Study of the Impact of Specialized Public High Schools of Science, Mathematics, and Technology

Principal Investigator: 
Project Overview
Background & Purpose: 

Why Study Selective Science, Mathematics and Technology (SMTl Schools and Programs?

  • Recognition that some students complete middle school with strong interests, skills, and knowledge in science and mathematics, and aspire to high level careers in STEM.
  • Advanced students are not being sufficiently challenged in many of today's high schools and SMT schools and programs offer one way to meet their educational needs.
  • Curricular and environmental components of SMT schools and Talent Search (TS) programs that appear to predict completion of STEM university degrees might be adapted more widely.

Study Research Goals

  • Overarching Research Question: What is the impact of participation in a selective SMT school or Talent Search program on completion of a STEM major in university?
  • Sub-research question 1: What characteristics of (a) the SMT schooi/TS program or (b) the participants contribute to this outcome?
  • Sub-research question 2: Do these characteristics differentially affect male and female graduates? Those whose parents are not college graduates? Those whose parents are not in STEM related careers?

The research team conducted a survey designed to address the study goals. Respondents included 4,113 individuals who had either graduated from one of 25 specialized science high schools between 2004-2007 or same age individuals who qualified for Talent Search (high SAT scores), did not attend a specialized school, and took mathematics or science classes in the summers.

Study Description: Participants

  • 2004-2007 graduates of 25 selective SMT (specialized science high school) schools (N=3,510)
  • Same age peers who did not attend a selective SMT school, attended Talent Search, and enrolled in mathematics or science classes in the summer (N=603)
  • For the overall study population, 69% were White, 23% Asian, 4% African-American, 4% Hispanic, and 5% reported other ethnic backgrounds
  • 79% reported English as their primary language
  • 53% of SMT participants were female as were 55% of Talent Search respondents

SMT School Model Descriptions

  • Residential: Draws from an entire state. All students reside on campus. (n=8 schools)
  • Half-Time: Students attend a regional center for SMT courses daily. (n=7 schools)
  • Full-Time Commuter: Whole school is SMT and draws from local metropolitan population. (n=4 schools)
  • School-within-School: An academy within a regular high school (n=6 schools)
  • Talent Search: Selective invitation to enroll in advanced summer SMT courses based on high standardized test scores.

Each school model is designed to provide advanced coursework, expert teachers, peers motivated and interested in SMT, and opportunities for independent research. However, each model also has additional design features that potentially affect outcomes for their graduates. First, state residential schools serve students from all around the state, ensuring that every county has a chance at representation in the entering class. Second, comprehensive or self-contained schools are usually located in large metropolitan areas, serving an entire district or city's most talented students. Third, Schools-within-schools are also typically located in urban areas, and were often founded to bring cohorts of academically talented students and additional resources to schools with fewer resources. The schools-within-schools model allows SMT students to participate in non-SMT classes with their schoolmates and for some non-SMT students in the school to take advanced courses designed for the SMT cohorts. The fourth category of specialized SMT schools is the half-day or part-time school model available in Virginia and Michigan, usually in poorly resourced or rural counties. These schools serve a geographical region, busing in students from a number of high schools to participate in advanced SMT coursework for part of the day. The rest of the day, students are enrolled in their home high school.

Research Design: 

This project is cross-sectional and comparative and is designed to generate descriptive [observational] correlational or associative [factor analysis, multi-level regression modeling, hierarchical linear modeling] evidence. Original data are collected through school records or policy documents and survey research.

The survey instrument for the Impact of Specialized Public High Schools of Science, Mathematics, and Technology study was employed using email solicitations. Contact information for the specialized high school alumni and the comparison group was obtained through a combination of school records and alumni association lists, Of the 9,461 email solicitations, 4,113 completed the online survey for an overall response rate of 43.5%.

For analysis purposes, the respondents are nested within school type. That is, the data for this study were clustered within clear groupings based on whether the respondents attended a residential SMT high school, comprehensive SMT high school, SMT school within a school, a half-day SMT high school, or were part of the comparison group. When analyzing data (i.e., survey responses) clustered within clear groupings (i.e., school model), it is necessary to calculate the portion of variance in the outcome variable that occurs at the individual respondent level and the portion that occurs at the school level If a significant proportion of the variance in the outcome variable occurs at the school level (> 10.0%), suggesting that the particular model of specialized SMT high school accounts for the variance in participants responses, multi-level models must be utilized to account for this variance, If a significant proportion of the variance in the outcome variable does not occur at the school level (< 10.0%), suggesting that the variance in participants' responses can be accounted for by individual differences in respondents and not the particular model of specialized SMT high school, multi-level models are not necessary.

For this reason, we examined the intra-class correlation (ICC) for this data set and calculated the ICC to be 0.0123 (or 1.23% of the variance) at the school level for the outcome variable of undergraduate degree major. As a result, analyses appropriately focus on variables at the level of individual respondents and utilize binary logistic regression as the analytic procedure.

With the ICC finding in mind, this study examines individual variables or factors and how they are associated with the graduates' of specialized STEM high schools and participants in Talent Search programs' completion of an undergraduate STEM degree. The outcome variable, undergraduate degree major, was re-coded into STEM related and non-STEM related areas, creating a binary outcome variable. Binary logistic regression models were then developed to address each of the research questions in this study. Each binary logistic regression model included background and demographic variables to act as controls for differences in the following: gender, race/ethnicity, immigrant status, parental/guardian educational level, primary language spoken at home, parental/guardian in a STEM related career, an expressed early interest in STEM related areas, and, as expected, attendance at a specialized SMT high school versus participation in a Talent Search program.


Completion of STEM Degree

  • 49.8% percent of graduates of selective SMT schools
  • 53.4% percent of Talent Search participants
  • 22.6% of all students entering college complete a STEM undergraduate degree (source NSF)
  • 26.5% of students scoring at the 95th percentile or higher in a sample of SAT-M test results (overall sample N=87,840)
  • 17.3% of students scoring at or above the 95th percentile on the combined SAT Critical Reading, Math, and Writing tests (sample size=87,740)

Completion of STEM University Degree by Specialized SMT School Model

  • School w School = 58.3%
  • Residential = 51.7%
  • Half-Time = 48.4%
  • Full-Time Commuter = 42.3%

Percentage of Participants Completing STEM degrees by Gender

  • Female
    • Specialized SMT School = 46.1%
    • Talent Search = 50.5%
  • Degree Fields Selected by Participating Female STEM Completers
    • Bio and BioMed = 33%
    • Engineering = 11%
    • Physical Science = 9%
    • Mathematics = 6%
  • Male
    • Specialized SMT School = 57.8%
    • Talent Search = 61.9%

Parent Working in a STEM Field

  • Graduates of specialized SMT schools with a parent in STEM are 1.37 times more likely to complete a STEM related major than SMT graduates with no parent working in STEM related fields.
  • Respondents from the Talent Search with a parent in STEM were twice as likely to complete a STEM degree.
  • Looking at those without a parent in STEM, there was no difference in odds of completing a STEM degree between graduates of specialized SMT School or Talent Search.
  • 20% of SMT school graduates who completed STEM degrees had parents with no education beyond high school; the same was true for 2% of the Talent Search participants.

Signature High School Factors that Predict STEM Degree Completion

  • Participation in an authentic high school research experience:
    • Overall, those who participated in high school research were nearly two times more likely to complete a STEM degree.
    • Among female students, those who participated in high school research had nearly twice the odds (1.95) of completing a STEM university degree compared to females who did not.
  • Feelings of belonging in the academic setting:
    • 55.8% of respondents that attended a specialized SMT high school said that their high school experiences in SMT helped them to be well prepared in their chosen major, compared to their university classmates.
    • 24.6% of Talent Search participants agreed.

Motivation and Interest

  • 40% became interested in topics related to their eventual major before high school. Respondents in this group were 52.9% more likely to report that they earned a STEM related major or concentration.
  • SMT school graduates motivated to attend their school based on interest in STEM were 2.5-4 times more likely to obtain a STEM degree than fellow graduates with different motivations.
  • For SMT graduates who reported other motivations for attending their SMT school, the study found:
    • Those primarily motivated by prestige and recognition were 30% less likely to obtain a STEM degree.
    • SMT graduates whose attendance was primarily motivated by getting into a good college were 25% less likely to obtain a STEM degree.
    • SMT graduates primarily motivated by the academic peer group available at the school were 30% less likely to receive a STEM degree.

Behavioral and Social Sciences (BSS) Degrees

  • 28% of SMT school graduates reported earning an undergraduate degree in the behavioral or social sciences; Talent Search participants were no more likely to obtain a degree in these disciplines.
  • SMT graduates with BSS undergraduate degrees, who reported that experiences in college (rather than high school) were the most important determinant of their major, were more than twice as likely to earn an undergraduate degree in BSS.
  • There were no differences between college graduates in life or physical sciences and BSS in terms of self-reported intellectual capacity for mathematics and science.
Publications & Presentations: 


Subotnik, R.F., Tai, R.H., Almarode, J , and Crowe, E. (2013). What are the Value Added Contributions of Selective Secondary Schools of Mthematics, Science, and Technology? Talent Development and Excellence Vol. 5, number 1. pp. 87-97.

Subotnik, R.F. & Tai, R.H. . (2011) Successful education in the STEM disciplines: Selective STEM schools. For workshop report conducted by the National Research Council’s Board on Science Education and board on Testing and Assessment on Successful STEM Education in K-12 Schools.

Subotnik, R.F., Edmiston, A., Lee, G.M., Almarode, J. & Tai, R.H. (2011). Exploring intensive educational experiences for adolescents talented in science. In A. Ziegler & C. Perleth (Eds.). Excellence: Essays in honor of Kurt A. Heller. Munich, Germany: LIT Verglag.

Subotnik, R.F., Tai, R.H., Rickoff, R. & Almarode, J. (2010). Specialized Public High Schools of Science, Mathematics, and Technology and the STEM Pipeline: What Do We Know Now and What Will We Know in Five Years? Roeper Review, 32, 7-16.


Since our last report to NSF, we have conducted the following status reports on the study:

  • November, 11, 2010 in Atlanta, at the professional conference of the National Consortium of Specialized Secondary Schools of Mathematics, Science and Technology
  • January 21, 2011 in Chicago, at a meeting of the Board of Directors of the Illinois Science and Mathematics Academy.
  • March 7, 2011 in Maryland, at a gathering of students in the research program at Montgomery Blair High School Science, Mathematics and Computer Science Magnet.
  • May 10, 2011, Workshop on successful STEM education in K-12 Schools, National Research Council, Washington DC
  • August 6, 2011, American Psychological Association convention, Washington DC
  • October 28, 2011 – specific date to be determined. The professional conference of the National Consortium of Specialized Secondary Schools of Mathematics, Science and Technology, Austin, Texas
  • November 5, 2011, Convention of the National Association for Gifted Children, New Orleans, LA
  • November 17, 2011, Invited presentation to meeting of National Sciences Resource Center
Other Products: 

Our primary target is to generate policy recommendations for high schools that reflect the best practices available to talented and interested students in STEM at specialized high school. Our second target is to provide data to support [or not support] the development of additional specialized science high schools. In addition, we have recruited a panel of advisors (including Norman Augustine, lead author of Rising Above the Gathering Storm; Kathryn Sullivan, a member of the National Science Board until she stepped into her current position as Assistant Secretary of Commerce; and Barry Bozeman, Ander Crenshaw Professor of Public Policy at the University of Georgia) who are highly visible in the science education policy world. They have agreed to help us frame the policy implications of the study.