Skip Navigation

Jay G. Chambers
American Institutes for Research

About the Author

During the late 1970s and early 1980s, Dr. Jay G. Chambers served as the Associate Director of the Institute for Research on Educational Finance and Governance (IFG) at Stanford University, a national R&D center funded by the National Institute of Education. He is a nationally recognized expert in the analysis of patterns of resource allocation and cost variations in educational organizations and in recent years has played a major role in cost analyses of special education programs. He also was the principal investigator for A Study of Chapter 1 Resources in the Context of State and Local Resources for Education, conducted for the U.S. Department of Education, Office of Policy and Planning.

Dr. Chambers has directed two major school finance reform studies in Illinois and Alaska; he has directed a large-scale cost analysis of early intervention services in California; and, he has completed a cost analysis of migrant summer school programs.

Dr. Chambers served as an expert witness in the Levittown school finance reform case in New York State. He has also directed a major comparative study funded by the National Institute of Education on differences in the patterns of resource allocation and decision-making in public and private schools.

Currently, Dr. Chambers is Co-Director of the Center for Special Education Finance (CSEF) at the American Institutes for Research (AIR).

Public School Teacher Cost
Differences Across the United States:
Introduction to a Teacher Cost Index (TCI)

Jay G. Chambers
American Institutes for Research


Since the mid-1970s, a number of studies have focused on the development of methodologies and empirical estimation of a cost-of-education index (CEI).1 A CEI is designed to adjust for differences in the purchasing power of the educational dollar in different locations. Because personnel expenditures account for about 80 percent of local school budgets, most of the previous studies of education cost differences have focused attention on the analysis of personnel costs.2 Thus, the development of an adequate methodology for addressing the differences in the costs of personnel would aid in the development of a CEI. But how are teacher cost differences defined?

Most educators readily acknowledge that school districts in different geographic locations encounter different costs in acquiring and retaining teachers with similar qualifications. That is, teacher salaries reflect not only the cost-of-living and desirability of a given geographic location, but also a school district's preference for teacher qualifications (e.g., educational preparation or experience levels). The studies of personnel cost differences address the following question:

How much more or less does it cost in different jurisdictions to recruit and employ school personnel with similar characteristics into similar jobs and job assignments?

Accurately measuring geographic cost differences has been one of the pre-eminent measurement challenges in education. The challenge of this task lies in identifying and measuring the factors that define similar characteristics and similar jobs, as well as those factors that reflect differences in the cost-of-living and desirability across geographic locations.

Until recently, no national data have been available to support a comprehensive analysis of the variations in teacher salaries. With the advent of the Schools and Staffing Survey (SASS), which was first conducted during the 1987-88 school year by the (NCES), a data source emerged that supports the empirical analysis required to develop a national, geographical teacher cost index (TCI).

The study described in this paper draws on a sample of over 40,000 public school teachers, derived from the SASS database for school year 1990-91, to conduct the statistical analysis of teacher salaries and cost. In conjunction with the SASS data, the analysis also uses data from the Census Bureau (e.g., population and population density), the Bureau of Labor Statistics (BLS) (e.g., unemployment rates), the Federal Bureau of Investigation (FBI) (e.g., crime rates), and the National Climatic Data Center (NCDC) (e.g., climatic conditions).

The Importance of Cost-of-Education Adjustments

The importance of developing a CEI is that it may be used in two significant ways. First, it may be used to adjust educational expenditure or teacher salary data for differences in the purchasing power of the educational dollar in different communities. For the most part, published information on educational expenditures and salaries of school personnel across states and other local jurisdictions is based on actual reported values.3 However, because of the existing variations in the costs of comparable educational resources across these jurisdictions, it is difficult to make comparisons of the level of educational services being provided in different locations. In order to make such comparisons, it is necessary to adjust reported data on average educational expenditures and teacher salaries for differences in the purchasing power of the educational dollar across jurisdictions.

Second, in addition to their importance for reporting expenditure and salary data, such cost-of-education adjustments play a significant role in analyses of the demand for educational services and inputs across communities.4 Studies of educational resource allocation have commonly had to use measures such as average teacher salaries or other proxy variables (e.g., opportunity wages in other occupations) to reflect relative costs of school resources. Unfortunately, variations in average teacher salaries reflect both variations in costs, as well as the qualifications of the teaching staff. Analysis of demand for school inputs requires an index of the relative cost of comparable inputs. The importance of accurately controlling for costs in these analyses is that ultimately such studies are often focused on addressing the impact of changes in state or federal policies or funding formulas. Without the ability to control for the impact of input costs on choices of local school district officials, it is not possible to isolate the effects of state and federal policies on patterns of resource allocation in local schools.

Toward the Development of a Teacher Cost Index (TCI)

Barro (1992) developed a model that adjusts the variations in average teacher salaries for variations in the levels of education and experience. Other researchers (e.g., McMahon and Chang 1991) have suggested using a cost-of-living adjustment to account for variations in the purchasing power of the education dollar. Unfortunately, neither of these alternatives is adequate. To capture such variations in teacher costs requires a comprehensive analysis of the patterns of teacher compensation. It requires a model that portrays the complexities of the employment transaction between an individual teacher and the school district: that is, one that accounts for school district preferences for teacher qualifications and individual teacher preferences for working and living conditions in local communities.

The model used for the development of the TCI--referred to by economists as the hedonic wage model--provides a comprehensive conceptual framework for understanding and sorting the various factors that underlie variations in the patterns of teacher salaries. This model is well suited to isolate the impact of regional amenities and costs of living on teacher salaries while controlling for various teacher and job characteristics.

In an earlier paper, Chambers (1981a) described the hedonic wage model as follows:

The intuitive notion underlying this theoretical structure is that individuals care both about the quality of their work environment as well as the monetary rewards associated with particular employment alternatives, and that they will seek to attain the greatest possible personal satisfaction by selecting a job with the appropriate combination of monetary and non-monetary rewards. Similarly, employers are not indifferent as to the characteristics of the individual to whom they offer particular jobs. The result of these simultaneous choices is the matching of individual employees with employers. It is the result of this matching process itself that reveals implicitly the differential rates of pay associated with the attributes of individual employees and the working conditions offered by employers. More formally, it is the supply of, and demand for, individuals with certain personal attributes to any particular kind of job assignment that determines the equilibrium wages of labor as well as the implicit market prices attached to the personal and job characteristics.

The implicit relationship observed between wages and the personal and job characteristics of individuals is referred to as a hedonic wage index. The word hedonic literally refers to the physical and psychic pleasures that one can derive from engaging in certain activities. In the context of labor markets, the word hedonic refers to the satisfactions or utility derived by employees from the characteristics of the work place and the profits or the perceived productive value derived by employers from the characteristics of employees they assign to certain jobs. The hedonic wage index permits one to decompose the observed variation in the wages paid to labor into the dollar values attached to each unit of the personal and workplace characteristics (p.51).

Ordinary least squares regression is used to estimate the parameters of the model. The estimated coefficients provide a foundation for determining the wage premiums (positive or negative) associated with particular personal, job, or location characteristics.

This analysis reveals wage premiums for attributes of the workplace and the employee that are not commonly included in regular salary schedules. For this reason, the coefficients are said to provide estimates of the implicit prices of particular attributes. The patterns of implicit prices for worker attributes result from the process of matching the combination of teacher characteristics embodied in a given individual teacher to the job characteristics embodied in a given job assignment. The employment of teachers in particular schools represents a process of choice for the teachers (on the supply side) and for the school district decision-makers (on the demand side) that reveals the trade-offs among the teacher attributes and job characteristics. These trade-offs provide the basis for the set of implicit prices.

Objectives of the Analysis

The TCI analysis presented here accomplishes two primary objectives. First, the TCI component extends the analysis of teacher salaries to include specific variables that reflect the costs of living and the amenities of the jurisdictions in which public school systems are located. Second, the empirical analysis of teacher salaries is used to estimate a TCI for each school district in the United States.

The TCI is designed to reflect variations in teacher salaries associated with factors that are outside the control of local school decisionmakers. Thus, calculation of the TCI requires controlling for (i.e., holding statistically constant) the differences in teacher qualifications and job assignments, while simulating the effects of the factors that reflect differences in costs of living and attractiveness of local jurisdictions. The analysis of teacher costs presented in this paper controls for two major categories of teacher and job attributes:5

Cost Factors: Regional and District Characteristics That Are Outside Local Control

How do teacher salaries vary with factors outside local control? These cost factors encompass variations in the costs of living, competitiveness of the labor markets, levels of crime, quality of the weather, availability of alternative job opportunities, and other attributes of the regions and districts that affect their attractiveness as places to live and work. It is anticipated that less attractive jurisdictions will have to pay relatively higher salaries to attract comparable teachers.

Highlights of the variations in teacher salaries in relation to each of the cost factors are presented below.

The differences in teacher salaries associated with these variables are cost differences. They reflect the variations in salaries paid to comparable teachers working in similar job assignments across local school systems. That is, all else equal, districts in more urbanized settings tend to pay higher relative salaries for comparable teachers. In addition, districts located in faster growing regions, regions with climates characterized by colder temperatures and greater quantities of snowfall, and regions with higher rates of crime pay higher relative salaries to teachers, holding all else constant. At the same time, districts in more remote regions pay somewhat higher-than-average salaries to compensate for reduced access to some of the amenities of living in more urbanized areas.6

Teacher Cost Differences by State

Table 1 presents state-by-state and overall patterns of variation in the TCI. It designates the number of districts for which data are included, along with the weighted mean, standard deviation, and minimum and maximum values of the index within each state. The overall mean value of the TCI for the United States is scaled to a value of 100. This means that the index is scaled so that the average student attends a district with a TCI value of 100. The variation in the TCI across the United States ranges from a low of 53 to a high of 137. This means that teacher costs for a district located in the lowest cost region of the United States are 53 percent of those faced by the district serving the average student. The highest cost district pays 37 percent higher teacher costs than the district serving the average student. The lowest cost region of the United States is located in South Dakota, while the highest cost region is located in Alaska. Another way of looking at these numbers suggests that districts in the highest cost regions of the country pay 2.6 times (= 137/53) as much to place comparable teachers in comparable classrooms and schools as districts in the lowest cost regions of the country. The standard deviation of the TCI is 11.7 percent; that is, most of the districts are within plus or minus 11.7 percent of the average.

Table 1. State-by-state estimates of the teacher cost index (TCI)

                                                     Descriptive statistics on TCI
                        Number of       Total               Standard
State                   districts       enrollment   Mean   deviation      Minimum    Maximum
United States             14,494        40,116,02     100    11.7              53     137
Alaska                        36           94,330     114       8              96     137
Alabama                      129          725,115      88     5.4              76      97
Arkansas                     322          431,490      87     4.1              78      97
Arizona                      205          630,816      97     6.9              84     106
California                   991         4,813,64     109     7.6              77     119
Colorado                     176          573,985      99     7.6              71     116
Connecticut                  160          453,468     114     3.9             103     118
District of Columbia           1           80,694     107       0             107     107
Delaware                      16           96,384     102     4.2              96     106
Florida                       67         1,862,18      95     5.4              79     107
Georgia                      184         1,150,17      92     8.8              71     106
Hawaii                         1          159,285      92       0              92      92
Iowa                         429          483,176      90     4.7              76      98
Idaho                        110          217,555      94     4.7              72     102
Illinois                     942         1,795,47     107    13.1              72     120
Indiana                      295          937,324      98     6.3              80     106
Kansas                       302          436,494      88     7.6              58      99
Kentucky                     176          630,091      89     5.5              76      99
Louisiana                     65          774,724      85     3.9              74      92
Massachusetts                269          730,024     114     3.8              88     120
Maryland                      22          669,620     104     5.7              86     112
Maine                        215          208,599     104     4.4              95     111
Michigan                     552         1,560,80     105     7.5              86     115
Minnesota                    429          751,268      99     8.9              73     111
Missouri                     538          805,029      95       9              71     107
Mississippi                  149          491,684      84     3.8              74      91
Montana                      503          148,411      94     5.2              76     119
North Carolina               133        1,084,489      93     5.1              80     101
North Dakota                 262          117,531      89     5.3              68     111
Nebraska                     728          269,106      90     6.9              58     118
New Hampshire                148          163,778     109     3.9             101     113
New Jersey                   534         1,007,16     113     4.5              96     119
New Mexico                    86          296,471      90     5.2              74      97
Nevada                        17          201,316      95     4.1              87     108
New York                     627         2,361,04     115    12.7              89     129
Ohio                         610         1,766,73     102     6.5              83     113
Oklahoma                     586          568,711      87     4.3              68      95
Oregon                       292          483,507     100     6.5              72     108
Pennsylvania                 499        1,629,157     106       8              86     120
Rhode Island                  36          136,086     111     1.7             108     112
South Carolina                80          451,308      90     5.9              79     102
South Dakota                 172          125,316      87     4.4              53      92
Tennessee                    132          819,229      90     4.8              77      98
Texas                      1,042         3,380,80      93     8.3              70     106
Utah                          40          444,832      97     4.4              73     110
Virginia                     129          984,702      96     8.2              75     108
Vermont                      236           90,215     101     2.7              95     108
Washington                   294          810,011     106     9.1              75     116
Wisconsin                    424          796,114      99     6.7              85     109
West Virginia                 54          318,577      86       3              78      93
Wyoming                       49           97,976      88     4.4              78     107

SOURCE: U.S. Department of Education, , Schools and Staffing Survey, 1990-91.

The five states with the highest average teacher costs are, in order from highest to lowest, New York (115), Massachusetts (114), Connecticut (114), Alaska (114), and New Jersey (113). Four of these states are located in the northeastern portion of the United States. Calculations of standard errors indicate that the differences among the top five states are not statistically significant.7

The five states with the lowest average teacher costs are, in order from lowest to highest, Mississippi (84), Louisiana (85), West Virginia (86), Oklahoma (87), and South Dakota (87). Four of these states are located in the South (using the state classification scheme provided in SASS). Once again, based on the standard errors of these estimates, none of the differences among the lowest five states is statistically significant.

The five states with the largest within-state variation in TCI values (based on the size of the standard deviations presented in the table) are Illinois, New York, Washington, Missouri, and Minnesota. The states with the lowest within-state variation8 are Rhode Island, Vermont, West Virginia, Massachusetts, and Mississippi. In general, states with larger numbers of school districts tend to have a larger variance. Among the five states that exhibited the highest within-state variation, the average number of districts per state is 566, while the average number of districts in the five states with the lowest within-state variation is 149.

Teacher Cost Differences, by Type of District

Table 2 presents the descriptive statistics for the TCI broken down by region of the United States, level of per-pupil revenue in the district, population of the metropolitan area or county of location, distance from the central city, district enrollment, type of city, and percentage of students living in poverty.

Table 2. --The TCI, by region, per-pupil revenue, metropolitan population, distance from the nearest central city, district size, type of city, and percentage of children living in poverty

                                                      Descriptive statistics on TCI
                          Number of  Total                Standard
Category                  districts  enrollment   Mean    deviation   Minimum  Maximum
Northeast                    2,724    6,779,53     112     9.7           86     129
Midwest                      5,683    9,844,37     100    10.4           53     120
South                        3,287   14,519,98      92     7.8           68     112
West                         2,800    8,972,13     104     9.7           71     137
Per-pupil revenue
Less than $4,000             3,695    8,903,340     91     8.1           65     119
4,0006,000                  7,122   22,072,043     99    10.0           53     120
6,0008,000                  2,316    7,257,311    110    10.9           58     129
8,00010,000                   797    1,382,471    111     7.6           62     123
More than 10,000               564      500,862    114     7.4           65     137
Metropolitan population
Less than 5,000                686      166,555     84     9.2           53     119
5,000-20,000                 3,341    2,521,938     85     6.3           66     137
20,000-50,000                2,999    4,170,191     88     5.8           72     123
50,000-100,000               1,519    3,201,822     92     5.9           78     112
100,000-500,000              2,458    8,348,309     95     7.2           73     119
500,000-1,000,00             1,188    5,578,574    100     7.8           85     118
More than 1,000,             2,303   16,128,638    110     8.7           84     129
Distance from the nearest central city
Less than 10 miles           2,018   15,477,412    102     9.9           73     120
10-20                        2,369     8,832,81    107    11.9           68     129
20-40                        3,973     8,186,29     97    10.5           68     120
40-80                        3,770     5,695,55     90     7.9           58     119
80-160                       1,885     1,576,68     90     7.7           53     119
More than 160                  479      347,281    100    11.2           65     137
District size
Less than 500 students       5,154    1,103,979     91    10.3           53     137
5011,000                    2,370    1,712,255     94    10.3           63     134
1,0015,000                  5,374   12,270,304     98    11.3           71     137
5,00110,000                   915    6,317,093     99    11.2           75     123
10,00125,000                  480    7,135,233    100    10.4           72     119
25,00150,000                  120    4,081,084    100     9.5           73     119
50,001100,000                  44    2,960,552    100     7.8           85     120
More than 100,00                21    4,523,514    111    12.2           91     129
Type of city
Large central city             811    8,579,610    108    10.9           83     129
Mid-size city                  806    9,187,913     97     8.5           73     119
Urban fringe of large city   1,287    5,921,311    110     7.7           81     120
Urban fringe of mid-size city  810    2,861,090     99    10.0           71     118
Large town                     418    1,154,387     92     8.5           58     119
Small town                   4,158    8,812,909     93     9.9           70     137
Rural                        6,174    3,596,914     92     8.7           53     137
Percentage of children living in poverty
Less than 10%                4,808   11,733,121    105     8.5           62     134
10-20                        4,834   13,299,197     97     9.0           53     137
20-40                        3,656   12,941,309     99    14.5           58     137
More than 40                   875    1,700,075     93    12.4           62     119

SOURCE: U.S. Department of Education, , Schools and Staffing Survey, 1990-91.

Highlights of the pattern of variations in the TCI are presented below.

A Comparison of Alternative Models: The Case for the TCI

In addition to the TCI derived from the hedonic wage model, two alternative cost adjustments have been proposed: (1) a geographic cost-of-living (COL) index (McMahon and Chang 1991), and (2) an index of variations in average teacher salaries adjusted for differences in teacher education and experience (Barro 1994). As one would expect, the correlations between the TCI, the McMahon-Chang COL, and Barro's adjusted-average-teacher salary index are positive and relatively high (i.e., 0.76 and 0.72, respectively). However, there are significant differences in the values of these indices and what they represent. The COL accounts only for variations in the cost of living which, while an important part of teacher cost differences, does not capture all of the relevant factors (e.g., the effects of labor market competition, climate, crime, and urban amenities). The Barro index controls for teacher education and experience, but fails to control for variations in other teacher and school attributes that are within local control.

The potential for distortion between the TCI and COL is evident from a more detailed examination of the patterns of difference across the United States. For example, the COL values for the highest COL regions of the country--the metropolitan areas surrounding the cities of San Francisco and San Jose, California--exceed the average values of the TCIs for these same regions by more than 28 percentage points. These values represent significant differences in the perception of what constitutes high costs and provide a very different picture of the real purchasing power of the educational dollar. These results suggest that teachers are willing to trade salaries for the amenities of living in the San Francisco-San Jose areas. The state of Hawaii shows a similar pattern.

The TCI presented in this paper represents an attempt to account for most of the factors that affect the ability of local school systems to recruit and employ teachers with similar characteristics hired into similar jobs and job assignments. It accounts systematically for the factors that underlie differences in the cost of living, and for differences in regional amenities that affect the attractiveness of places to live and work. Despite the high correlations among these three models, there are some important differences in the ordering of regions of the country according to these alternative indices, as well as the magnitudes themselves. Using an inappropriate index for adjusting salary or expenditure data can lead to significantly different conclusions about the levels of educational services being provided in different regions of the country.

Future Work

Future work on the analysis of teacher compensation could be improved along two dimensions. First, additional data items are needed to control for teacher quality (e.g., the quality of colleges attended, scores on Scholastic Assessment Tests (SATs), or national teacher exams). A second area in which data could be improved is benefits received by teachers. The current SASS does not report data that would permit determination of the value of benefits, which could easily add as much as 30-40 percent to teacher salaries. In addition, future research in this area should expand the analysis of teacher salaries to other certified and noncertified personnel, as well as nonpersonnel resources. While it is expected that patterns of school administrator costs will be similar to those for teachers, noncertified personnel tend to operate in more localized labor markets, and in the past have been found to have somewhat different patterns of cost variation than certified personnel (Chambers 1978). Finally, in order to develop a comprehensive cost-of-education index, it will be necessary to obtain some data on the variations in the costs of nonpersonnel resources, which account for approximately 15 percent of school budgets.


Augenblick, J., and K. Adams. 1979. An analysis of the impact of changes in the funding of elementary/secondary education in Texas: 1974-75 to 1977-78. Denver, CO: Education Commission of the States.

Barro, S. M. 1974. "The impact of intergovernmental aid on public school spending." Diss. Stanford University.

Barro, S. M. 1992. Cost-of-education differentials across the states. Washington, DC: SMB Economic Research, Inc.

Brazer, H. E. 1974. "Adjusting for difference among school districts in the costs of educational inputs: A feasibility report." In Selected Papers in School Finance. Washington, DC: Department of Health, Education, and Welfare, Office of Education.

Chambers, J. G. 1975. "The impact of collective negotiations for teachers on resource allocation in public school districts." Unpublished Diss. Stanford University.

Chambers, J. G. Fall 1978. "Educational cost differentials and the allocation of state aid for elementary/secondary education." The Journal of Human Resources 13(4).

Chambers, J. G. 1979. "School district behavior, markets for educational resources, and the implications for public policy: A survey." In Economic Dimensions of Education. Washington, DC: National Academy of Education.

Chambers, J. G. 1980. "The development of a cost of education index: Some empirical estimates and policy issues." Journal of Education Finance 5(3): 262-281.

Chambers, J. G. 1981a. "Cost and price level adjustments to state aid for education: A theoretical and empirical review." In Perspectives in State School Support Programs, second annual yearbook of the American Educational Finance Association. Ed. K. Forbis Jordan. Cambridge, MA: Ballinger Publishing Co.

Chambers, J. G. Winter 1981b. "The hedonic wage technique as a tool for estimating the costs of school personnel: A theoretical exposition with implications for empirical analysis." Journal of Education Finance 6(3): 330-354.

Chambers, J. G. and W.J. Fowler 1995. Public school teacher cost differences across the United States. Washington, DC: U.S. Department of Education, . Publication No. 95-758.

Chambers, J. G., and T. Parrish. 1984. The development of a program cost model and a cost-of-education index for the state of Alaska: Final report. Volumes I-IV. Stanford, CA: Associates for Education Finance and Planning, Inc.

Feldstein, M. S. 1975. "Wealth neutrality and local choice in public education." American Economic Review 65(1): 765-789.

Grubb, W. N., and J. Hyman. 1975. "Constructing teacher cost indices: Methodological explorations with California unified school district." In Selected Papers in School Finance. Washington, DC: U.S. Department of Health, Office of Education, and Welfare.

Johnson, Frank. 1992. Public elementary and secondary state aggregate data, for school year 1990-91 and fiscal year 1990. NCES Report No. 92-033. Washington, DC: U.S. Department of Education, National Center for Education Statistics.

Kenny, L., D. Denslow., and I. Goffman. 1975. "Determination of teacher cost differentials among school districts in the state of Florida." In Selected Papers in School Finance, 1975. Ed. E. Tron. Washington, DC: U.S. Department of Health, Education, and Welfare.

Ladd, H. F. 1975. "Local education expenditure, fiscal capacity, and the composition of the property tax base." National Tax Journal: 145-158.

McMahon, W., and S. Chang. 1991. Geographical cost-of-living differences: Interstate and intrastate, Update. Normal, Illinois: Center for the Study of Educational Finance, MacArthur/Spencer Special series of November 20.

Wendling, W. 1979. Cost-of-education indices for New York state school districts. Denver, CO: Education Commission of the States.


  1. See Chambers (1981a, 1981b) for methodological discussions. See Chambers and Parrish (1984) and Chambers (1978, 1980) for a comprehensive empirical study of educational cost differences. For work of other authors on the CEI, see Augenblick and Adams (1979); Brazer (1974); Grubb and Hyman (1975); Kenney, Denslow, and Goffman (1975); and Wendling (1979).
  2. Since some portion of the remaining 20 percent of school district budgets is allocated to personal service contracts (e.g., psychological services, physical and occupational therapy, consultants, repair services, and legal services), expenditures allocated to personnel actually exceed 80 percent.
  3. For example, see the NCES publication, Public Elementary and Secondary State Aggregate Data, for School Year 1990-91 and Fiscal Year 1990 (NCES Report No. 92-033).
  4. Example of studies of the demand for educational expenditures and demand for educational resources (e.g., staff/pupil ratios) at the local level include Barro (1974), Chambers (1975 and 1979), Feldstein (1975), and Ladd (1975).
  5. A more detailed presentation of the analysis of the relationship between teacher salaries and these teacher and job attributes is contained in Chambers and Fowler (1995).
  6. The TCI described in this paper is based on what was referred to as the regional-level TCI in Chambers and Fowler (1995). The regional-level TCI includes only regional- or county-level variables in the computation of the index. A district-level TCI, which is calculated in the full report (Chambers 1995), includes both regional, as well as district-level variables in the index. Only the regional-level TCI is presented in this paper, since the regional factors are more easily interpreted and the standard error of the regional-level TCI is smaller than for the district-level TCI.
  7. The standard errors of the statewide average index values are generally below 1 percentage point for all but 13 states. Thus, a 95 percent confidence interval for the vast majority of states would be smaller than plus or minus 2 percent. With the exception of Alaska, the standard errors of the five highest cost states range from 0.6997 in Connecticut to 1.3580 in New York. The standard error for the statewide average in Alaska is 2.1859. The standard error is higher for districts further away from the overall average.
  8. This excludes Hawaii and the District of Columbia, each of which have only one district.

[Introduction and Overview] Previous Table of Contents Next[Meeting the Challenge of Devolution: How Changing Demigraphic and Fiscal Contexts Affect State Investments in Education]