CALDER Policymakers Council Opinion Brief Understanding the Early Teacher Pipeline: What We Can (and, Importantly, Can't) Learn From National Data Dan Goldhaber American Institutes for Research/CALDER Kris Holden American Institutes for Research/CALDER Suggested citation: Goldhaber, D. & Holden, K. (2020). Understanding the Early Teacher Pipeline: What We Can (and, Importantly, Can't) Learn From National Data. (CALDER Policy Brief No. 21-1120). Washington, D.C.: National Center for Analysis of Longitudinal Data in Education Research. The crafting and dissemination of this opinion brief was supported by the National Center for the Analysis of Longitudinal Data in Education Research (CALDER), which is funded by a consortium of foundations. For more information about CALDER funders, see www.caldercenter.org/about-calder. We wish to thank Hannah Putman, Nell Sedransk, and Roddy Theobald for helpful comments. WARNING: this brief contains the author’s unmoderated opinions about controversial issues, which may cause dizziness, nausea, and/or seizures. Note that the views expressed are those of the authors and do not necessarily reflect those of funders or the institutions to which the authors are affiliated. Understanding the Early Teacher Pipeline: What We Can (and, Importantly, Can't) Learn From National Data Dan Goldhaber American Institutes for Research/CALDER Kris Holden American Institutes for Research/CALDER CALDER Policy Brief No. 21-1120 Understanding the nature of the early teacher pipeline, how many and what types of individuals are pursuing a teaching credential, is critically important. Unfortunately, we conclude that the two national data collections that can be used to explore the early teacher pipeline provide an incomplete and contrasting pictures of the supply of teacher candidates. Specifically, we find that Title II and Integrated Postsecondary Data System (IPEDS) information about the early teacher pipeline diverge significantly, frequently by over 30,000, in the reported number of individuals obtaining a teaching credential. Title II is a data collection explicitly intended to describe the early teacher pipeline, but it pretty clearly undercounts teacher candidates. IPEDS provides a longer window dating back to 1979–80, includes more information about race/ ethnicity and degree level, but it too provides an incomplete picture as, for instance, it likely suffers from “double counting” individuals already in the teacher workforce who obtain an advanced degree. In the concluding section we describe changes to data collection that could lead to more accurate and detailed information about the early teacher pipeline. Why We Need a National Picture of the Early Teacher Pipeline There are compelling reasons to better understand more about the early teacher pipeline—from the point at which individuals express a clear interest in teaching by taking basic licensure tests or enrolling in teacher preparation programs (TPPs) to the point at which they are fully credentialed and eligible to teach.1 Policymakers are rightfully interested in the number and types of people who are pursuing a teaching career and in how the early pipeline is influenced by teaching salaries or other workforce factors that affect the desirability of becoming a teacher. As recent evidence shows, the quality of the teacher workforce has a profound impact on individuals (Chetty et al., 2014) and society as a whole (Hanushek, 2011; Hanushek et al., 2019), as well as the challenges schools face in attracting talent into teaching (Dee & Goldhaber, 2017). There are at least four reasons why it is important to get a national perspective on the early teacher pipeline: 1. As described above, the quality of the teacher workforce has clear implications for the social and economic health of the country. And gaining a better understanding of what influences the pursuit of a teaching career is key to influencing the quality and quantity of the teacher workforce. This is particularly important today because there is evidence that interest in pursuing a career in teaching is sliding; for example, for the first time, a majority of parents surveyed by PDK International in September 2018 stated that they did not want their children to pursue teaching as a career (Kappan, 2018).2 2. Debates about the teacher pipeline often play out nationally, even while using data from selected states to make a point.3 The problem with drawing national inferences from particular states is that states may differ substantially from one another because key teacher policies (e.g., licensure, evaluation, and tenure) are state functions (Goldhaber et al., 2015). 3. Increasing attention is being placed on the diversity of the teacher workforce and the mismatch between the proportion of students and teachers of color.4 Media reports tend to emphasize hiring and retention as key drivers for a lack of teacher workforce diversity (e.g., Anderson, 2018; Meckler & Rabinowitz, 2019; Walk-Morris, 2017). While it is good to understand how these later points in the teacher pipeline are related to workforce diversity, a growing body of work indicates that teacher diversity is greatly limited prior 1 The early teacher pipeline is somewhat complicated because the timing of licensure tests and entry into TPPs varies across states. For example, most TPPs in North Carolina require teacher candidates to take a Basic Skills Test before enrolling (92%), while most TPPs in Massachusetts are not required to do so (33%). Since TPPs in most states require a Basic Skills Test, we refer to this as the earliest point of the early teacher pipeline. 2 There are contrasting views, for example, on the extent to which pursuit of a teaching credential is driven by policies in education, such as school or teacher evaluation and tenure reform (Kraft et al., forthcoming), or economic conditions (Blom et al., 2015; Nagler et al., 2020). 3 A 2015 New York Times article (Rich, 2015), for example, focused on the teacher credentialing and enrollment slowdown from 2007–08 to 2011–12 in California to sound an alarm about teacher shortages, an alarm that has, until more pressing events took hold (i.e., the COVID-19 pandemic), continued to ring in the press and policy circles. 4 Students of color compose nearly half of all public school students, while teachers of color make up less than 20% of teachers nationwide (Goldhaber et al., 2015). 1 to individuals’ employment in the teacher labor market (e.g., U.S. Department of Education, 2016, p.15), and the prospects for significantly addressing the “diversity gap” through policies that focus on current teachers as opposed to earlier pipeline prospective teachers, appear limited (Putman et al., 2016). 4. There are existing and proposed federal policies that target the early teacher pipeline at the federal level. The TEACH grant, for example, which has been in place since 2008– 09, provides federal student aid aimed at encouraging students to serve as highly qualified teachers in high-need fields for schools with low-income student populations. More recently, presidential aspirant (and now vice-presidential candidate) Kamala Harris has proposed to increase teacher salaries by 23% to close the “teacher pay gap” relative to comparable college educated professionals to increase the desirability of the teaching profession. As stated by Harris, “bright college graduates are not choosing this path of service because they need to pay their student loans (Harris, 2019).” But while there is a clear need for a national picture of the early teacher pipeline, getting a detailed national perspective turns out to be quite challenging. There are only two national sources of annual information on the number of individuals pursuing teaching: Title II reports, which are explicitly intended to get a national picture of the early teacher pipeline and thus focus narrowly on the enrollment of individuals in TPPs and whether they complete their programs’ requirements; and the Integrated Postsecondary Data System (IPEDS), which focuses more broadly on the number of college students by areas of study. As we describe below, these two data sources provide an incomplete and sometimes contradictory picture of the early teacher pipeline. Moreover, both data sets are missing information related to key policy debates about the teacher workforce. IPEDS provides a longer window of dating back to 1979-80, includes more information about race/ethnicity and degree level, but is likely prone to “double counting” individuals who are continuing their teacher education and is unclear about reporting of alternative programs. Title II is explicitly intended to describe the early teacher pipeline and likely avoids double counting problems but is an undercount of the number of individuals in the pipeline as it misses some teacher candidates who pursue out-of-state teaching positions and those who do not pass licensure tests. While it is quite likely that the true number of teacher candidates lies somewhere between the two, the discrepancy between the two datasets in counts of individuals in the pipeline is not trivial. Title II and IPEDS: History and Purpose At a high level, Title II and IPEDS appear to measure similar features of the teacher pipeline: the number of teacher candidates enrolled in and completing their education in teacher preparation programs (TPPs). But a closer inspection shows they provide related but distinct information. These data are, in fact, collected by separate agencies for different federal requirements at different times and for different purposes. Title II focuses on one specific area of postsecondary education—teacher preparation—and reports from Title II collections are “intended to provide Congress, aspiring teachers, the education community, researchers and policymakers, and the general public with information that Congress has identified as important to a basic understanding of teacher preparation in America (U.S Department of Education, Office of 2 Postsecondary Education, 2011).” As described by the National Center for Education Statistics, “IPEDS provides basic data needed to describe—and analyze trends in—postsecondary education in the United States, in terms of the numbers of students enrolled, financial aid used, staff employed, dollars expended, and degrees and certificates earned.”5 Title II data collection on teacher candidates began in 2010–11 and is collected from TPPs as authorized by Title II of the Higher Education Act. Administrators at TPPs complete surveys about their programs—including general information about location, requirements, and counts of teacher candidates who are enrolled or completing their training—and these surveys are submitted by the end of April each year. IPEDS data collection began in 1979–80 and includes surveys of administrators at every college, university, and technical/vocational institution that participates in any federal financial assistance program authorized by Title IV of the Higher Education Act. Information about teacher candidates is only part of this broad survey about higher education, which includes institution characteristics, admission processes, graduation, etc., and is collected annually in the fall, winter, and spring of each school year (the specific collections that we focus on are reported in the fall). Information from these data collections can be broadly categorized as focusing on counts of “enrollment,” i.e., those who are currently preparing for teaching; “completion,” i.e., those who have finished their preparation in the current year; and “licensed,” i.e., those who have received a teacher license in the current year. Table 1 provides an overview of how these data collections differ in terms of the specific definitions of enrollment, completion, and licensed, coverage over time, and types of statistics reported. Perhaps most notably, the data collections define enrollment and completion using different criteria. Enrollment for Title II is defined as participating in a TPP as a teacher candidate, while IPEDS reports students with a “field of study” in education.6 Completion is more complicated because for Title II reporting, TPPs set the standards by which candidates are considered to have “completed” the program, while IPEDS completion simply records the receipt of a degree or credential. So, for example, we would expect the two data collections to diverge in the case of students who are enrolled in TPPs and complete the coursework required to graduate from college but choose not to take in-state subject licensure tests. Those students will be counted as completing in IPEDS but will not be counted in Title II. Similarly, teacher candidates completing one-year TPPs are not recorded as “enrolling” in teacher education so will be missed in Title II enrollment, but not necessarily IPEDS.7 Moreover, Title II reports the number of teacher 5 U.S. Department of Education, National Center for Education Statistics. (2020). IPEDS 2020–21 data collection system. Retrieved from https://surveys.nces.ed.gov/ipeds/public/purposes-of-survey-data 6 It is challenging to identify the correct field of study to capture teacher candidates. The broad, 2-digit CIP code for Education includes many nonteaching positions (e.g., educational leadership and administration, student counseling) and training outside of K–12 ages (e.g., adult education, pre–K). We follow Kraft et al. (2018) and use the following 6-digit CIP codes to identify education degrees: Education, General (13.0100-13.0101); Bilingual, Multi-lingual, and Multicultural (13.0201-13.0299); Curriculum and Instruction (13.0301); Special Education and Teaching (13.1000-13.1099); Teacher Education and Professional Development (13.1200-13.1299 and 13.1300-13.1399); ESL (13.1401-13.1499); and Education, Other (13.9999). We think these are sensible restrictions, but the degree to which these codes exclude valid teacher candidates or include individuals who are not teacher candidates is unclear. 7 Title II reporting requires that teacher candidates are not counted as both enrolled and completing in the same year. For example, see https://title2.ed.gov/public/Webinar/FAQsQandA.pdf, questions 31 and 105. 3 licenses awarded in a given year, while IPEDS does not report licenses at all. As mentioned above, IPEDS records cover a long period of time (back to 1979–80), although not all elements are available each year.8 In contrast, Title II data is fairly recent, with enrollment and completion records starting in the 2000s. Incomplete and Sometimes Contrasting Information As suggested by Table 1, the data sets provide incomplete information related to four contemporary policy concerns: the number of individuals pursuing a teaching career and what they portends for staffing challenges faced by K–12 schools; the number of individuals pursuing a teaching career; the extent to which teacher candidates are being prepared in traditional and alternative institutions; and the diversity of teacher candidates relative to the growing disparity between student and teacher diversity. We focus mainly on the number of individuals in the early teacher pipeline, but several distinctions between Title II and IPEDS are worth noting as they relate to the latter two issues. Title II reports results only for race/ethnicity for enrollment data, while IPEDS reports race/ethnicity for both enrollment and completion data. For reasons we return to in the concluding section of this brief, we would like to know more about the race and ethnicity at both of these stages of the pipeline. Alternative preparation reporting differs substantially between Title II and IPEDS. IPEDS does not report enrollment or completions separately for alternative programs at institutions of higher education (IHEs), and due to the nature of the IPEDS survey collection, does not include enrollment or completion counts for non-IHE based alternative programs. While Title II reports enrollment and completion separately for traditional, IHE-based alternative, and non-IHE based alternative programs, defining what constitutes an alternative program is left to the state.9 This indicates that there is much ambiguity in how many teacher candidates are prepared in each pathway. We return to this point in the concluding section of this brief. Focusing now on what these datasets suggest about the number of individuals pursuing a teaching career, it is interesting to ask the degree to which the two datasets appear to align with one another in terms of counts of individuals in the early teacher pipeline. While Title II and IPEDS differ in some ways, at a high level we would expect them to provide similar information, at least in terms of teacher candidate “completions” (as discussed above).10 Figure 1 shows the reported completions of teacher candidates over time generated from each dataset. While IPEDS data covers years back to 1979–80 (Cowan et al., 2016), we present only years for which Title II completion data is available. It is clear from both datasets that the number of completions were roughly constant following the 2008 financial crisis until 2009-10. Then, both datasets show a large and relatively similar drop in completions: From 2010–11 to 2014–15, the number of completions declines by about 45,000 for Title II and by about 40,000 for IPEDS. 8 College enrollment data, for example, is available only with field-of-study information, which is essential to identify teacher candidates every two years 9 See https://title2.ed.gov/public/Webinar/FAQsQandA.pdf, p.25, question 112. 10 One additional restriction we place on the data is limiting IPEDS completions to bachelor and master degrees in an attempt to match the Title II focus on initial completions. As noted above, we follow prior work by Kraft et al. (2018) and drop CIP codes that do not pertain to K–12 teaching. 4 In Figure 2, we track this period of declining completions, reporting the percentage change in completions in each state from 2010–11 to 2014–15 using Title II data. This figure is very much aligned with the national narrative that school systems across the country have encountered severe staffing challenges (often reported as “teacher shortages”), at least partially due to declines in the enrollment in teacher preparation.11 Almost all states, 48, appear to have experienced declining completions of teacher candidates over this period. Interestingly, however, the picture is not uniform—declines are quite severe in states like Illinois and Pennsylvania, around 40%, while others like New Jersey, Utah, and West Virginia have much smaller declines in completions, in the neighborhood of 0 to 10%. This figure is important both for highlighting the fact that states varied substantially in terms of changes in the early teacher pipeline, and in setting up a comparison with IPDES data. In particular, in Figure 3 we compare the percent change in completions for Title II and IPEDS for each state. The x-axis of the figure is organized by the size of the percent change in completions for each state according to Title II (the blue dot) and the corresponding figure for the change in completions according to IPEDS (the red dot).12 Visual inspection across states suggests that there are sometimes large differences between the percentage changes in completions according to Title II and IPDES (the vertical distance between the blue and red dots), and there is little consistent pattern across states and in some cases the states provide contradictory information about the change in completions from 2010-11 to 2014-15.13 In California, for example, Title II reports a decline of about 25% while IPEDS reports an increase in completions of about 3%. More broadly, and in contrast to Figure 2, 41 states are below zero for IPEDS, indicating more increases in completions.14 Another important area in which the two datasets diverge is in the aggregate number of completions (as opposed to trends over time). In particular, returning to Figure 1, we see that the overall number of completions in each year is quite different depending on the data used to estimate it. The gap between Title II and IPEDS completions (represented by the vertical distance between the lines) varies from 15,000 to 40,000 teacher candidates per year. How large is this gap? Prior work suggests that, typically, about 100,000 to 150,000 new teachers are hired without having prior teaching experience (Cowan et al., 2016). The difference between Title II and IPEDS, then, is in the range of 10% to 40% of all positions that are filled by novice teachers each year. What might explain the differences between Title II and IPEDS in teacher candidate completions? There are several explanations for how discrepancies could arise: 11 For more background on this, see Dee and Goldhaber (2017). 12 The horizontal lines indicate the national percent change for Title II and IPEDS. 13 The correlation between the Title II and IPEDS state-level measures is only 0.26. Casual observation of Figure 3 would appear to suggest that there is more variation in IPEDS completions relative to Title II. However, this is only a result of sorting by Title II percent change; in fact, the standard deviation of these measures is 16% and 19% for Title II and IPEDS, respectively. 14 In most cases, the percent change reported by IPEDS is substantially higher than Title II. 5 • Differences could be related to the issue of double counting existing teachers who continue their higher education. IPEDS does not distinguish between teacher candidates who are seeking their first license and teachers who are currently in the teacher workforce and seeking more advanced credentials. In contrast, Title II reporting focuses on teacher candidates who are seeking initial teacher licensure. IPEDS would thus tend to inflate counts of individual teacher candidates relative to Title II.15 • Differences could be related to the way completions in alternative programs in each dataset are classified. In particular, it is not clear whether alternative IHE programs are included in IPEDS data, and IPEDS staff have informed us that the decision to include these individuals is made by the institutions themselves (IPEDS Help Desk, personal communication, February 27, 2020).16 • Differences could be due to the definition of program completion as this differs across datasets: IPEDS counts education degrees, while Title II allows programs to define completion as having passed in-state licensure examinations (Putman & Walsh, 2019). There may be some segment of individuals who receive education degrees but who do not seek an endorsement or do not pass the necessary licensure examinations. These individuals would be seen as exiting the preparation pipeline prior to completing their preparation programs according to the Title II definition, but they still would be counted as completions according to IPEDS.17 • The definition of completion in IPEDS could also undercount relative to Title II completions because, in some states, teacher candidates pursue teaching as a minor rather than a major degree area. In these cases individuals would not be counted in IPEDS. In Texas, for instance, institutions of higher education have been explicitly prohibited from offering an undergraduate degree in education.18 Not surprisingly, then, examination of IPEDS data for Texas indeed suggests very few completions in education relative to Title II (around 6,000 in IPEDS versus about 20,000 to 25,000 in Title II). • Relatedly, Title II may also underreport completions in cases where teacher candidates finish their education degrees but seek out-of-state teacher licenses. For example, a candidate who completes their coursework and student teaching in California and takes a 15 As discussed above, it is also possible that because IPEDS completions does not track individuals but rather the receipt of degrees, a student receiving two education degrees could hypothetically be counted twice. 16 Given the ambiguity around whether IPEDS completions include or exclude alternative IHE programs, we have opted to exclude alternative IHE programs in Title II completions for Figure 1. That said, if IPEDS does include alternative IHE programs, then excluding alternative IHE programs from the Title II count will tend to exaggerate this gap. When alternative programs from IHE are included in Title II data, this reduces the gap between Title II and IPEDS completions by approximately 33% to 50%, depending on the year. 17 A less likely possibility is that teacher candidates finish their TPPs but do not graduate from their universities, which would tend to overstate Title II completions relative to IPEDS. This possibility is difficult to verify directly, although Title II reports of exit requirements indicate that about 90% of TPPs require candidates to finish their coursework to be considered a completion. Consequently, this would likely affect only the remaining 10% of institutions. 18 This has recently changed with the passage of House Bill 3217; https://capitol.texas.gov/BillLookup/History.aspx?LegSess=86R&Bill=HB3217. 6 teacher licensure test for Washington, but does not take an “in-state” licensure subject test in California would likely not be counted as completing by his or her California TPP. In fact, about 75% of TPPs define completion as having passed such a test. We can explore some of these hypotheses. In particular, we can consider the extent to which inconsistency between datasets is due to double counting and the ambiguity in IPEDS about institutional reporting on alternative IHE programs. For double counting, we use the fact that double counting should disproportionately affect master’s degree completions in IPEDS data relative to bachelor’s degrees.19 We believe that double counting helps explain some of the discrepancy between Title II and IPEDS. Specifically, the NCES beginning teacher survey indicates that we should expect about 80% of new teachers to have only a bachelor’s degree, and about 17% should have a master’s degree (about 3% are reported to have either a doctorate or less than a bachelor’s; see https://nces.ed.gov/surveys/btls/cohort.asp). We find that less than 50% of IPEDS completions are bachelor’s degrees, suggesting that IPEDS completion counts include a large number of individuals already in the workforce who are returning to school to obtain a master’s degree. To explore reporting on alternative programs, we examine whether there are greater discrepancies between Title II and IPEDS completions counts in states with greater shares of completions from alternative IHE providers. The logic here is that if dataset discrepancies are driven by the discretion that IHEs have in reporting completions in alternative programs, we should see more of them in states that have a greater share of alternative IHE providers.20 Surprisingly, however, there is only a small and statistically insignificant correlation between discrepancies and the share of alternative IHE completions. Unfortunately, it is not possible to say much of anything about the latter two hypotheses— whether discrepancies are related to the definition of completion or to out-of-state mobility. Neither dataset allows for the tracking of individuals in and out of teacher education programs, nor do they follow teacher candidates across state boundaries.21 In short, beyond the crude back-of-the-envelope assessments we described above, we do not believe that one can make significant progress in terms of reconciling the differences between the two datasets in completion counts at either the national or state level. Conclusions: Getting A More Comprehensive Picture 19 Licensure requires that teachers have at least a bachelor’s degree, and in most school systems, teachers are financially rewarded if they obtain a master’s degree. Double counting could occur for individuals in the teacher workforce with a bachelor’s degree who are seeking a master’s, but it is unlikely that those in the workforce would seek a second bachelor’s degree given that they probably would not receive additional compensation for doing so. 20 We focus on IHE-based alternative programs because, as described above, IPEDS does not include completions at non-IHE based alternative programs. We consider the percent gap—e.g., IPEDS minus Title II relative to total IPEDS completions—to account for large differences in the size of states. The correlation between the percent gap and the share of IHE-based alternative program completions is -0.2, and is not statistically significant. 21 There is a small body of evidence on rates of cross-state mobility of inservice teachers (e.g., Goldhaber, 2015; Podgursky et al., 2016), but we are unaware of any quantitative research on cross-state mobility of prospective teachers at the point at which they are seeking or obtain state licensure credentials. 7 Title II and IPEDS offer a useful but somewhat limited view of the early teacher pipeline. They provide the only annual national picture of the number of individuals by state who are likely to represent the majority of tomorrow’s teachers. Yet important pieces of information are missing from each dataset. This includes information on race/ethnicity of teacher candidates in Title II and alternative program information in IPEDS. And the lack of alignment, and for some states, outright contrasting pictures, of the number of teacher candidates being prepared, is troubling. While Title II and IPEDS suggest a similar decline over time in the number of individuals in the early teacher pipeline, the picture differs drastically across some states. Examining one versus the other dataset could easily lead one to reach a different conclusion about whether the number of individuals preparing to teach is sufficient to meet the needs of a state’s K–12 school system. This of course is problematic as teacher preparation is a state function. Title II is clearly an undercount of the number of individuals obtaining credentials to teach. It is not surprising therefore that IPEDS data suggests that the number of completions is much higher than Title II. But, as we noted above, IPEDS may be an over-estimate of the new supply of potential teachers given the double counting issue. Unfortunately, it is not possible to pin down precisely why the two datasets diverge, or which one is likely to be more accurate. Additionally, some of the data that are not collected in enough detail by either dataset are relevant to a number of teacher pipeline policy debates and discussions related to, for instance, the diversity of the teacher workforce (Goldhaber et al., 2019) and the degree to which teacher training is occurring in traditional or alternative settings (Zeichner & Schulte, 2001). Surprisingly, we do not really know the basic national numbers about diversity (or lack thereof) in the early teacher pipeline. For example, a great deal of the lack of teacher diversity appears to happen as teacher candidates exit the early pipeline (U.S. Department of Education, 2016), which could be related to the fact that teacher candidates of color are less likely to pass licensure tests that TPPs may require for classifying a teacher candidate as a completion (Putman & Walsh, 2019). However, as described above, Title II reports do not include data for completions by race/ethnicity that are required to contrast the diversity of enrolled candidates with the diversity of completing candidates. Similarly, the definition of alternative programs is murky because, as described in a National Academies of Sciences (NAS) report (NAS, 2020), “even at a national level, accurately gauging how many teachers are coming into the profession through different routes is difficult due to the fact that each state defines for itself what constitutes a traditional or alternative route (U.S. Department of Education, 2019).” Indeed these two issues are likely intertwined given that a significant (albeit unknown at the national level) share of teachers of color are prepared in alternative programs (U.S. Department of Education, 2016). Resolving discrepancies and getting a clear national (and state-by-state) perspective on what the early teacher pipeline looks like is not complicated and may not be terribly costly. Many of the concerns discussed in this brief could be addressed by slightly modifying IPEDS and/or Title II survey collection efforts. IPEDS survey questions could provide clear direction to institutions about whether alternative programs should or should not be included, and the scope of IPEDS surveys could be expanded to cover non-IHE alternative preparation programs. IPEDS could also collect information on whether individuals are continuing their education or pursuing their first 8