THE IMPACT OF THE HIGH SCHOOL JUNIOR ROTC PROGRAM: DOES TREATMENT TIMING AND INTENSITY MATTER?

Abstract

The Junior Reserve Officers’ Training Corps is a high school program that combines classroom teaching with extracurricular activities. The program is located primarily in inner city schools and serves at‐risk students. Its goals are multidimensional and include military preparation and improving academic achievement. Using High School and Beyond data we find that the program’s effects depend on the timing and intensity of involvement. Test scores, graduation rates, and enlistments are higher for students who participate early in high school and for those who persist in the program. Conversely, we find few effects for students participating in the last two years of high school.

Keywords:

• JROTC
• High school achievement
• At‐risk youth
• Enlistments

JEL Classification:

• I2

ACKNOWLEDGEMENTS

The authors thank David Neumark, Barry Hirsch, Alice Crawford, Linda Bailey, Dahlia Remler, Yu‐Chu Shen, and seminar participants at Baruch College/CUNY, and at the 2008 WEAI Conference for helpful comments. Yee‐Ling Ang and Janet Days provided research assistance.

Notes

¹ Indicative of the controversy are anti‐JROTC campaigns in some major cities (Nazario, 2007) and the decision in 2006 by the San Francisco School Board to eliminate JROTC from all city high schools (Tucker, 2006).

² Surveys indicate that 30–50% of JROTC graduates intend to enlist (Bailey et al., 1992; Center for Strategic and International Studies, 1999). However, only a fraction of those with positive intentions actually enlist, and an even smaller percentage ‘graduate’ (i.e. complete the 4‐year program).

³ By way of comparison, JROTC (with 3300 schools) is nearly twice the size of the career academy program (1500 schools) and far larger than Upward Bound (727 schools) (Kemple and Snipes, 2000; Myers et al., 2004).

⁴ Information on JROTC in other youth surveys, such as NELS or NLSY, can only be inferred from transcripts, which yields smaller samples of participants. In addition, other surveys do not collect information on whether a school offers JROTC.

⁵ HSB asks students whether they have ever heard of or participated in a list of programs, one of which is JROTC. We define participants as those reporting to have participated, not simply heard of the program.

⁶ Our continuous participation indicator is not as precise as we would like. Due to the wording of the question in HSB, it is possible that some who respond positively in either 1980 or 1982 remain in the program as long as those who respond positively in both years. However, we believe the number of such individuals is small because continuous participants appear very different in their observable characteristics from both early and late participants. In terms of estimation, the measurement error in the continuous participant indicator would, if anything, bias the program effect toward zero. In addition, one solution to measurement error is to instrument for the mismeasured variable, which is the approach we take in estimating causal effects.

⁷ Since students most likely drop out due to poor academic outcomes, including them in the control group when treatment includes late JROTC participants would bias the results in favor of JROTC. A similar problem arises if we investigate program effects on dropout behavior by defining the early participant group to include only those who participate in 9th or 10th grade but no longer continue with the program. Removing continuous participants, who, by default, stay enrolled in high school, would negatively bias program estimates.

⁸ About 1452 students (9.8% of the sample) have missing information on JROTC. They appear to be no different from non‐participants in their observed characteristics.

⁹ Table I presents unweighted summary statistics for the sample used in regressions. Regression results are also not weighted. Weighted summary statistics indicate that students who ever participate represent about 3.3% of the population of high school sophomores in 1980.

¹⁰ Where possible, we fill in missing information in the base year with answers to the same questions collected in later follow‐ups.

¹¹ We focus on the sum of the ‘formula’ score on the vocabulary, reading, and the first part of the mathematics test.

¹² To be consistent with the measurement of the rest of the outcomes we also obtained results for enlistment rates two years after high school. The results were very similar to those that used the later follow‐ups, suggesting that most enlistments occur soon after high school.

¹³ Talent Search targets disadvantaged students and aims to reduce dropout rates and increase postsecondary education. Upward Bound serves students from low‐income families, students whose parents have less than a bachelor’s degree, and low‐income, first‐generation military veterans who are preparing to enter postsecondary education. Upward Bound aims at boosting high school graduation and postsecondary enrollments. Some funds were also earmarked for occupational training to potential high school dropouts and for developing work‐study programs. ESEA is a precursor to No Child Left Behind and Title I targeted schools with disadvantaged students.

¹⁴ The average school offers seven federal programs. The 18 schools that do not offer any federal programs are very small (maximum enrollment 282). If we include JROTC as one of these programs, only 1% of the school sample does not participate in one of these initiatives at some point in time. JROTC is offered by 191 schools (18% of the sample), of which 23 add or drop the program during 1980–1982.

¹⁵ This result is obtained from a regression of the number of federal programs on ‘school offers JROTC,’ which also includes controls for school size, urban location, private ownership, and region (results available on request).

¹⁶ Note that we are not assuming away the existence of aggregate effects of these programs, only that aggregate effects are confined to the specific outcomes targeted by each program.

¹⁷ Including a single dummy for whether the school ever had a JROTC program made no difference in the estimated program effects.

¹⁸ The full range of results for equation (2) for all sub‐groups and outcomes are available on request.

¹⁹ To assess whether graduation effects derived from the IV approach are sensible, we follow Altonji et al. (2005) and construct bounds around the estimates. This approach assumes that any unobserved heterogeneity is, at most, of the same magnitude as observed heterogeneity. Since JROTC students are negatively selected, unobserved characteristics would bias graduation effects downward. Therefore, OLS estimates would provide a lower bound, and estimates obtained under the Altonji et al. assumption would provide an upper bound. After implementing this approach, we find that graduation effects for early participants could be as large as 10 points higher, which suggests that our selection‐adjusted estimates are in the right direction and, if anything, may underestimate the true program effect.

²⁰ Both are significant. In a probit enlistment model for early participants SJROTC_j,1980 has a partial effect of −0.028 (0.11), whereas SJROTC_j,1982 has a partial effect of 0.03 (0.15). Results are available upon request.

²¹ The variation obtained from schools adding and dropping the program over time is too small when working with the school survey.

²² To reduce measurement error we average over time the percentages of students who enlist, the percentage disadvantaged, and the percentage of faculty with advanced degrees.

²³ If we assume an enrollment of 500,000 students each year, that roughly 13% graduate as JROTC participants, and that among these 30% enlist (a lower bound estimate, see Center for Strategic and International Studies, 1999), the cost to DOD per new recruit via JROTC would be about $11,350, which compares favorably with the average cost via direct recruitment programs.