Special Education Teacher Shortage: Differences Between High and Low Shortage States

Abstract

In this study, using Office of Special Education Programs (OSEP) personnel data from 2006 to 2014, we identified seven states with consistently low shortages of highly qualified special education teachers and seven states with persistently high shortages. We employed Guarino et al.’s framework to guide our assumptions and selection of demographic, supply, and demand variables and compared two groups in this descriptive analysis. We found significant differences across supply and demand variables. Low shortage states make greater investments in per pupil expenditures; have higher teacher salaries, generally; have greater preparation capacity; and produce more special education graduates. Taken together, our findings suggest that special education teaching is a relatively better job in low shortage states than in high shortage states. We situate the discussion of our findings within policy recommendations that states may use to address shortages. Limitations of our analysis are addressed, and implications for future research are proposed.

Keywords

special education teacher shortage supply and demand fully certified certification/licensure teacher preparation policy/service delivery educational policy

An insufficient supply of fully qualified special education teachers (SETs) undermines the Individuals with Disabilities Education Act (IDEA, 2004) promise of an appropriate and individualized educational program for all students with disabilities (SWDs). Although the SET shortage declined during the Great Recession through 2012 (Dewey et al., 2017), shortages have increased since then to 6.8%, leaving approximately 23,000 positions in special education without a qualified teacher. These vacancies appear to be related to geography (Levin et al., 2015), disability category (Katsiyannis et al., 2002), school setting (Mason-Williams et al., 2017), and school characteristics (Mason-Williams, 2015). Shortages have been attributed to many factors, including high SET attrition rates (Goldring et al., 2014) and an insufficient supply of new teachers entering the workforce from educator preparation programs (EPPs). Regardless of its provenance or extent, the SET shortage undermines the promise of IDEA.

National Evidence of the Limited Supply and Growing Demand

National SET employment steadily increased as the number of SWDs increased (Boe, 2006). From 1987 to 2003, the number of SETs reached 400,000 (Boe, 2006), paralleling the growth in the number of SWDs in U.S. schools. At the same time, the number of fully certified SETs grew commensurately until 2000, when the SET shortage—defined originally as the discrepancy between all SETs and all fully certified SETs—began to grow, reaching 12.6% by 2002 (Boe, 2006). Shortage of this scale is concerning given that SWDs have been shown to benefit when their teachers are fully prepared (Feng & Sass, 2013) and preparation yields improved teaching performance (Nougaret et al., 2005; Sindelar et al., 2004).

Replicating methods used by Boe (2006), Dewey et al. (2017) reported national demand¹ for SETs continued growing until 2005, when the number of SETs exceeded 420,000. Beginning in 2006, however, SET employment declined steadily to roughly 346,000 in 2012 (Dewey et al., 2017). This decline appeared to be fueled by decreased numbers of students eligible (particularly students with learning disabilities), a redistribution of school funding from special education to general education, and the impact of the Great Recession in 2008 on school budgets. Ironically, the decline in SET employment had an upside: It led to increases in the proportion of fully certified (now labeled highly qualified) SETs, which peaked in 2012 at just over 95%. Since then, as the U.S. economy emerged from recession, schools have hired teachers and SWD prevalence has grown—but SET employment has not changed commensurately (Sindelar, Bettini, et al., 2018). In fact, in 2014–2015, the number of highly qualified SETs had dropped to 340,000, a level not seen since the early 1990s. Apparently, despite increased demand because of recent growth in SWD prevalence, SET supply has continued to lag and many SWDs are still taught by individuals who fail to meet full professional standards.

The Role of Policies on Supply and Demand

A complicated network of national and state policies influences the supply of SETs. Although federal policy plays a small role in SET supply, state-level policies impact SET supply and demand more directly. For instance, state-level certification policies based on broad certification categories (i.e., cross-categorical or covering grades K-12) allow all new candidates to be eligible for all positions (Sindelar, Fisher, & Myers, 2018). Furthermore, variations in states’ workforce policies may lead to higher teacher pay, other incentives for entering teaching, and better benefits packages, impacting supply by making special education teaching a more attractive professional option (Aragon, 2016; Feng & Sass, 2018a).

Although investigations of supply and demand data aggregated nationally offer few insights about how to address shortages, states may benefit from an examination of the state-level contexts that influence supply and demand more directly. For instance, the availability of graduates from special education programs directly influences the supply. In addition, state-level policy solutions proffered to address this problem may include better pay and benefits, forgivable loans, improved administrative support and working conditions, and pension portability, to name only a few (for a comprehensive overview, see Kolbe & Strunk, 2012; Sutcher et al., 2016). Aragon (2016) and Dee and Goldhaber (2017) emphasized the importance of focusing such initiatives on disciplines and school contexts where shortages are most severe, to increase cost-effectiveness. For example, Cowan et al. (2016) found that focused bonuses and loan forgiveness programs proved successful in reducing attrition of STEM (science, technology, engineering and mathematics) and SETs (Feng & Sass, 2018), especially in some high-poverty schools (Clotfelter et al., 2008). Nevertheless, recent research by Feng and Sass (2018) has found the degree to which these programs were effective across teacher shortage groups was dependent upon the size of the fiscal investment. Specifically, payments of $2,500 were necessary to retain SETs, compared with $500–$1,000 in other shortage areas (e.g., ESL [English as a second language], STEM).

As Dee and Goldhaber (2017) noted, many recommended policy solutions have a downside: funding. Even successful policy approaches (e.g., induction) may lie out of reach due to political and funding realities. For example, although rigorous induction has been shown to stem attrition (and reduce demand; Smith & Ingersoll, 2004), adequate funding may lie beyond a state’s political or economic reach. In this sense, state and district policymakers may find themselves in the awkward position of having an abundance of ideas about how to solve the problem of teacher shortage but lacking the wherewithal to make them happen. For this reason, targeted and strategic solutions, such as those recommended by Cowan et al. (2016), designed intending to minimize cost, are a necessity.

State governments are well positioned to assess the extent of shortages, navigate the vagaries of regional differences, and carry out effective policies. Therefore, it appears valuable to compare state-level characteristics that may either directly or indirectly affect SET supply. Thus, the focus of the current study is to examine contextual and policy-malleable differences between states with relatively and consistently low and high SET shortages. We founded our analysis on the assumptions of critical supply and demand variables rooted in the teacher labor market literature (Lovenheim & Turner, 2018).

Conceptual Framework

To frame our comparison of state characteristics and policies related to SET shortages, we draw on Guarino et al.’s (2006) review of the empirical literature on teacher recruitment and retention practices. Employing an economic frame of reference, Guarino et al. (2006) assessed the effectiveness of recruitment and retention policies based upon the concept of opportunity cost for selecting teaching as an occupation relative to all other available occupations. Furthermore, Guarino et al. (2006) nest the teaching profession within a larger labor market, in direct competition with all other occupations “requiring similar levels of education or skill” (p. 175). To capture the dynamic of the teacher market, they operationalize supply and demand as it applies to teaching and subsequent policy levers. They define demand as the number of teaching positions offered at a given level of compensation, impacted by policies concerned with student enrollments, class-size targets, and teaching load norms. Supply, or attractiveness, is the number of qualified individuals willing to teach at a given level of overall compensation. Policies that address ease of entry and overall compensation (e.g., salaries, benefits, working conditions) can manipulate supply and drive demand. Conceptually, teacher shortage is the difference between demand and supply of teachers, mitigated in part by associated policy levers (Lovenheim & Turner, 2018).

We use this heuristic to shape our comparative analysis of states with consistently low and persistently high SET shortages. Although researchers have reported findings that suggest SETs respond differently to recruitment and retention policies versus science, math, and teachers of students who speak English as a second language (Feng & Sass, 2018), there is little evidence to suggest that entirely different policies are required. Thus, Guarino and colleagues’ framework is an appropriate starting point for our analysis. We posit that states with low SET shortages will have more favorable supply policies (i.e., higher salaries), labor market conditions, and infrastructure (e.g., preparation capacity) in place to address demand, all else being equal. Through this analysis, we hope to discern differences that would suggest state-level policy strategies for addressing SET shortage. In the following sections, we (a) categorize state-level variables into supply and demand categories, (b) articulate our assumptions about each, (c) situate our assumptions within the literature of teacher shortage, and (d) draw comparisons between low and high shortage states across each variable. Our investigation focused on three research questions:

Research Question 1: How do high and low SET shortage states differ in terms of demographics?

Research Question 2: How do high and low SET shortage states differ on variables that affect SET supply (i.e., number of new special education graduates)?

Research Question 3: How do high and low SET shortage states differ on variables that drive demand (i.e., student enrollments, teacher/student ratios, per-pupil expenditures, mobility, and commitment)?

Taken together, these questions allow us to address a more general question: Is special education teaching a more attractive occupation in low shortage than high shortage states?

Method

The purpose of this investigation was to identify differences between high and low SET shortage states that might suggest state-level policy strategies for coping with shortage. For over 30 years, OSEP has collected and reported the number of SETs employed by state, differentiating between those who are highly qualified and those who are not. (Until 2006, they used the passage of No Child Left Behind, fully certified and not fully certified.) Cook and Boe (2007) defined quality shortage as the proportion of SETs who, according to OSEP, are not highly qualified. Thus, states with severe shortages are those with a high percentage of SETs who do not meet the highly qualified standard (of being enrolled in a high-quality alternative route preparation program). Subsequently, in this paper, we use the term shortage to refer to the Cook and Boe (2007), OSEP-related quality shortage definition. Given this definition, we identified high and low shortage states and used data collected by several federal agencies (e.g., OSEP, NCES [National Center for Education Statistics]) to describe the demographics of relevant states, as well as variables related to supply and demand. In this section, we describe the process for identifying high and low shortage states, the variety of data sources used, and variables we derived from the data.

Identifying High and Low Shortage States

We identified high and low SET shortage states with a three-step process. Using state-level OSEP data from 2006 to 2015 (the most recent year available at https://www2.ed.gov/programs/osepidea/618-data/static-tables/index.html#partb-pen) and focusing only on the 50 states, we defined SET shortage as the percent of SETs (for SWDs aged 6–21) considered not highly qualified. We first examined state total SET counts year by year to identify irregularities (e.g., missing data, unexplained spikes or dips) not readily explained by national policy or other system shocks. For example, the introduction of the highly qualified teacher requirement and the change in the OSEP unit from not fully certified to not highly qualified were associated with declines in fully certified SETs in 2006. Furthermore, SET employment dropped universally in 2008 with the onset of the Great Recession. In states where data were missing for a single year, we interpolated missing values. If a state had missing data for consecutive years or interpolation involved use of anomalous data points, we opted to eliminate the state from consideration (n = 22).

To ensure there were no significant differences between those with anomalous data points and those with consistent data points, we performed Mann–Whitney U test to assess differences between our samples across four demographic areas: (a) population, (b) population density, (c) the percentage of White school-age children, and (d) per capita gross domestic product (GDP). We found no significant differences between the included sample of states (M_ss rank = 25.11) and those not included (M_NI = 26.00), U = 297, p = .83 with regard to population; percent of White school-age children (M_ss rank = 25.66, M_NI = 25.30), U = 303, p = .930; population density (M_ss rank = 27.04, M_NI = 23.55), U = 265, p = .401; and per capita GDP (M_ss rank = 28.29, M_NI = 21.95), U =230, p = .127.

In the second phase, we parsed the remaining states into high and low shortage groups. Low shortage states included those consistently below the annual national average with no more than 2 years at or above the national average (n = 21). High shortage states included those whose shortages persistently exceeded national averages (n = 7). We then ranked states based on their deviation from the national average and chose the seven most extreme for the low shortage group. This created groups for which differences in variables of interest might be most readily observed. The national average of not highly qualified SETs ranged from a high of 11% in 2006 to a low of 5% in 2011. We identified states that deviated every year by at least 3% above or below the annual national average to define our groups. We display these shortage trends in Figure 1. The seven states with consistently lower annual percentages of not highly qualified SETs included New Hampshire (NH), Connecticut (CT), Pennsylvania (PA), Michigan (MI), Kentucky (KY), Illinois (IL), and North Dakota (ND). The seven states with persistently higher annual percentages of not highly qualified SETs included West Virginia (WV), Maryland (MD), Kansas (KS), Utah (UT), Nevada (NV), Alaska (AK), and Hawaii (HI).

Figure 1.

Trends of uncertified SETs from 2006 to 2015 as reported by OSEP in their annual report to congress in high and low shortage states as in relation to national trends.

Data Collection

To capture differences in supply and demand variables between high and low shortage states, we identified variables from among accepted measures within the teacher labor market literature (Lovenheim & Turner, 2018). Teacher labor markets and subsequent supply and demand variables operate within each state context. In the following section, we summarize each construct of interest and identify the variables selected for examination and the rationale for inclusion.

Data sources

We focused data collection on demographic information (Table 1), supply (Table 2), and demand variables (Tables 3 and 4), primarily collected in the year 2014 (to correspond with the OSEP SET counts). When data were not available for 2014, we opted for years in closest proximity. We drew demographic information from various data sources collected by the Bureau of Labor Statistics, including the American Community Survey and the Council for Community and Economic Research. Several national data sets derived from the NCES and the Office of Special Education Programs (OSEP) to support our analysis of supply and demand variables. Specifically, we extracted data from the Elementary and Secondary Information System (ELSI), the Integrated Postsecondary Data System (IPEDS), and the School and Staffing Survey (SASS). In the following section, we provide a broad overview of the data collected and detailed descriptions of the derived variables we created for our analysis.

Table 1.

U.S. and State Context Variables.

States	Population estimate^a	% White (school-aged)^b	Population density (people per sq. mile)^a	Per capita GDP
USA	6,252,099	66.6	87	$49,329
Low shortage
CT	3,592,053	65	287	$70,115
IL	12,868,747	52	90	$57,370
KY	4,383,272	81	43	$43,541
MI	9,889,024	70	68	$49,331
NH	1,321,069	90	57	$66,382
ND	704,925	84	4	$55,617
PA	12,758,729	71	110	$53,357
Median	4,383,272	71	68	$53,357
High shortage
AK	728,300	52	1	$72,083
HI	1,392,704	16	85	$68,858
KS	2,882,946	69	13	$52,038
MD	5,887,776	44	237	$74,350
NV	2,761,584	40	10	$53,588
UT	2,858,111	78	14	$60,056
WV	1,853,881	82	30	$41,663
Median	2,761,584	52	14	$60,056

Note. GDP = gross domestic product; CT = Connecticut; IL = Illinois; KY = Kentucky; MI = Michigan; NH = New Hampshire; ND = North Dakota; PA = Pennsylvania; AK = Alaska; HI = Hawaii; KS = Kansas; MD = Maryland; NV = Nevada; UT = Utah; WV = West Virginia.

2014 U.S. Census Bureau. ^bAmerican Community Survey 5-Year Estimates, 2013.

Table 2.

Variables Related to the Supply of Teachers in Low and High Shortage States.

States	Ratio #produced/#uncertified^a,b	Median household income (2015)^d	Median household net wealth (2013)^d	Per pupil expenditure^c	M salary (with COLA)^d	SET salary differential
USA	1.41	$53,889	$96,679	$11,508	$55,443	1.00
Low shortage
CT	11.71	$70,331	$147,278	$18,401	$58,028	1.18
IL	5.63	$57,574	$102,768	$13,213	$61,037	1.09
KY	6.22	$43,740	$87,998	$9,411	$61,216	1.01
MI	45.8	$49,576	$87,983	$10,649	$73,592	1.11
NH	196^f	$66,779	$148,468	$14,601	$51,043	0.95
ND	145^f	$57,181	$103,615	$12,032	$52,770	0.93
PA	7.10	$53,599	$113,131	$13,824	$64,668	1.12
Median	11.71***	$57,181	$104,950	$13,213	$61,037*	1.09*
High shortage
AK	1.21	$72,515	$ 120,365	$18,466	$55,100	1.14
HI	0.56	$69, 515	$153,570	$12,400	$42,859	0.88
KS	0.76	$52,205	$96,608	$10,011	$54,482	1.03
MD	0.70	$74,551	$136,853	$14,217	$57,861	0.99
NV	0.52	$51,847	$63,224	$8,275	$57,000	0.91
UT	0.60	$60,727	$104,950	$6,546	$49,614	0.83
WV	0.27	$41,751	$92,262	$11,371	$51,256	0.89
Median	0.60	$60,727	$103,615	$11,371	$54,482	0.91

Note. COLA = cost-of-living adjustment; SET = special education teacher; CT = Connecticut; IL = Illinois; KY = Kentucky; MI = Michigan; NH = New Hampshire; ND = North Dakota; PA = Pennsylvania; AK = Alaska; HI = Hawaii; KS = Kansas; MD = Maryland; NV = Nevada; UT = Utah; WV = West Virginia.

Integrated Postsecondary Data System, 2009–2015. ^bOffice of Special Education Programs. ^cNational Center for Education Statistics, 2016. ^dThe Council for Community and Economics Research. ^eOSEP reports all teachers were highly qualified for the given year, the number reflects the number of available special education graduates for the given year.

p < .05, one-tailed. **p < .01, one-tailed. ***p < .001, one-tailed.

Table 3.

Variables Related to the Demand of Teachers in Low and High Shortage States.

States	Average growth of children <18 (2006–2015)^a	SWDs:SETs ratio^b (2014)
USA	−0.10	20.6
Low shortage
CT	1.10%	13.5
IL	0.89%	12.9
KY	0.11%	15.6
MI	1.37%	17.5
NH	1.66%	13.3
ND	−1.68%	14.8
PA	0.65%	15.1
Median	0.10%*	14.8
High shortage
AK	−0.10%	17.4
HI	0.16%	8.7
KS	−0.34%	16.5
MD	−0.22%	12.2
NV	0.24%	19.7
UT	−1.59%	26.4
WV	0.36%	16.0
Median	−0.89%*	16.5

Note. SWDs = students with disabilities; SET = special education teacher; CT = Connecticut; IL = Illinois; KY = Kentucky; MI = Michigan; NH = New Hampshire; ND = North Dakota; PA = Pennsylvania; AK = Alaska; HI = Hawaii; KS = Kansas; MD = Maryland; NV = Nevada; UT = Utah; WV = West Virginia.

American Community Survey 5-year estimates, 2013–2017. ^bNational Center for Education Statistics, 2014.

p < .05, one-tailed. **p < .01, one-tailed. ***p < .001, one-tailed.

Table 4.

Variables Related to the Demand of Teachers in Low and High Shortage States: SASS Data.

	Would you still become a teacher?						Do you plan to remain in teaching?
	SETs			GETs			SETs		GETs
States	Definitely yes	Maybe	Definitely not	Definitely yes	Maybe	Definitely not	Long-term	Short-term	Long-term	Short-term
USA	70.5	14.9	14.6	65.4	17.4	17.1	78.2	21.8	74.8	25.2
Low shortage
CT	77.8	13.0	9.3	67.9	17.4	14.7	81.8	18.2	79.3	20.7
IL	74.7	13.9	11.4	79.9	10.0	10.2	77.1	22.9	83.1	16.9
KY	78.7	13.1	8.2	67.5	14.2	18.3	85.2	14.8	81.8	18.2
MI	66.4	21.1	12.5	66.4	15.1	18.5	82.9	17.1	76.2	23.8
NH	58.8	23.5	17.6	75.0	11.0	14.0	64.7	35.3	79.8	20.2
ND	88.9	11.1	0	75.4	16.4	8.2	88.9	11.1	75.4	24.6
PA	76.7	11.4	11.9	68.9	19.1	12.0	78.2	21.8	80.2	19.8
M	74.6	15.3	10.1	71.6	14.7	13.7	79.8	20.2	79.4	20.6
High shortage
AK	70.0	10.0	20.0	75.5	15.1	9.4	70.0	30.0	71.7	28.3
HI	33.3	55.6	11.1	75.3	10.4	14.3	41.7	58.3	63.6	36.4
KS	70.3	21.6	8.1	72.5	12.9	14.6	77.8	22.2	77.4	22.6
MD	48.2	16.5	35.3	56.8	21.8	21.5	62.4	37.6	67.5	32.5
NV	62.1	6.9	31.0	54.1	22.7	23.2	69.0	31.0	74.7	25.3
UT	85.2	7.4	7.4	72.8	15.0	12.1	81.5	18.5	74.3	25.7
WV	85.3	5.9	8.8	60.4	20.8	18.8	88.2	11.8	74.0	26
M	64.9	17.7	17.4	66.8	17.0	16.3	70.1	29.9	71.9	28.1

Source. Schools and Staffing Survey, 2011–2012. Demographic variables.

Note. SASS = Schools and Staffing Survey; SET = special education teacher; GET = general education teacher; CT = Connecticut; IL = Illinois; KY = Kentucky; MI = Michigan; NH = New Hampshire; ND = North Dakota; PA = Pennsylvania; AK = Alaska; HI = Hawaii; KS = Kansas; MD = Maryland; NV = Nevada; UT = Utah; WV = West Virginia.

Demographic variables

To compare the demographic make-up of each state, we selected variables that were indicators of differences in population, geography, and wealth (Table 1). Our intent was to frame each state context and the larger samples of high and low shortage states. Our rationale was straightforward. Because the principles of supply and demand in a teacher labor market do not operate in a vacuum, state contexts matter. For example, a wealthier state is more likely to have funds to invest in public education initiatives (Baker, 2017), just as a state with a substantial remote rural population is likely to struggle with attracting individuals into vacant teaching positions (Showalter et al., 2017). The demographic variables we reported include (a) total state population, (b) student population diversity, (c) population density, and (d) per capita GDP. Our intent was to compare included states across these key contextual variables to examine the degree to which our samples differed from one another. In collecting these data, we assumed states with lower shortages of SETs would have a significantly larger population to contribute to the supply pool of potential SETs, greater fiscal resources to invest in public education initiatives, and greater population density, all things being equal. Demographic data are displayed in Table 1.

Supply variables

To capture differences pertaining to supply, we again relied on analysis of teacher labor markets (Lovenheim & Turner, 2018), in which teacher compensation, resource allocation and availability of resources as elements of working conditions, and production of SETs all influence supply. We used two variables to estimate compensation: special education teacher salaries adjusted for cost of living (COLA) and a derived variable we call SET salary differential.

Compensation

We included both COLA and SET salary differential variables in our analysis for two reasons. First, although teacher salary with COLA and SET Salary differential are similar measures, salary adjusted for COLA provides easily understood values. In addition, salary adjusted for COLA provides a starting place to explore the more nuanced measure of SET Salary differential (SET state salary differential calculations, data sources, and results are offered as an online supplement).

We included SET salary differential because it is an estimate of how well SETs are paid relative to what we would predict if special education teacher wages were adjusted for differences across states—for example, in housing prices or amenities such as good weather. (Such factors affect all occupations within a state comparably). For a given occupation, state-to-state wage differences reflect what economists call compensating differentials. Compensating differentials are differences that develop due to the market forces of supply and demand, to compensate for other inherent differences (Rosen, 1986). For example, all else equal, states with better weather are likely to have lower wages, since the better weather boosts the supply of workers and pushes down wages, leaving workers compensated, in part, in sunshine rather than money.

To illustrate, the value of 1.18 for a state such as Connecticut means that actual special education teacher wages there are 1.18 times higher than would be predicted for SETs given the pattern of compensating differentials, statewide changes exhibited by other occupations, and the national level of special education teacher pay.

Of course, SET Salary differential may reflect any number of underlying differences, for example, more restrictive hiring criteria or a more challenging teaching assignment. That is, it may reflect, in part, compensating differentials for differences in the characteristics of SET jobs across states. Although assuming characteristics of the job are roughly the same across states, it reflects the difference between actual pay and pay adjusted for location-related compensating differentials. Thus, the more the actual wage difference between two locations exceeds the compensating differential, the smaller the SET shortage one would expect.

Production

To account for differences across our samples in production of special education degrees, we created a derived variable that represents the ratio of annual special education degree production (Goff et al., 2018; Guarino et al., 2006; Lovenheim & Turner, 2018) from institutions of higher education in each state, to the number of not highly qualified (not HQ) SETs. The intent is to examine a state’s capacity to address SET shortage. In essence, in a given year, this variable represents the number of new college graduates with special education degrees for each not HQ SET.

Availability of resources

In addition, we estimated each state’s availability of resources by exploring differences between median household income and median family net wealth of each state. According to the Pew Research Center, Social & Demographic Trends (2012), the combination of both measures represents a more complete picture of the economic status of families in a state. Recent research has also established that availability of fiscal resources is correlated with the ability of districts to offer competitive wages (Baker & Weber, 2016).

We assumed states with lower SET shortages would have more attractive supply variables, such as greater degree production, higher salaries, and a higher SET salary differential. All supply variables on which we obtained significant or otherwise pertinent differences are found in Table 2.

Resource allocation

To assess differences in resource allocation across samples of high and low shortage states, we examined per pupil expenditures (PPE). PPE represent investment in critical school inputs that benefit students and teachers. Increases in PPE have been identified as effective for increasing both postschool outcomes for students as well as key school inputs related to working conditions, such as lower student to teacher ratios, increased salaries, and extended school years (Greenwald et al., 1996; Jackson et al., 2016). Thus, we investigate differences between our samples in the investment of school inputs that impact teacher working conditions.

Demand variables

To estimate differences within the construct of demand, we relied on labor market–identified variables that shift demand for teachers, chief among them are student population and student:teacher ratios (Baker, 2017; Goff et al., 2018; Lovenheim & Turner, 2018; Pew Research Center, Social & Demographic Trends, 2012; Sutcher et al., 2016). In addition, we included factors that drive demand for SET, including variables measuring commitment and mobility. Demand variables described in this section (on which we obtained significant or otherwise interesting differences) are found in Table 2.

Student population

We evaluated differences in the student population by examining the average growth rate of children under the age of 18 for each state. Our rationale was as states increase in population of school-age children under the age of 18, so too will the demand for additional SETs (Lovenheim & Turner, 2018). We reported this variable as the average rate of growth from 2006 to 2015 in each state.

Student to teacher ratios

Finally, we examined the ratio of SWDs to SETs and assumed that higher ratios suggest increased teacher caseloads and thus demand for SETs (Russ et al., 2001).

Mobility and commitment

Retaining SETs in school limits the need for new teachers. Conversely, teachers who choose to leave the classroom drive demand. Measures of teacher mobility and commitment were derived from the 2012 administration of the Schools and Staffing Survey (SASS:12). Administered every 2 to 4 years from 1987 through 2011, the SASS provides the most complete picture of U.S. teachers, schools, and districts available to researchers. Appropriate sampling weights, recentered to adjust for the reduced sample size (Thomas & Heck, 2001), were applied at all stages of this analysis.

In this investigation, we limited the sample to only regular, full-time teachers who identified their main teaching assignment as either in special education (any; n = 4,142) or in general education (including elementary education, English/language arts, foreign languages, mathematics and computer science, natural sciences, and social sciences; n = 23,055), removing all other sampled teachers. This provided an opportunity to differentiate SET-specific effects from variables that affected all teachers similarly.

Responses to two questions provided an indication of the overall commitment of SETs and GETs to the profession. First, a question regarding whether respondents would enter teaching if given an opportunity to return to college (T0472) was recoded from five response categories to three (definitely yes, maybe, and definitely no). Second, a question regarding how long a respondent planned to remain in teaching (T0473) was recoded from eight response options to two (long-term and short-term commitment). Separately, a question regarding the number of different schools at which respondents had taught provided an indicator of mobility.

Data Analysis

The purpose of this analysis was to identify differences between states with high and low SET shortages. Because our analysis was exploratory, we relied on descriptive and inferential statistics to guide it. Approached correctly, descriptive analysis can establish the characteristics of a place or population and diagnose real-world problems that need to be addressed by policy.

We evaluated the differences in demographics, supply, and demand variables between states in high and low shortage groups by calculating mean ranks, medians, and standard deviations. Because of our small sample size, risk of outliers inflating the mean for each group, and increased likelihood of Type 1 error, we conducted Mann–Whitney U tests to compare our samples (Argesti & Finlay, 2009; Nachar, 2008). The Mann–Whitney U test assumes all dependent variables are ordinal or continuous and are not normally distributed. The Mann–Whitney U tests examine groups based upon the distributions of the two samples to derive one of two possible paths for analysis. In the first case, samples had similar distributions (e.g., estimated by visual analysis from histograms) and we make the comparison from the mean ranks of the two samples. In the second case, distributions are markedly different across samples, and we estimated differences using the median of each group (Nachar, 2008). We used one-tailed tests for all variables for which direction was specifically predicted based on our conceptual framework and literature review and two-tailed tests for variables for which we could not predict direction a priori. In this way, we adopted “low-inference, low-assumption” methods (e.g., central tendency, variation, frequency counts) to cover patterns within the data (Loeb et al., 2017, p. 22). We set the alpha at .05.

Results

In this section, we present the data we gathered organized by demographics, supply, and demand. We report data for individual states, group medians, and national averages. To achieve efficiency in presenting the results of our analyses, we report all statistically significant differences between the groups of high and low shortage states and only those nonsignificant findings of some special import or interest, which we describe on case-by-case basis.

State-Level Demographics

In Table 1, we present state-level demographic variables. States are represented by postal codes and organized by low and high shortage status in this, and all other tables. The demographic variables serve to establish differences that exist along estimates of population, geography, and economic health of our samples. In other words, we sought to ascertain if, in fact, our states were substantially different from one another.

A Mann–Whitney U test indicated no statistically significant differences between the medians of high shortage states (Mdn_H = 2,761,548) and low shortage (Mdn_L = 4,383,272) states on measures of population, U = 15.0, p = .26. Similarly, we found no statistically significant differences in percentage of White students in high (Mdn_H = 52%) and low (Mdn_L = 71%) states, U = 11.5, p = .10. In addition, we found no statistically significant differences between the medians of our samples (Mdn_H = 14, Mdn_L = 14) with regard to population density, U = 14.0, p = .21. Finally, we found no differences between high shortage (Mdn_H = $60,056) and low shortage states (Mdn_L = $53,357) with regard to per capita GDP, U = 17.0, p = .37. In sum, across key state-level demographic variables, we found no statistically significant differences between our samples.

State-Level Supply Data

In assessing supply data, we posited that low shortage states would have access to a more robust pipeline of candidates to address SET demand and thus employ policies and practices that make the occupation of special education a more attractive profession, all things being equal. Thus, we again used one-tailed Mann–Whitney U tests for our analysis. In evaluating production, we calculated the number of graduates from state teacher preparation programs compared with the number of not highly qualified teachers, effectively assessing the degree to which a state is able to address the shortage through graduates of their own preparation systems. The median of low shortage states (Mdn_L = 11.71) was substantially and significantly greater than the ratio in high shortage states (Mdn_H = 0.60), U = 49, p = .001, with low shortage states outproducing high shortage states.

Resources

To capture elements of working conditions, we estimated differences in the availability and allocation or resources as an estimate of school inputs that benefit students and teachers (e.g., reduced class sizes). On measures of wealth, we found no statistically significant differences on median household income for high (Mdn_H = $60,727) and low (Mdn_L = $57,181) shortage states, U = 22, p = .39. We also found no differences in the medians (Mdn_H = $104,950, Mdn_L = $103,615) of the two samples on household net worth, U = 24, p = .50. Mann–Whitney U test indicated no statistical significance between high (Mdn_H = $11,371) and low (Mdn_L = $13,213) shortages states for PPE, U =18.0, p = .228. Although this difference was not statistically significant, it may well be meaningful; a difference of $1,842 converts to $36,840 per class of 20 students.

Compensation

To investigate the role of compensation in explaining shortages, we provide several pieces of information. We evaluated differences in salaries with COLA and the more sensitive SET salary differentials, as reported in Table 2. Due to similar distributions, we used mean rank in the Mann–Whitney test to assess differences between our samples. On average, the mean rank of SETs’s adjusted salary in high shortage states (M_L rank = 9.71) significantly exceeded those in low shortage states (M_H rank = 5.57), U = 9.00, p = .03. In addition, the Mann–Whitney U test indicates a statistically significant difference between the mean rank of high (M_H rank = 5.57) and low (M_L rank = 9.43) shortage states for SET salary differential, U =11.0, p =.049, such that SET salary differentials in low shortage states significantly exceeded those in high shortage states. SET salary differentials in low shortage states (Mdn_L = 1.09) indicated that SETs are paid more than would be predicted on the basis of COLA and compensating differentials; in high shortage states, they pay SETs less than predicted (Mdn_H = 0.91).

State-Level Demand Drivers

We present state-level demand data in Tables 3 and 4. Our analysis of these data was guided by the assumption that high shortage states would have greater values on variables of demand than low shortage states, and, as a result, we used one-tailed Mann–Whitney U tests, which guided our analysis. In examining the average growth of the population of students under the ages of 18, we found statistically significant differences between the medians of high (Mdn_H = −0.10) and low (Mdn_L = .89) shortage states, U = 10.0, p = .036, such that low shortage states have a higher growth rate for children under 18. In measuring differences between medians of high (Mdn_H = 16.5) and low (Mdn_L =14.8) shortage states on student to special education teacher ratio, a rough proxy for caseload, we again found no statistical differences, U =17, p = .19, although a difference of nearly two students per case load may be of some import depending on the nature of disability, resources available to support the student(s), and training of the special educator.

Employing data from SASS:12, we considered the extent to which commitment and mobility drive demand for SETs in our samples. We investigated the extent to which respondents would (a) still choose to enter teaching if given an opportunity to return to college and (b) were committed to remain in teaching. We also compared their responses with GETs. Nationally, 70.5% of SETs and 65.4% of GETs would still enter teaching, if given an opportunity to go back to college. On average, 74.6% of SETs in low shortage states versus 64.9% of SETs in high shortage states would return. The same pattern held for GETs. Thus, GETs in low shortage states (M = 71.6%) were substantially more likely than GETs in high shortage states (M = 66.8%) to indicate a desire to reenter teaching. In terms of plans to remain in teaching, SETs in low shortage states (M = 79.8%) were substantially more likely than SETs in high shortage states (M = 70.1%) to indicate plans to remain in teaching for the long term. For GETs, the differences were nearly identical (M_low = 79.4% and M_high = 71.9%). Overall, low shortage states benefit from teachers, both GETs and SETs, who are more likely to reenter teaching as a profession and are committed to it long term. In both high and low shortage states, however, SETs would reenter teaching at higher rates than their general education counterparts. It is interesting to note that, contrary to findings on higher attrition rates for SETs than GETs, SETs in both high and low shortage states appeared to express more commitment to the profession than GETs. Nonetheless, the benefits of a committed workforce favor low shortage over high shortage states.

Discussion

To address SET shortages, states must have access to accurate information on the wide array of factors that directly and indirectly impact teacher recruitment and retention (Dee & Goldhaber, 2017). Analysis of state-level data may suggest some policy solutions for SET shortages (Cowan et al., 2016; Loeb et al., 2017). In this study, we drew primarily upon publicly available, state-level data to discern differences between high and low SET shortage states in supply policies, labor market conditions, demographics, and infrastructure. Given high state-to-state variability on most measures, we reasoned that differences would be most readily discernible in extreme groups like those we identified. Furthermore, if differences were observed, we anticipated that they would conform to patterns established in the teacher labor market literature (e.g., better pay, greater public investment in education in low shortage states). To guide our analysis, we adopted Guarino et al.’s (2006) supply and demand framework.

Although our aim was to describe differences between low and high shortage states, we recognize that group differences may not apply to all states within a group. Indeed, on most variables, we found overlap between our samples, with one or two states belying trends. Clearly, no two states were precisely the same, and each presented a unique constellation of demographic, demand, and supply variables impacting SET shortage. Such incongruence is to be expected. That said, our results support two general findings of note. First, we found few differences between high and low shortage states in state-level demographics and demand variables. Second, with regard to supply variables, the differences we observed are consistent with both what is known about teacher labor markets (e.g., higher wages) and with our assumption that differences in educational expenditures, including salaries, and preparation capacity will impact supply.

On demographic variables, we found no statistically significant differences between high and low shortage states. High shortage states, however, were more rural and more diverse than low shortage states (even if those differences did not reach statistical significance). The staffing problems in rural areas and in schools serving highly diverse student populations are well known (Showalter et al., 2017; Sindelar, Pua et al., 2018) and most certainly contribute to shortages in our states. Among the demand variables, only average annual growth of children (18 or younger) differentiated the two groups. On average, this subpopulation is growing (by 0.10% per year); by contrast, in high shortage states, the number of children is declining (by nearly 1% per year). This finding is important because declining child counts may temper political appetite for investment in public education.

Although our analysis yielded no significant differences between high and low shortage states on demographic variables, there were substantial differences likely to impact the kinds of policy solutions states might adopt. For example, differences in where residents live may lead to different policy solutions. In WV, a high shortage state, 51% of residents live in rural areas, compared with only 6% in NV. These demographic realities necessitate differentiated policy solutions (Engel & Cannata, 2015). To this point, recent efforts in many states and cities have been launched to create affordable housing and mortgage options for teachers as one localized approach to attract and to retain teachers. For instance, in North Carolina, the state credit union, local nonprofits, and school districts worked together to develop attractive housing options for attracting and retaining individuals willing to work in rural areas (Verdin & Smith, 2013). High shortage states may consider adopting such policies to meet local demands.

In other states, differences in the SET supply pool may require different policy solutions. For instance, although MD reported nearly twice the proportion of adults with bachelor’s degrees than WV, MD schools compete for college-educated labor with such other employers as the federal government and private corporations. To compete for capable individuals to enter teaching, this reality demands competitive wages and easy-entry preparation alternatives. In comparison, WV may want to work closely with school districts to develop “grow your own” programs that target high school students interested in living and working in rural areas. Thus, the problem of SET shortage and the constellation of factors that contributes to it are likely to be distinct. Solutions must be differentiated as well. Mason-Williams (2015) and Dee and Goldhaber (2017) both linked shortages specifically to geographic location and socioeconomic well-being of districts and communities. These findings suggest that addressing SET shortages requires both state and local policy approaches (Dee & Goldhaber, 2017).

Along with socioeconomic status and geography, race of students seems to differentiate high and low shortage states. On average, high shortage states have a substantially greater percentage of students of color. Although this 20% discrepancy was not statistically significant, it suggests an important and unfortunate disparity in the allocation of highly qualified SETs, disadvantaging students of color. This finding is consistent with those of previous researchers who have found that the distribution of teachers with special education credentials is driven by geography, school poverty, student achievement, and racial composition (Cooc & Yang, 2016; Dee & Goldhaber, 2017; Mason-Williams, 2015).

Our analysis uncovered clear and distinct differences favoring low shortage states on educational expenditures, including salaries, and preparation capacity. PPE in low shortage states exceeded PPE in high shortage states by more than $1,800. Although this difference was not statistically significant, it represents a substantial and practically significant investment—$36,000 for a classroom of 20 students. Unfortunately, state and local invest-ments in U.S. public schools declined precipitously during the Great Recession (Leachman et al., 2017) and only recently have begun to recover. Only 21 states spent more in 2015 on education than they did in 2008, and 17 states lagged by 10% or more from 2008 to 2015 (Leachman et al., 2017). In these states, ameliorating the shortfall in PPE may enhance their ability to improve working conditions and retain teachers.

Teacher salaries also differed in high versus low shortage states, favoring the latter. Furthermore, on average, in 2017, U.S. teachers earned $340 less per week than other college graduates, a “teacher pay penalty” (Allegretto & Mishel, 2018, p. 1). Of course, increasing teachers’ salaries is a hard sell in a political environment in which spending on public education is anathema to many. Recent teacher work stoppages, however, have demonstrated that public opinion and political will might well be changing (Cheng et al., 2019)—and that salary and public investments may be malleable policy variables. Besides, we have 30 years of data that affirm both the persistence of SET shortages and the futility of some approaches—lowering admissions standards, redefining fully certified, and providing alternatives to lure career changers into the field. We believe that any serious effort to address SET shortage must have salary increases at its core. Furthermore, although we believe all teachers should be making more money, across-the-board salary increases are not likely to address field-specific shortages. We believe special education is best served politically by alignment with other teacher shortage disciplines.

Our findings align with what is known of teacher labor markets: Lower teacher pay relative to other occupations with comparable education requirements results in people viewing teaching less favorably (Han et al., 2018), and when teaching is viewed less favorably, enrollments in teacher preparation programs are likely to suffer. In our sample of low shortage states, SET salary differential averaged 1.09. Thus, special educators in low shortage states were paid more than predicted (adjusted for COLA and compensating differential), and the supply of special education graduates outpaced numbers of available positions. In high shortage states, by contrast, the SET salary differential averaged .91. In these states, SETs are paid less than predicted, and EPPs prepare one SET for every two available positions.

Low shortage states have twice the number of preparation programs and produce more special education graduates than high shortage states, but do high shortage states lack capacity—or enrollment? In instances where programs are at full capacity with more candidates then they are able to serve, a reasonably low-cost solution might be to add faculty or programs. Nevertheless, given the current consensus about low EPP enrollments (Dee & Goldhaber, 2017), it seems reasonable to assume that most EPPs—in both high and low shortage states—are operating below capacity—and that the increasing capacity solution offers little promise in the current context. To increase enrollments in rural areas where access to bricks and mortar programs is limited, however, states may benefit from establishing robust online or hybrid programs (Sindelar, Pua et al., 2018). That said, our findings and other current research (Allegretto & Mishel, 2018; Han et al., 2018) suggest that a lack of available candidates in high shortage states is tied to poor compensation relative to other employment options and a resulting lack of social status for special education teaching. We believe that improving the valence of special education teaching as a profession may be a useful antidote to low enrollments. Increasing the valence of special education teaching (or any shortage discipline, for that matter) cannot be achieved by across-the-board salary increases. Dee and Goldhaber (2017) have argued that although across-the-board compensation increases are attractive, the overall cost of such policies is prohibitive and is unlikely to address field-specific shortages. Simply put, across-the-board increases do little to increase the attractiveness of a career in special education relative to all other teaching fields.

Limitations and Implications

Our study was limited in several important ways. First, our findings are descriptive and consequently do not allow for causal inference. Thus, the statistically significant difference we obtained in average salaries (adjusted for COLA), for example, offers no guarantee that increasing salary will reduce shortage. Second, given the small number of states in each group and the relatively high within-group variability on many variables, our analyses had low statistical power. As a result, even some substantial mean differences were not statistically significant (e.g., PPE), on the contrary, low power reduced the threat of our making a Type 2 error. Relatedly, when aggregating across states, within-state nuance and subtlety are lost. As a result, some factors related to SET shortage (e.g., school poverty)—and the policy solutions associated with them—are obscured in state-level aggregations. Next, in describing shortages, we adopted the quality shortage definition by Cook and Boe (2007), which underestimates the magnitude of shortage. Underestimation may occur for both high and low shortage states, although the degree of underestimation will vary with the proportion of SETs enrolled in alternative route programs. For one thing, many SETs are considered to be highly qualified merely by virtue of being enrolled in alternative route programs, and OSEP counts fail to identify the number of highly qualified SETs who are not fully certified and presumably underprepared. For another, OSEP long ago stopped reporting vacancies, largely because there were so few. We wonder whether this assumption holds today, given the magnitude of current shortages and opportunity schools have to hire part-time substitutes who never appear in OSEP counts. Finally, our analysis does not account fully for working conditions, such as workload manageability, which recent re-search (Bettini et al., 2017) has related to career intentions. Although we captured teachers’ intent to stay in special education with SASS data—and found no statistically significant differences between SETs in our samples—future studies should account for workplace conditions, such as school climate and administrative support.

Despite the limitations of this study, there are a number of important implications for the work moving forward. Although this analysis revealed significant state-level differences, there is much to be learned from deeper dives into state data. All 14 of our states—indeed, all 50 states—have regional and district-to-district differences that affect SET shortages. More focused analysis requires harnessing the power of large datasets to generate substantive reports on the supply and demand needs of localized areas (see Goff et al., 2018). More nuanced, within-state analysis may provide answers to critical questions, such as the extent of within-state trends in attrition and mobility, trends in the teacher supply, and district responses to shortages and retention efforts.

For these efforts to be effective, researchers must work collaboratively with state departments of education, where possible, to analyze data for policy solutions responsive to local need. The need for accurate data is paramount. As we mentioned previously, there are significant limitations to OSEP personnel data, which led us to look carefully for consistent year-to-year reporting and responsiveness to system shocks before including a state in our analysis. OSEP personnel data are plagued by missing data, anomalous data points, and the suspicion of using work-arounds to portray states more favorably. Accurate data would enhance national and state reports about shortages of SETs and help inform federal and state policy development.

Conclusion

In sum, our study has contributed to the growing knowledge base on the drivers of special education teacher shortages by describing state-level variables that differentiate high and low shortage states. Our analysis suggests that compensation, special education degree production, and investment in public education are among the variables that differentiate these two groups. Some of these differences have policy implications (e.g., salaries, PPE, preparation capacity), and others do not (e.g., compensating differential). Full understanding awaits district-level analysis on variables that are heterogeneous within states.

Supplemental Material

SPED_Pay_Diff – Supplemental material for Special Education Teacher Shortage: Differences Between High and Low Shortage States

Supplemental material, SPED_Pay_Diff for Special Education Teacher Shortage: Differences Between High and Low Shortage States by David J. Peyton, Kelly Acosta, Alexandria Harvey, Daisy J. Pua, Paul T. Sindelar, Loretta Mason-Williams, Jim Dewey, Tiffany L. Fisher and Emily Crews in Teacher Education and Special Education

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

David J. Peyton

Daisy J. Pua

Loretta Mason-Williams

Supplemental Material

Supplemental material for this article is available online.

Notes

Author Biographies

David Peyton is an Assistant Professor of Special Education at Appalachian State University.

Kelly Acosta is a doctoral candidate in Special Education at the Univeristy of Florida.

Alexandria Harvey is a doctoral candidate in Special Education at the Univeristy of Florida.

Daisy Pua is a doctoral candidate in Special Education at the Univeristy of Florida.

Paul Sindelar is a Distinguished Professor at the University of Florida.

Loretta Mason-Williams is an Associate Professor at Binghamton State Univeristy.

James Dewey is an Assistant Professor at Florida Polytechnic Univeristy.

Tiffany Fisher is a doctoral candidate in Special Education at the Univeristy of Florida.

Emily Crews is a doctoral candidate in Special Education at the Univeristy of Florida.

References

Allegretto

Mishel

(2018). The teacher pay penalty has hit a new high: Trends in the teacher wage and compensation gaps through 2017. https://www.epi.org/publication/teacher-pay-gap-2018/

Aragon

(2016). Teacher shortage series: Teacher shortages: What we know. https://www.ecs.org/teacher-shortages/

Argesti

Finlay

(2009). Statistical methods for the social sciences. Pearson.

Baker

B. D.

(2017). How money matters for schools. https://learningpolicyinstitute.org/product/how-money-matters-report

Baker

B. D.

Weber

(2016). State school finance inequities and the limits of pursuing teacher equity through departmental regulation. Education Policy Analysis Archives, 24(47). https://doi.org/10.14507/epaa.v.24.2230

Bettini

E. A.

Jones

Brownell

M. T.

Conroy

Park

Leite

Benedict

(2017). Workload manageability among novice special and general educators: Relationships with emotional exhaustion and career intentions. Remedial and Special Education, 38(4), 246–256. https://doi.org/10.1177/0741932517708327

Boe

E. E.

(2006). Long-term trends in the national demand, supply, and shortage of special education teachers. The Journal of Special Education, 40(3), 138–150. https://doi.org/10.1177/00224669060400030201

Cheng

Henderson

Peterson

P. E.

West

M. R.

(2019). Results from the 2018 EdNext poll. Education Next, 19(1), 9–26.

Clotfelter

Glennie

Ladd

Vigdor

(2008). Would higher salaries keep teachers in high-poverty schools? Evidence from a policy intervention in North Carolina. Journal of Public Economics, 92(5–6), 1352–1370. https://doi.org/10.1016/j.jpubeco.2007.07.003

10.

Cooc

Yang

(2016). Diversity and equity in the distribution of teachers with special education credentials: Trends from California. AERA Open, 2(4), 1–15. https://doi.org/10.1177/2332858416679374

11.

Cook

L. H.

Boe

E. E.

(2007). National trends in the sources of supply of teachers in special and general education. Teacher Education and Special Education, 30(4), 217–232. https://doi.org/10.1177/088840640703000402

12.

Cowan

Goldhaber

Hayes

Theobald

(2016). Missing elements in the discussion of teacher shortages. Educational Researcher, 45(8), 460–462. https://doi.org/10.3102/0013189X16679145

13.

Dee

T. S.

Goldhaber

(2017). Understanding and addressing teacher shortages in the United States. https://www.brookings.edu/wp-content/uploads/2017/04/es_20170426_understanding_and_addressing_teacher_shortages_in_us_pp_dee_goldhaber.pdf

14.

Dewey

Sindelar

P. T.

Bettini

Boe

E. E.

Rosenberg

M. S.

Leko

(2017). Explaining the decline in special education teacher employment from 2005 to 2010. Exceptional Children, 83(3), 315–329. https://doi.org/10.1177/0014402916684620

15.

Engel

Cannata

(2015). Localism and teacher labor markets: How geography and decision making contribute to inequality. Peabody Journal of Education, 90, 84–92. https://doi.org/10.1080/0161956X.2015.988533

16.

Feng

Sass

T. R.

(2013). What makes special-education teachers special? Teacher training and achievement of students with disabilities. Economics of Education Review, 36, 122–134. https://doi.org/10.1016/j.econedurev.2013.06.006

17.

Feng

Sass

T. R.

(2018a). The impact of incentives to recruit and retain teachers in “hard-to-staff” subjects. Journal of Policy Analysis and Management, 37(1), 112–135. https://doi.org/10.1002/pam.22037

18.

Feng

Sass

T. R.

(2018b). Teacher quality and teacher mobility. Education Finance and Policy, 12(3), 396–418. https://doi.org/10.1162EDFP_a_00214

19.

Goff

Carl

Yang

(2018). Supply and demand for public school teachers in Wisconsin (WCER Working Paper No 2018-2). https://wcer.wisc.edu/docs/working-papers/Working_Paper_No_2018_02.pdf

20.

Goldring

Taie

Riddles

(2014). Teacher attrition and mobility: Results from the 2012-13 teacher follow-up survey (NCES 2014-077). https://nces.ed.gov/pubs2014/2014077.pdf

21.

Greenwald

Hedges

L. V.

Laine

R. D.

(1996). The effect of school resources on student achievement. Review of Educational Research, 66(3), 161–396. https://doi.org/10.2307/1170528

22.

Guarino

C. M.

Santibanez

Daley

G. A.

(2006). Teacher recruitment and retention: A review of the recent empirical literature. Review of Educational Research, 76(2), 173–208. https://doi.org/10.3102/00346543076002173

23.

Han

S. W.

Borgonovi

Guerriero

(2018). What motivates high school students to want to be teachers? The role of salary, working conditions, and societal evaluations about occupations in a comparative perspective. American Educational Research Journal, 55(1), 3–39. https://doi.org/10.4135/9781446215074

24.

Individuals with Disabilities Education Act of 2004, P.L. 108-446.

25.

Jackson

C. K.

Johnson

R. C.

Persico

(2016). The effects of school spending on educational and economic outcomes: Evidence from school finance reforms. Quarterly Journal of Economics, 131(1), 157–218. https://doi.org/10.1093/qje/qjv036

26.

Katsiyannis

Zhang

Conroy

(2002). Availability of special education teachers: Trends and issues. Remedial and Special Education, 24(4), 246–253. https://doi.org/10.1177/07419325030240040701

27.

Kolbe

Strunk

K. O.

(2012). Economic incentives as a strategy for responding to teacher staffing problems: A typology of policies and practices. Educational Administration Quarterly, 48(5), 779–813. https://doi.org/10.1177/0013161X12441011

28.

Leachman

Masterson

Figueroa

(2017). A punishing decade for school funding. https://www.cbpp.org/sites/default/files/atoms/files/11-29-17sfp.pdf

29.

Levin

Berg-Jacobson

Atchison

Lee

Vontsolos

(2015). Massachusetts study of teacher supply and demand. https://www.air.org/sites/default/files/downloads/report/Massachusetts-Study-of-Teacher-Supply-and-Demand-December-2015_rev.pdf

30.

Loeb

Dynarski

McFarland

Morris

Reardon

Reber

(2017). Descriptive analysis in education: A guide for researchers (NCEE 2017–4023). U.S. Department of Education, Institute of Education Sciences, National Center for Education Evaluation and Regional Assistance.

31.

Lovenheim

Turner

(2018). Economics of education. Worth Publishers.

32.

Mason-Williams

(2015). Unequal opportunities: A profile of the distribution of special education teachers. Exceptional Chil-dren, 81(2), 247–262. https://doi.org/10.1177/0014402914551737

33.

Mason-Williams

Bettini

E. A.

Gagnon

J. C.

(2017). Access to qualified special educators across elementary neighborhood and exclusionary schools. Remedial and Special Education, 38(5), 297–307. https://doi.org/10.1177/0741932517713311

34.

Nachar

(2008). The Mann-Whitney U: A test for assessing whether tow independent samples come from the same distribution. Tutorials in Quantitative Methods for Psychology, 4(1), 13–20. https://doi.org/10.20982/tqmp.04.1.p013

35.

Nougaret

A. A.

Scruggs

T. E.

Mastropieri

M. A.

(2005). Does teacher education produce better special education teachers? Exceptional Children, 71(3), 217–229. https://doi.org/10.1177/00144020507100301

36.

Pew Research Center, Social & Demographic Trends. (2012). Fewer, poorer, gloomier: The lost decade of the middle class. https://www.pewresearch.org/wp-content/uploads/sites/3/2012/08/pew-social-trends-lost-decade-of-the-middle-class.pdf

37.

Rosen

(1986). Chapter 12: The theory of equalizing differences. In Handbook of labor economics (Vol. 1, pp. 641–692). https://doi.org/10.1016/S1573-4463(86)01015-5

38.

Russ

Berttram

Rylance

B. J.

Bongers

(2001). Caseload in special education: An integration of research findings. Exceptional Children, 67(2), 161–172. https://doi.org/10.1177/001440290106700202

39.

Showalter

Klein

Johnson

Hartman

S. L.

(2017). Why rural matters 2015-2016: Understanding the changing landscape. https://www.ruraledu.org/user_uploads/file/WRM-2015-16.pdf

40.

Sindelar

P. T.

Bettini

Fisher

Mason-Williams

Rosenberg

(2018, November). Back to the future: Special education, teacher supply, and dynamic labor markets [Paper presentation]. Presentation at the annual meeting of the Teacher Education Division of the Council for Exceptional Children, Las Vegas, NV, United States.

41.

Sindelar

P. T.

Daunic

Rennells

M. S.

(2004). Comparisons of traditionally and alternatively trained teachers. Exceptionality, 12(4), 209–223. https://doi.org/10.1207/s/15327035ex1204_3

42.

Sindelar

P. T.

Fisher

Myers

(2018). The landscape of special education licensure (2016). Teacher Education and Special Education, 42(2), 101–116. https://doi.org/10.1177/0888406418761533

43.

Sindelar

P. T.

Pua

D. J.

Fisher

Peyton

D. J.

Brownell

M. T.

Mason-Williams

(2018). The demand for special education teachers in rural schools revisited: An update on progress. Rural Special Education Quarterly, 37(1), 12–20. https://doi.org/10.1177/8756870517749247

44.

Smith

T. M.

Ingersoll

R. M.

(2004). What are the effects of induction and mentoring on beginning teacher turnover? American Educational Research Journal, 41(3), 681–714. https://doi.org/10.3102/00028312041003681

45.

Sutcher

Darling-Hammond

Carver-Thomas

(2016). A coming crisis in teaching? Teacher supply, demand, and shortages in the U.S. https://learningpolicyinstitute.org/product/coming-crisis-teaching

46.

Thomas

S. L.

Heck

R. H.

(2001). Analysis of large-scale secondary data in higher education research: Potential perils associated with complex designs. Research in Higher Education, 42, 517–540. https://doi.org/10.1023/A:1011098109834

47.

Verdin

Smith

(2013). The impact of housing on teacher recruitment and retention in rural areas: A case study. https://pdfs.semanticscholar.org/8837/865b1d4337fd5168c39f20c535608dacbff1.pdf?_ga=2.208070939.844214293.1580531636-607929785.1576734677

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.04 MB