# Teach For America

GiveWell aims to find the best giving opportunities we can and recommend them to donors. We tend to put a lot of investigation into the organizations we find most promising, and de-prioritize others based on limited information. When we decide not to prioritize an organization, we try to create a brief writeup of our thoughts on that charity because we want to be as transparent as possible about our reasoning. The following write-up should be viewed in this context: it explains why we determined that (for the time being), we won't be prioritizing the organization in question as potential top charity. This write-up should not be taken as a "negative rating" of the charities. Rather, it is our attempt to be as clear as possible about the process by which we came to our top recommendations.

The last time we examined the charities working primarily in the U.S. was in 2010. As of 2011, we have de-prioritized further work on this cause. The content we created in 2010 appears below. This content is likely to be no longer fully accurate, both with respect to what it says about the organization and with respect to what it implies about our own views and positions.

Published: 2010

Previous GiveWell reviews of TFA:

## What do they do?

Teach For America (TFA) recruits recent college graduates,1 facilitates their training,2 and places them in schools across the country, where they teach while completing coursework to earn full teaching certification.3 TFA aims to recruit high-performing and ambitious students who have graduated from top universities and "who are highly likely to be leaders in all sectors of society in the course of their careers."4 TFA seeks to hire students with strong academic records, regardless of prior teaching experience.5

## Does it work?

### Evidence of impact

There are many studies of TFA's impact,6 only one of which is a randomized controlled trial.7 These can be separated into (a) studies assessing TFA teachers' short-term impact on student achievement, generally by examining scores on standardized math and reading exams, and (b) studies of what participants do after they complete the TFA program.

#### Short-term impact

Only one of the studies of TFA's impact on student academic achievement that we've reviewed uses a randomized controlled trial design, and this study appears to be a high-quality study examining TFA across several of its regions. For reasons laid out here, we tend to rely on randomized controlled trials as evidence where they are available. For the case of TFA, we do not discuss the results of the non-experimental studies. The randomized controlled trial -- Decker, Mayer, and Glazerman (2004) -- compares test scores of students in grades 1-5, some of whom (by random assignment) had TFA teachers and some of whom did not. The study evaluated 100 teachers' classes distributed over 6 of TFA's urban and rural program regions (Baltimore, Chicago, Los Angeles, Houston, New Orleans, and the Mississippi Delta).8 The students in each group were comparable on observable characteristics such as sex, racial composition and free-lunch eligibility; in addition, baseline test score comparisons and the experiment's design (using random assignment) imply that students placed in TFA teachers' classrooms did not differ systematically from students in non-TFA classrooms, prior to their experiences with different teachers.9 Students were tested in the beginning and end of the school year (the study thus investigated the impact of having a TFA teacher for one school year).10 At the end of the year, students in classes taught by TFA corps members had improved their math standardized test scores more than students in other classes and the difference in improvement between the two groups was statistically significant; the advantage gained by students in TFA classrooms corresponded roughly to having an additional month of instruction.11 In other words, the study found that having a TFA teacher for one school year had an impact of 2.4 normal curve equivalents (NCEs),12 corresponding to an effect size of 0.15 standard deviations, or about 10 percent of a grade equivalent.13 On the reading exam, students in TFA classes outperformed others by only half a NCE (0.03 standard deviations), a non-statistically significant result.14 The study also found no statistically significant impact on the probability that students were held back or sent to summer school.15 It found that student discipline, absenteeism, and negative behavior either remained constant or increased in TFA teachers' classrooms16 (although higher rates of poor behavior in TFA teachers classrooms may have been a result of the greater likelihood that TFA teachers would report such behavior).17 We note the following possible concerns about the study:
• While we know that regions and schools were randomly selected, we don't know how school districts were selected within each region. If the districts chosen had particularly poor teachers, the study could overestimate TFA teachers' effectiveness in general (and likewise, if they had particularly good teachers the study could underestimate TFA teachers' effectiveness).
• The authors note that there was no difference in the rates at which students left TFA teachers' classrooms vs. non-TFA teachers' classrooms: about 87% of students in both groups remained all year, with about 4% switching from non-TFA to TFA (or vice versa), and 9% leaving the school entirely.18 However, this still leaves open the possibility that students leaving TFA classrooms were worse performers than those leaving non-TFA classrooms, which could account for asymmetric score increases.
However, these concerns are relatively minor and we see good reason to believe that that TFA teachers are roughly on par (in terms of short-term impact on academic performance) with the teachers they replaced, and perhaps slightly better (particularly in math).

#### Long-term impact

One of Teach For America's goals is to increase its corps members' interest in education, and increase the likelihood that they stay in the field or improve education through positions in other sectors. Related publications provide information such as prominent examples of alumni,19 how long alumni stay in education,20 civic attitudes and behavior of alumni,21 and how many KIPP principals and teachers are TFA alumni.22 We're encouraged by the fact that this data is available, but ultimately don't find it compelling as evidence of impact. We would guess that students who apply to TFA are already more interested in education and public service than average, and don't find any of this data to make a compelling case that TFA itself has had a long-term impact on their interest. We are also ambivalent as to whether shifting more talented people into U.S. education would be a good thing, assuming TFA were succeeding at it. (For example, we feel that international aid offers more promising opportunities to donors, and may benefit more from human capital as well.) We recognize that this is very much a judgment call, and encourage donors who disagree with our perspective to consider the long-term potential of Teach For America to recruit talented people into U.S. education.

#### Interpretation

Weighing the empirical evidence, we would guess that TFA is succeeding in recruiting and training teachers whose personal qualities (intelligence, motivation, etc.) are enough to compensate for their relative lack of experience and certifications, with the result that these teachers are as good as - or better than - those they're replacing. However, while this accomplishment may be impressive, we question how much it is worth in terms of creating significantly better outcomes for disadvantaged students. We share Prof. Richard Murnane's concern that the teachers replaced by TFA members may be "remarkably ill prepared to educate children and especially children needing the nation's best teachers;"23 thus, recruiting teachers who are as good or slightly better - and dispersing these teachers across disadvantaged schools - may have only a small and dispersed impact on disadvantaged students. We feel that TFA is one of the most transparent and well-evaluated groups working in the field of education, with the result that there is a large body of evidence about its work whose interpretation requires substantial judgment calls. We strongly encourage donors interested in this area to consider TFA and the evidence we have discussed here.

### Ongoing monitoring

We are still in the process of assessing TFA's ongoing monitoring of its work.

## What do you get for your dollar?

The following table shows the yearly breakdown of TFA's total expenses divided by their entering corps sizes.24 Note that the figures cited do not include the teachers' salaries, which TFA does not pay.25
Fiscal yearExpenses# of incoming corps members who start teachingCost per incoming corps member who start teaching
2005$40,931,6962,176$18,811
2006$56,203,7432,417$23,254
2007$83,392,2982,892$28,836
2008$124,550,6963,613$34,473
2009$155,035,5704,075$38,046
It appears that as TFA has grown, its cost per incoming corps member has risen significantly. A TFA representative emailed us the following regarding the increase in cost per incoming corps member:26
• Between 2006 and 2009 we invested in improving corps member and alumni effectiveness while growing our corps by 18% annually, and took steps to ensure our organizational health at increased scale. These investments set the stage for efficiencies as we scale from 2010-2015.
• Between 2009-11, cumulative per corps member leadership investment steadied at 2% as we realized efficiencies. Beginning in 2010, we realized \$2-3 million annual savings in conferences, communications, shipping, and other areas as well as a nearly 20% reduction in cost per recruit. We anticipate continued efficiencies in 2011 as we automate back-office services.
• For the next four years, we are planning for 1% annual investments in quality while we continue to seek efficiencies. We will continue investing in program effectiveness, but we also intend to reduce costs through innovation, technology investments, and scale.

## Room for more funding

TFA gives a general outline of its plans for the future,27 but we do not have a concrete sense of how it would use additional donations. A TFA representative emailed us the following regarding their room for more funding:28
We have launched an exciting campaign for our 2015 strategic plan that outlines our priorities for the next five years. (the growth plan on the website ends in 2010). Over the next five years, we aim (1) to double in the size of our incoming corps while increasing diversity, (2) increasing the effectiveness of our teachers, and (3) fostering leadership among some 44,000 alumni (4) while continuing to building an enduring institution.

## Sources

• 1. “About Teach For America Teach For America is the national corps of outstanding recent college graduates who commit to teach for two years in urban and rural public schools and become lifelong leaders in expanding educational opportunity.” Teach For America, “Teach For America Announces the Schools Contributing the Most Graduates to Its 2010 Teaching Corps.” “Corps Recruiting and Admissions Teach For America recruits top college graduates of all academic majors, career interests, and backgrounds who demonstrate achievement, leadership, and a commitment to expanding opportunity for children in low-income areas.” Teach For America, “About Teach For America.”
• 2. "Teach For America provides a comprehensive two year program of training and support." Teach For America, "Training and Support."
• 3. "Certification requirements and length varies across our regions. In most cases, the ongoing coursework that corps members take as part of their region's alternative teaching program leads to full certification by the end of the two years. In other cases, corps members need to take additional coursework beyond their two-year commitment to become fully certified. In all cases, corps members are responsible for completing and submitting all coursework and certification program-related deliverables on time and for fulfilling all financial obligations to the certifying agency or institution prior to receiving their formal certification, or as agreed upon between the corps member and relevant entity." Teach For America, "Certification."
• 4. “We target the most highly sought college students of all academic majors and career interests from top colleges – the future leaders of America – who are deeply committed to our mission and who bring the necessary mindset and skills to drive effectively toward our vision.... We select individuals who are highly likely to be leaders in all sectors of society in the course of their careers, both within and outside of education. Those selected to enter Teach For America's corps are among the most highly sought-after graduating college seniors. Specifically, the first-year corps members who began teaching in New York City last fall had an average GPA of 3.57, an average SAT score of 1324, 94% held at least one leadership position in college.” Teach For America, "Clear Fund Grant Application," Pg 4.
• 5. “Corps Recruiting and Admissions Teach For America recruits top college graduates of all academic majors, career interests, and backgrounds who demonstrate achievement, leadership, and a commitment to expanding opportunity for children in low-income areas.” Teach For America, “About Teach For America.”
• 6. The studies we have considered for our review of TFA are:
• Boyd et al. (2005).
• Boyd et al. (2007).
• Boyd et al. (2009).
• Darling-Hammond et al. (2005).
• Decker, Mayer, and Glazerman (2004).
• Donaldson (2008).
• Donaldson and Johnson (2010).
• Henry et al. (2010).
• Kane, Rockoff and Staiger (2006).
• Laczko-Kerr and Berliner (2002).
• Mac Iver and Vaughn (2007).
• Noell and Gansle (2009).
• Raymond, Fletcher, and Luque (2001).
• Teach For America, "Alumni Social Impact Report (2008)."
• Teach For America, "Alumni Social Impact Report (2009)."
• Teach For America, "National Principal Satisfaction With Teach For America Teachers."
• Xu, Hannaway, and Taylor (2007).
The literature review Heilig and Jez (2010) reviews TFA's evidence of effectiveness. TFA also links to and summarizes literature at Teach for America, "Research on Teach For America."
• 7. Decker, Mayer, and Glazerman 2004.
• 8. “We randomly assigned students to classrooms in order to ensure that the TFA and control teachers have essentially identical classes of students.” Decker, Mayer, and Glazerman 2004, Pg 7. “To facilitate the use of random assignment, our study included only elementary students (grades 1 to 5).... We first conducted a pilot study in one region—Baltimore—during the 2001-2002 school year; then conducted a full-scale evaluation during the 2002-2003 school year in five additional regions—Chicago, Los Angeles, Houston, New Orleans, and the Mississippi Delta. The sample includes 6 of the 15 regions where TFA placed teachers at the time the study was being designed.... The final research sample, which is summarized in Table III.1, consisted of 17 schools, 100 classrooms, and nearly 1,800 students.” Decker, Mayer, and Glazerman 2004, Pg 8.
• 9. “An important feature of the study is the use of random assignment to produce equivalent groups of students across classrooms within each block (grade within school). Table V.3, which compares the average baseline characteristics of students in TFA (treatment) and non-TFA (control) classes, shows that random assignment did indeed produce equivalent groups in terms of demographic characteristics, baseline test scores, and class characteristics. All of the treatment-control differences were small and none is statistically significant [p-values for differences between the treatment and control groups at baseline range from 0.400 to 0.998].” Decker, Mayer, and Glazerman 2004, Pg 25.
• 10.
• “The evaluation was conducted in two stages. We first conducted a pilot study in one region—Baltimore—during the 2001-2002 school year; then conducted a full-scale evaluation during the 2002-2003 school year in five additional regions—Chicago, Los Angeles, Houston, New Orleans, and the Mississippi Delta. The sample includes 6 of the 15 regions where TFA placed teachers at the time the study was being designed.” Decker, Mayer, and Glazerman 2004, Pg 8.
• “Our measures of student achievement were based on standardized mathematics and reading test scores. Using the Iowa Test of Basic Skills (ITBS), we administered a baseline achievement test in the fall and a follow-up test in the spring in each of the classes included in the study.” Decker, Mayer, and Glazerman 2004, Pg 9.
• “The impact estimates are based on scores from tests we administered at the end of the school year, accounting for any preexisting differences based on the test we administered at the beginning of the school year.” Decker, Mayer, and Glazerman 2004, Pg 29.
• 11. “The positive impact of TFA on math scores is statistically significant, but is it large enough to imply that TFA teachers produce meaningfully greater math achievement? When expressed in grade equivalents, the math achievement advantage TFA teachers offered appears to be meaningful. The impact translates into about 10 percent of a grade equivalent, suggesting that the advantage to TFA students corresponds roughly to an additional month of instruction.” Decker, Mayer, and Glazerman 2004, Pgs 30-31. According to Decker, Mayer, and Glazerman (2004, Table VI.1, Pg 32), the impact on math scores was statistically significant at p = 0.002.
• 12. “Normal curve equivalent (NCE) scores were created to alleviate the problem of unequal interval scales presented by percentile ranks. This equal interval scale has a mean of 50, and range of 1 to 99, just like NPR [national percentile rank]. A standard deviation of 21.06 forces the intervals to be equal, spreading the scores away from the mean. NCEs have the same meaning across students, subtests, grade levels, classrooms, schools, and school districts. Fifty (50) is what one would expect for an average year's growth.” Bernhardt 2006, Pg 208.
• 13. “All calculations in this report used normal curve equivalent (NCE) scores, which are translated into percentile rankings for ease of interpretation.... We used regression methods to adjust for any background differences between treatment and control groups that might remain after random assignment. Table VI.1 shows the resulting impact estimates. We report all impact estimates of NCEs, which are scaled so that a nationally representative population has a mean of 50 and standard deviation of 21.06. Using this metric, the impact on math achievement is 2.4 NCEs, which is significantly different from zero. The same impact can be expressed in a different type of unit known as an “effect size.” An effect size is the fraction of a standard deviation in the underlying measure (test score) and is a popular metric for comparing results of studies that use different outcome measures. An impact on mathematics scores of 2.4 NCEs corresponds to an effect size of 0.15, or equivalently, 15 percent of a standard deviation. The positive impact of TFA on math scores is statistically significant, but is it large enough to imply that TFA teachers produce meaningfully greater math achievement? When expressed in grade equivalents, the math achievement advantage TFA teachers offered appears to be meaningful. The impact translates into about 10 percent of a grade equivalent, suggesting that the advantage to TFA students corresponds roughly to an additional month of instruction. Comparisons with other evaluation findings also suggest that the TFA impacts on math achievement are meaningful. An often-cited benchmark for assessing impacts on education performance is the effect of reducing elementary school class size from an average of 23 to 15 students, which has been reported to have a single-year effect size of about 0.23, based on a large-scale experimental study in Tennessee (Finn and Achilles 1999). Therefore, when compared with the effect of reduction in class size, the magnitude of the TFA impact on math scores—an effect size of 0.15—is about 65 percent of the effect of a reduction in class size of eight students.” Decker, Mayer, and Glazerman 2004, Pgs 29-31.
• 14. “TFA teachers did not have an impact on average reading achievement.” Decker, Mayer, and Glazerman 2004, Pg xv. “The estimated impact on reading scores, also shown in Table VI.1, was very close to zero and was not statistically significant. The point estimate of 0.56 NCEs corresponds to an effect size of 0.03.” Decker, Mayer, and Glazerman 2004, Pg 31. According to Decker, Mayer, and Glazerman (2004, Table VI.1, Pg 32), the estimated impact on reading scores was 0.56 normal curve equivalents (NCEs) and not statistically significant (p = 0.37).
• 15. “The differences between TFA and control students—less than one percentage point in grade retention (TFA students being held back with slightly greater frequency) and less than one percentage point in summer school attendance—were not statistically significant [p-values 0.536 and 0.884].” Decker, Mayer, and Glazerman 2004, Pg 43.
• 16. “Through objective data from the school records and from teacher impressions, we were able to estimate the impact of TFA on a variety of outcomes related to classroom management. The evidence is inconclusive on whether TFA teachers had an easier or more difficult time than their colleagues in managing their classrooms.” Decker, Mayer, and Glazerman 2004, Pg 44. According to Decker, Mayer, and Glazerman (2004, Table VII.2, Pg 44), the TFA students had somewhat more absence and disciplinary incindents but none of these differences were statistically significant at at p â‰¤ 0.05 (p-values ranged from 0.177 to 0.794). According to Decker, Mayer, and Glazerman (2004, Table VII.3, Pg 45), TFA teachers reported more classroom problems than other teachers but none of these differences were statistically significant at p â‰¤ 0.05 (two were statistically significant at p â‰¤ 0.1).
• 17. “There are at least two potential explanations for the findings of no impacts based on school reported outcomes and potentially harmful impacts based on teacher self-reports. One is that TFA teachers had different expectations and perceptions than control teachers about student behavior, which could lead them to interrupt the class more often for disruptive students or be more prone to describing their students' behavior as problematic. This explanation seems plausible, since, as we have demonstrated, TFA teachers and control teachers come from substantially different backgrounds before teaching. Another possibility is that TFA teachers actually had more difficulty managing their classrooms, which resulted in an objective increase in physical conflicts, verbal abuse, and disruption of class time. Because the results presented here on classroom management are inconclusive, further research is needed to fully understand the impacts of TFA on student behavior in the classroom.” Decker, Mayer, and Glazerman 2004, Pg 45.
• 18. Decker, Mayer, and Glazerman 2004, Pgs 26, 27, 48.
• 19. Teach For America, "Alumni Social Impact Report (2008)." Teach For America, "Alumni Social Impact Report (2009)." Kelly 2006.
• 20. Boyd et al. 2009. Donaldson 2008. McAdam and Brandt 2009. Noell and Gansle 2009. Teach For America, "Alumni Social Impact Report (2008)." Teach For America, "Alumni Social Impact Report (2009)."
• 21. McAdam and Brandt 2009.
• 22. "MF: KIPP grew out of Teach For America. Dave Levin and I started KIPP after completing our commitment to TFA in 1994. Currently 60% of KIPP principals and 33% of KIPP teachers are TFA alumni." Russo 2006.
• 23. Murnane 2004, Pg 1.
• 24. TFA's total expenses are from TFA's 990 forms:
• Teach For America, "IRS Form 990 (2005)," Pg 1.
• Teach For America, "IRS Form 990 (2006)," Pg 1.
• Teach For America, "IRS Form 990 (2007)," Pg 1.
• Teach For America, "IRS Form 990 (2008)," Pg 1.
• Teach For America, "IRS Form 990 (2009)," Pg 1.
The # of incoming corps members comes from TFA's annual reports:
• Teach For America, "Annual Report (2006)," Pg 5. (Has figures for FY 2005 and FY 2006.)
• Teach For America, "Annual Report (2007)," Pg 5. (Has figures for FY 2007 and FY 2006; the FY 2006 figure (2,417) differs slightly from the FY 2006 figure in the 2006 Annual Report (2,426).)
• Teach For America, "Annual Report (2008)," Pg 6.
• Teach For America, "Annual Report (2009)," Pg 5.
• 25. “Corps members receive the same salaries and health benefits as other beginning teachers, and they are paid directly by the school districts for which they work.” Teach For America, “Financial Arrangements.”
• 26. Danille Giusto, email to GiveWell, November 20, 2010.
• 27. Teach for America, "Our Growth Plan."
• 28. Danille Giusto, email to GiveWell, November 20, 2010.