Men			All
	First-Years 1	Second-Years 2	All 3	First-Years 4	Second-Years 5	All 6	First-Years 7	Second-Years 8	All 9
(Hypothetical) program earnings	7.22 (58.4)	60.0 (68.7)	33.5 (44.2)	77.6 (73.2)	22.8 (77.9)	36.8 (52.7)	22.7 (45.2)	54.1 (51.4)	33.0 (33.9)
Average grades	1.44 (0.917)	0.344 (1.17)	0.844 (0.736)	1.36 (1.49)	−2.16 (1.46)	−0.448 (1.06)	1.35 (0.803)^{^*}	−0.618 (0.912)	0.299 (0.603)
GPA	0.148 (0.079)^{^*}	0.019 (0.096)	0.082 (0.062)	0.083 (0.127)	−0.144 (0.122)	−0.037 (0.088)	0.119 (0.068)^{^*}	−0.041 (0.074)	0.033 (0.050)
Number of courses with grade of at least 70 percent	0.196 (0.163)	0.166 (0.184)	0.180 (0.121)	0.224 (0.226)	0.072 (0.230)	0.127 (0.162)	0.197 (0.132)	0.131 (0.141)	0.145 (0.096)
Total grade percentage points over 70 percent	−0.620 (2.32)	2.17 (2.69)	0.776 (1.75)	2.76 (2.74)	0.782 (3.02)	1.21 (1.99)	0.152 (1.75)	2.05 (2.02)	0.921 (1.32)
N	395	334	729	209	165	374	604	499	1,103

Notes: “Number of courses with grade of at least 70 percent” is the total number of courses in which the student received a grade at 70 percent or higher. “Total grade percentage points over 70 percent” is the sum of the percentage points by which the student’s grades exceeded 70 percent. Each row reports coefficients from regressions of the indicated variable on a treatment dummy, with sampling strata controls (gender, year in school, and high school grade quartile) and controls for high school grade average, whether students’ first language is English, parents’ education, and whether students answered questions on program rules correctly. Full-year courses are excluded from the calculation of all five dependent variables. “First-Year” and “Second-Year” continue to refer to the students’ standing during the program period. Robust standard errors are in parentheses.
↵* significant at 10 percent;
** significant at 5 percent;
*** significant at 1 percent

View popup

Table 8

IV Estimates for Participants

	Women			Men			All
	First-Years 1	Second-Years 2	All 3	First-Years 4	Second-Years 5	All 6	First-Years 7	Second-Years 8	All 9
Panel A: Full Sample
First stage (any contact)	0.901 (0.029)^{^***}	0.891 (0.032)^{^***}	0.897 (0.022)^{^***}	0.844 (0.037)^{^***}	0.874 (0.035)^{^***}	0.858 (0.025)^{^***}	0.876 (0.023)^{^***}	0.882 (0.024)^{^***}	0.878 (0.017)^{^***}
Second stages:
(Hypothetical) Program earnings	−89.0 (104)	186 (131)	36.8 (81.3)	8.31 (139)	292 (156)^{^*}	63.9 (108)	−73.4 (83.6)	204 (101)^{^**}	46.8 (65.4)
Average grades	−0.359 (1.05)	0.527 (1.02)	0.084 (0.727)	−0.276 (1.38)	1.34 (1.18)	−0.171 (0.956)	−0.523 (0.840)	0.696 (0.795)	−0.029 (0.587)
GPA	−0.023 (0.079)	0.062 (0.086)	0.020 (0.058)	0.023 (0.110)	0.144 (0.105)	0.024 (0.080)	−0.022 (0.065)	0.084 (0.068)	0.021 (0.047)
Number of courses with grade of at least 70 percent	−0.037 (0.283)	0.473 (0.362)	0.206 (0.225)	0.152 (0.407)	1.09 (0.437)^{^**}	0.394 (0.304)	−0.011 (0.234)	0.648 (0.277)^{^**}	0.272 (0.180)
Total grade percentage points over 70 percent	−4.27 (4.11)	6.92 (5.05)	0.809 (3.16)	−0.344 (5.23)	9.14 (5.96)	1.22 (4.12)	−3.62 (3.24)	6.97 (3.89)^{^*}	0.981 (2.53)
N	441	339	780	242	181	423	683	520	1,203
Panel B: Students Who Calculated Awards Correctly
First stage (any contact)	0.922 (0.033)^{^***}	0.907 (0.035)^{^***}	0.915 (0.024)^{^***}	0.863 (0.043)^{^***}	0.900 (0.037)^{^***}	0.875 (0.030)^{^***}	0.896 (0.027)^{^***}	0.895 (0.028)^{^***}	0.895 (0.019)^{^***}
Second stages:
(Hypothetical) program earnings	−237 (139)^{^*}	241 (164)	−10.2 (108)	119 (158)	411 (178)^{^**}	183 (123)	−89.8 (106)	274 (123)^{^**}	71.2 (82.0)
Average grades	−1.34 (1.16)	1.10 (1.19)	−0.176 (0.835)	0.972 (1.66)	1.92 (1.35)	0.862 (1.10)	−0.392 (0.997)	1.15 (0.950)	0.245 (0.696)
GPA	−0.116 (0.094)	0.123 (0.101)	−0.002 (0.069)	0.143 (0.129)	0.186 (0.120)	0.117 (0.091)	−0.008 (0.079)	0.130 (0.080)	0.049 (0.057)
Number of courses with grade of at least 70 percent	−0.368 (0.353)	0.788 (0.432)^{^**}	0.181 (0.282)	0.497 (0.475)	1.32 (0.511)^{^**}	0.729 (0.356)^{^**}	−0.009 (0.289)	0.908 (0.332)^{^***}	0.394 (0.222)^{^*}
Total grade percentage points over 70 percent	−9.99 (5.58)^{^*}	8.13 (6.34)	−1.41 (4.25)	3.45 (5.91)	14.0 (6.71)^{^**}	5.51 (4.62)	−4.44 (4.16)	9.15 (4.73)^{^*}	1.59 (3.19)
N	274	236	510	166	127	293	440	163	803

Notes: “First stage (any contact)” rows report coefficients from a regression of a dummy variable equal to one if the student made any program-related contact (see Table 2) on a treatment dummy “Second stage” rows report coefficients from IV regressions, instrumenting for the program contact dummy with the treatment dummy All regressions include sampling strata controls (gender, year in school, and high school grade quartile) and controls for high school grade average, whether students’ first language is English, parents’ education, and whether students answered questions on program rules correctly Full-year courses are double-weighted in the calculation of second-stage dependent variables. The sample used for this table includes students with grades in fall and spring. Standard errors are in parentheses.
↵* significant at 10 percent;
↵** significant at 5 percent;
↵*** significant at 1 percent

View popup

Table 9

Randomized Evaluations of College Achievement Awards

	Study 1	Sample 2	Treatment 3	Outcome 4	Effects
	Study 1	Sample 2	Treatment 3	Outcome 4	All 5	Men 6	Women 7
1.	Angrist, Lang, and Oreopoulos (2009) [The Student Achievement and Retention Project]	First year students at Canadian commuter university in 2005–2006, except for top HS grade quartile	$1,000 for C+ to B− first year performance, $5,000 for B+ to A performance (varies by HS grade)	First year GPA	0.010 (0.064) [1.805]	−0.110 (0.103) [1.908]	0.086 (0.084) [1.728]
				First year credits earned	−0.012 (0.064) [2.363]	−0.157 (0.106) [2.453]	0.084 (0.082) [2.298]
				Second year GPA	−0.018 (0.066) [2.040]	−0.081 (0.108) [2.084]	0.030 (0.085) [2.008]
				Second year credits earned	0.027 (0.108) [2.492]	0.155 (0.180) [2.468]	−0.024 (0.137) [2.509]
			Incentives and support services	First year GPA	0.210 (0.092)^{^**} [1.805]	0.084 (0.162) [1.908]	0.267 (0.117)^{^**} [1.728]
				First year credits earned	0.092 (0.087) [2.363]	−0.196 (0.015) [2.453]	0.269 (0.108)^{^**} [2.298]
				Second year GPA	0.072 (0.091) [2.040]	−0.170 (0.161) [2.084]	0.276 (0.106)^{^***} [2.008]
				Second year credits earned	0.072 (0.130) [2.492]	−0.240 (0.206) [2.468]	0.280 (0.172) [2.509]
2.	Angrist, Oreopoulos, and Williams (2013) [Opportunity Knocks]	First-year students on financial aid at Canadian commuter university in 2008–2009	Over two semesters and for each semester-long course, $100 for attaining at least 70 percent and $20 for each percentage point higher than this (full course load = 10 semester courses)	First year GPA	−0.019 (0.058) [2.42]	0.019 (0.096) [2.55]	−0.021 (0.073) [2.37]
				GPA, fall term of year after program	0.119 (0.068)^{^*} [2.60]	0.083 (0.127) [2.58]	0.148 (0.079)^{^*} [2.61]
		Second-year students on financial aid at Canadian commuter University in 2008-2009		First year GPA	0.075 (0.061) [2.65]	0.126 (0.097) [2.67]	0.055 (0.079) [2.64]
				GPA, fall term of year after program	−0.041 (0.074) [2.83]	−0.144 (0.122) [2.79]	0.019 (0.096) [2.85]
3.	Barrow et al. (2012) [Opening Doors Louisiana]	Low-income parents beginning community college in Louisiana between 2004 and 2005	For each of two semesters, $250 for at least half-time enrollment, $250 for C average or better at end of midterms, and $500 for maintaining a C average, plus optional enhanced college counseling	First semester credits earned	1.222 (0.285)^{^***} [4.609]	sample is mostly female
				Second semester credits earned	1.126 (0.265)^{^***} [2.77]
				Credits earned, year after program	0.343 (0.456)
4.	Cha and Patel (2010) [Ohio Performance-Based Scholarship Demonstration]	Low-income Ohio college students in 2008 with children and eligible for TANF	$1,800 for earning a grade of C or better in 12 or more credits, or $900 for a C or better in 6 to 11 credits, with payments at end of each semester	First year credits attempted	0.5 (0.4) [19.5]	sample is mostly female
4.				First year credits earned	2.0 (0.5)^{^***} [13.4]	sample is mostly female
5.	De Paola, Scoppa, and Nistico (2012)	First-year business students at the University of Calabria in 2008-2009	$1,000 for students with the 30 highest cumulative scores on all exams	Cumulative exam score	6.023 (3.059)^{^**}	5.390 (4.615)	5.841 (4.061)
				Credits earned	2.335 (1.197)^{^**}	1.759 (1.854)	2.490 (1.518)^{^*}
			$350 for students with the 30 highest cumulative scores on all exams	Cumulative exam score	5.350 (3.164)^{^*}	2.354 (4.877)	6.157 (4.207)
				Credits earned	2.194 (1.266)^{^*}	0.714 (1.970)	2.766 (1.655)^{^*}
6.	Leuven, Oosterbeek, and van der Klaauw (2010)	First-year economics and business students at the University of Amsterdam in 200–2002	$600 for completion of all first year requirements by start of new academic year	Met first year requirements	0.046 (0.065) [0.195]	not reported
				Total “credit points” in first three years	−1.2 (9.8) [84.3]
			$200 for completion of all first year requirements by start of new academic year	Met first year requirements	0.007 (0.062) [0.195]
				Total “credit points” in first three years	−2.5 (9.6) [84.3]
7.	Leuven et al. (2011)	First-year economics and business students at the University of Amsterdam in 2004–2005 and 2005–2006	$1,250 for the student with the top microeconomics exam score	Microeconomics exam score	0.974 (0.877) [18.7]	not reported
			$3,750 for the student with the top microeconomics exam score	Microeconomics exam score	1.184 (0.617)^{^*} [18.9]
			$6,250 for the student with the top microeconomics exam score	Microeconomics exam score	−0.629 (0.644) [21.2]
8.	MacDonald et al. (2009) [Foundations for Success]	At-risk students beginning community college in Ontario, Canada, between 2007 and 2008	$750 each of three semesters for 1) obtaining 2.0 GPA or higher, 2) eligible to continue in a full program the following semester, and 3) completing at least 12 hours of tutorial, case management, or career workshops	First semester GPA during program (missing imputed)	0.07 p > 0.1 [2.11]	not reported	0.12 p > 0.1 [2.20]
				Second semester GPA during program (missing imputed)	0.12 p < 0.05^{^**} [1.88]		0.14 p < 0.05^{^**} [2.04]
				Third semester GPA during program (missing imputed)	0.01 p > 0.1 [2.09]		0.12 p < 0.05^{^**} [2.16]
				Fourth semester GPA (after program)	0.07 p > 0.1 [2.18]		0.16 p < 0.05^{^**} [2.33]
9.	Miller et al. (2011) [New Mexico Performance-Based Scholarship Demonstration]	Low-income students starting at the University of New Mexico in fall, 2008, and fall, 2009	$1,000 each of four semesters for 1) obtaining 2.0 GPA or higher, 2) enrolling full time, and 3) completing two extra advisor meetings per semester	First semester credits earned	0.0 (0.2) [12.8]		not reported
9.				Second semester credits earned	0.6 (0.3)^{^*} [11.1]		not reported
10.	Richburg-Hayes Sommo, and Welbeck (2011) [New York Performance-Based Scholarship Demonstration]	New York City community college students aged 22–35 who required remediation fall, 2008, through fall, 2009	Up to $1,300 each of two or three semesters, paid in installments for achieving 1) registration, 2) continued mid-semester enrollment, and 3) 2.0 GPA in at least six credits	First semester credits earned	0.6 (0.3)^{^*} [8.1]		not reported
10.				Second semester credits attempted	0.6 (0.4) [9.3]		not reported

Notes: The table reports treatment effects for grades, credits earned, and measures of persistence. Standard errors are shown in parentheses. Control means are shown in square brackets.
↵* significant at 10 percent level.
↵** significant at 5 percent level.
↵*** significant at 1 percent level

View popup

Table 10

Quasi-Experimental Evaluations of Merit-Based College Scholarships

	Study 1	Treatment 2	Methodology 3	Outcome 4	Effects 5
1.	Castleman (2014) [Bright Futures Scholarship, Florida Medallion Scholars (FMS) and Florida Academic Scholars (FAS)]	FMS: 75 percent of public college tuition and fees for students with a 3.0 high school GPA and at least 20 on the ACT or 970 on the SAT	Difference in differences, non-eligible students as controls, HS graduates in 2000	FL public college credits, four years	−0.87 (1.90)
				FL public college BA in four years	0.00 (0.02)
		FAS: 100 percent of public college tuition and fees for students with a 3.5 high school GPA and at least 28 on the ACT or 1270 on the SAT	Difference in differences, non-eligible students as controls, HS graduates in 2000 Difference in differences, non-FMS students as controls, HS graduates in 2000	FL public college credits, four years	9.05 (1.75)^{^***}
				FL public college BA in four years	0.07 (0.02)^{^***}
2.	Cohodes and Goodman (Forthcoming) [John and Abigail Adams Scholarship Program (MA)]	MA public school tuition waived (excluding substantial fees) for students who score in the top 25th percentile of their school district and attain minimum absolute benchmarks on the statewide tenth grade test; must maintain 3.0 GPA in college	Regression discontinuity on tenth grade test score	Enrolled on time at a four-year college	0.009 (0.008) [0.716]
2.			Regression discontinuity on tenth grade test score	Graduated in four years from a four-year college	−0.017 (0.010)^{^*} [0.433]
3.	Cornwell, Lee, Mustard (2005) [Georgia HOPE]	Full tuition/fees at GA public colleges for students with a 3.0 high school GPA; must maintain 3.0 GPA in college	Differences in differences, non-GA-resident students as controls	Enrolled in full freshman course load at University of Georgia	−0.042 (0.016)^{^***} [0.812]
3.	Cornwell, Lee, Mustard (2005) [Georgia HOPE]			Completed full freshman course load at University of Georgia	−0.060 (0.019)^{^***} [0.588]
4.	Dynarski (2008) [Georgia HOPE and Arkansas merit aid program]	$1,000 at inception (now $2,500) for tuition/fees at AR colleges for students with at least 19 on the ACT and a 2.5 core high school GPA; full tuition/fees at GA public colleges for students with a 3.0 high school GPA; for AR and GA, must maintain 3.0 GPA in college	Difference in differences, other state populations as controls	Fraction of age 22-34 population with a college degree	0.0298 (0.004)^{^***} [0.337]
5.	Scott-Clayton (2011) [West Virginia PROMISE]	Full tuition/fees at WV public colleges for students with a 3.0 overall and core high school GPA and at least 21 on the ACT or 1000 on the SAT	Regression discontinuity on ACT score	Four-year GPA, WV public college students	0.099 (0.045)^{^**}
			Regression discontinuity on ACT score	Earned BA in four years	0.094 (0.022)^{^**}
			Event study, program introduction (small sample T-distribution critical values)	Four-year GPA, WV public college students Earned BA in four years	0.039 (0.018)
				Earned BA in four years	0.067 (0.005)^{^***}
6.	Sjoquist and Winters (2012a) [Georgia HOPE and Arkansas merit aid program]	$1,000 at inception (now $2,500) for tuition/fees at AR colleges for students with at least 19 on the ACT and a 2.5 core high school GPA; full tuition/fees at GA public colleges for students with a 3.0 high school GPA; for AR and GA, must maintain 3.0 GPA in College	Difference in differences, other state populations as controls; increased sample and updated clustering compared with Dynarski (2008)	Fraction of age 22–34 population with a college degree	0.0091 p=0.216 [0.3567]
7.	Sjoquist and Winters (2012b)	25 state merit aid programs with requirements on high school GPA, ACT/SAT scores, and college credit enrollment and GPA Nine strongest state merit aid programs with requirements on high school GPA, ACT/SAT scores, and college credit enrollment and GPA	Difference in differences, non-merit state populations as controls	Fraction of age 24-30 population with a college degree	−0.0008 (0.0028) [0.388]
7.	Sjoquist and Winters (2012b)			Fraction of age 24-30 population with a college degree	0.0011 (0.0037) [0.388]

Notes: The table reports main baseline sample outcomes for grades and measures of persistence. Standard errors are shown in parentheses. Control means are shown in square brackets.
↵* significant at 10 percent level.
↵** significant at 5 percent level.
↵*** significant at 1 percent level