Chapter II

REVIEW OF THE RELATED LITERATURE

Education Production Function

Since passage of The Elementary and Secondary Education Act in 2002, better known as the No Child Left Behind law or NCLB, a significant amount of educational and economic research has been devoted to studying every aspect of this controversial law.  The objective of much of this research has been to offer prescriptions for schools and teachers that can help improve student achievement.  Research findings of conditions that affect student achievement tend to focus on a single condition or small number of conditions.  The variety of findings and suggestions are quite impressive.  Fortunately or unfortunately, schools are constrained by resources and cannot try every prescription.

A review of the contemporary literature produces a myriad of different suggestions put forth by well meaning researchers.  Some researchers feel very passionate about their proposed improvements to the public education system.  It is naive to expect that any one body of research will produce a “silver bullet” that can improve student achievement in every school in this country.  The complexity of the public education system in this country is beyond the comprehension of the vast majority of the people who use it, not to mention the people who are charged with making it work.

The idea that one can control the conditions of the educational environment to produce expected results is borrowed from manufacturing or, more specifically, industrial engineering (Eden, 1994) .  One of the goals of industrial engineers is to identify critical costs and potential bottlenecks that affect the production cycle.  From this information, industrial engineers can strive to optimize the flow of raw materials and applied labor to create the most efficient production facility.  This process is referred to as a production function.  Authors such as Skinner (Skinner, 1969), Argote (Argote & Epple, 1990), and Brynjolfsson (Brynjolfsson, 1993) make extensive use of the production function in various articles related to industrial engineering.

Controllable Factors in the Education Production Function

Many education researchers have been influenced by the value of the industrial production function to adapt it to the process of creating the best possible educational environments.  A search of the phrase “education production function” in Google Scholar returns 1,400 peer-reviewed articles that contain the exact phrase (Google Scholar, 2008a).  The authors identified as “Key authors” on the topic include Greenwald, R., Hanushek, E., Hedges, L., Krueger, A., and Laine, R.  (Google Scholar, 2008b).  Hedges, Laine, and Greenwald employed the education production function in “Does Money Matter?” (Greenwald, Hedges, & Laine, 1994) and in “The Effect of School Resources on Student Achievement” (Greenwald, Hedges, & Laine, 1996).  The later study contained a review of 60 education production functions studies performed by other researchers.  More than 880 other peer-reviewed papers have cited these two articles.

Hanushek’s most cited works that employ the education production function include “Assessing the Effects of School Resources on Student Performance: An Update” (Hanushek, 1997) and “Teachers, Schools, and Academic Achievement” (Rivkin, Hanushek, & Kain, 2001).  Together these works are cited over 900 times in other peer-reviewed papers.

The most cited works by Krueger that employ the education production include “Experimental Estimates of Education Production Functions” (Krueger, 1999) and "Reassessing the View that American Schools are Broken" (Krueger, 1998) .  These two articles by Krueger are cited over 600 times in other peer-reviewed literature.    These are just some of the many examples of the use of the education production function.

The education production function has its proponents and detractors.  There is also considerable debate over which resources of the education production function model lend themselves to being controlled by schools, districts, and teachers.  Some elements of the education production function are more controllable than others; and, some are not controllable at all (Krueger, 1999) .

In an article published in “Education Policy”, authors Rau, Shelley, and Beck (2001) examined Illinois' high-stakes assessment system.  Their research used the results from the state's Illinois Goals Assessment Program (IGAP), which was the predecessor to the ISAT tests.  They arrived at the conclusion that 80% of the variance in schools' test scores is beyond individual school control.  That is to say the vast majority of conditions that influence student achievement in Illinois cannot be controlled or changed by teachers or administrators (Rau, Shelley, & Beck, 2001) .  Some of the factors identified in this study that account for up to 80% of this variance include the percentage of students who are poor, the percentage of minority students, and the residential mobility of the students’ families.  The authors argue further that the simplistic notion of pass or fail for a school or district is tantamount to racial classification and may violate the 14th amendment of the U.S. Constitution.  To support their argument, they offered these statistics from the state's own data:  schools that are labeled as "exceeding expectations" are, on average, made up of only 11% low-income and 4% black students; schools that are labeled as being on the "academic watch" are 95% low-income and 90% black.  NCLB sanctions schools based more on their levels of poverty and racial characteristics than they are on gain scores or performance indicators that take into account the characteristics of their students.

In another report published in “Contemporary Economic Policy”, Hoerandner & Lemke (2006) studied the perceived achievement gaps across socio-demographic groups of students.  The purpose of the study was to quantify the degree to which low-performing school could be expected to close the achievement gap with average performing schools, provided the low-performing schools were funded and equipped comparable to average performing schools.  The data used for the study came from ISAT scores.

The authors found that 30% to 50% of the gap is due to non-controllable school characteristics such as racial composition, mobility rate, and percentage of low-income students (Hoerandner & Lemke, 2006) .  After examining the non-controllable characteristics that affect the gap between pass and fail on the ISAT test, the researchers then turned their attention to the factors that schools can control.  They found that factors such as class size, teacher quality, and time devoted to subject area account for about 10–25% of the gap in pass rates.  The authors concluded that there is a non-controllable portion of the education production function that is characterized primarily by the students' socio-demographic status.  This status is reflected in areas of lower tax revenues, higher percentages of low-income status, greater mobility rates and lower attendance rates all contributing to greater failure rates on the ISAT test.

Another study published in the “American Education Research Journal” focused on the relationship between educational resources and student outcomes (Hill, Rowan, & Loewenberg-Ball, 2005) .  The study measured the resources utilized by students, teachers, and school districts and compared them to student achievement on standardized math tests.  The students' family background and socio-economic status played a significant role in the students' performance.  Although significant, this characteristic is not in the control of the teachers and school districts and thus cannot be factored into the education production function.  The factors that were found to be significant to students’ performance included teacher salaries, teacher quality, teacher-to-pupil ratios, material resources including books, supplies, manipulatives, and facility characteristics (Hill, Rowan, & Loewenberg-Ball, 2005) .

The balance of this literature review focuses on the research pertaining to the effects of teacher and school district controllable factors on student achievement.  These reviews correspond to controllable factors that are represented on the Illinois School Report Card, which is prepared annually for every public elementary and secondary school in Illinois.

Influences on Student Achievement

Teacher Compensation and Student Achievement

One of the metrics reported in every Illinois School Report Card is average teacher salary (ISBE, 2007) .  The National Education Association (NEA), which represents 3.2 million teachers from thousands of teachers’ unions in the United States, has a strong opinion on the topic of teacher salaries.  The NEA believes that teachers are subject to the same free market forces of supply and demand as most other professionals.  As with other professions, higher compensation tends to attract greater numbers of teachers for positions and also higher quality teachers.  Unlike other professions that have suffered recently due to downsizing, demand for teachers has increased.  Unfortunately, the low compensation compared to other professions causes half of all new teachers to leave the profession in five years.  This amounts to a $2 billion attrition cost to American schools each year (National Education Association, 2007) .

Several factors impact the calculation of teacher compensation.  One of the largest factors is years of experience, which will be addressed later in this literature review.  Some conditions may skew the correlation between teacher salary and student achievement on standardized tests.  For example, student quality and other non-pecuniary characteristics that teachers value are sometimes “capitalized ” into teacher wages as compensating differentials.  The effect is that teachers may be willing to accept lower wages in districts with better working conditions or higher student quality, both of which are likely to be correlated with higher student outcomes (Loeb & Page, 2000) .  Despite this, the report concluded that when controlling for job characteristics, local wage opportunities, and/or local amenities, teacher pay has a large effect on student achievement.

RAND Education (2006) produced a report measuring the effects of teacher pay on student performance in Illinois.  The study covered approximately 3,800 schools in Illinois from 1987 through 1998.  Student achievement on standardized reading and math tests for 3rd and 8th grades were used.  Controlling for factors such as average teacher education and experience, the report indicated that higher teacher salaries were associated with higher student performance (RAND Education, 2006) .

Evidence relating teacher salaries to student achievement is not always strongly convincing.  A study (Hanushek, Kain, & Rivken, 1999) performed by the University of Texas at Dallas for the Texas public schools demonstrated this.  The project was a multi-year study to obtain a better understanding of the determinants of student performance.  The long-term objective was to provide knowledge and a research base of ways to improve the performance of public schools.  When it came to the relationship between teacher salaries, teacher quality and their impact on student performance, the evidence about the strength of any such relationship is inconclusive.  The only significant relationship between salaries and student achievement holds for existing experienced teachers.  Teacher salary for new and probationary teachers had no statistical significance on student performance (Hanushek, Kain, & Rivken, 1999) .

Teacher Experience and Student Achievement

Teacher experience is another metric gathered in the Illinois School Report Card  (ISBE, 2007) .  Teacher experience is strongly correlated to teacher salary.  But the question is whether experience translates into higher achievement for students.

There are other difficulties associated with comparing teachers' experience to student achievement.  For instance, unless held constant, more experience is often associated with graduate work or a master’s degree.  If it can be generalized that a more educated teacher makes a better teacher than this would confuse the data comparing teacher experience to student achievement.

Supply and demand of teachers, particularly those in high-demand fields such as math, science, and special education, also play a role in confounding the relationship between teacher experience and student achievement.  School districts that accept the hypothesis that more experienced teachers are better qualified to increase student achievement would naturally choose to hire more experienced teachers.  It is these teachers, particularly those certified in math and science, who tend to leave the profession for employment in the private sector.  Turnover is greater for these types of teachers and gathering accurate longitudinal data comparing their experience to student achievement is difficult (Wayne & Youngs, 2003) .

Another study (Goldhaber & Brewer, 1999) examined teacher experience and student achievement by comparing the type of secondary license held by the teacher.  Teachers in this study held one of the following licenses:  initial, standard, emergency, or alternative.  An initial license indicated a teacher with 0 to 4 years experience.  A standard license indicated a teacher with 5 or more years experience and the average experience in this group was high, almost 18 years experience.  An emergency license indicates a teacher who is not state licensed as a subject matter teacher but, due to his or her experience (averaging seven years) and the severe shortage, was granted an emergency teaching license.  Finally, the alternative license indicated a teacher who had completed an alternative certification program and was, essentially, in his or her first year of teaching.

This study employed multiple linear regression analyses on these and other student and school environmental characteristics to explain students’ scores on twelfth-grade standardized mathematics and science tests.  The results from this study indicate that, with 90% confidence, the model indicated that the coefficient of teacher-years experience has a positive correlation.  For example, teachers with 10 years of experience in the subject area could be expected to influence student achievement scores in that subject by as much as 21%.  Similarly, teachers with 20 years of experience in the subject area could be expected to influence student achievement scores in that subject by as much as 42% (Goldhaber & Brewer, 1999) .  This study of teacher experience and its influence on student achievement is valuable because of its use of multiple linear regression analysis to account for and control the factors that influence student achievement.

Another study (Darling-Hammond, 1999) prepared by the Center for Teaching Policy at the University of Washington-Seattle found much evidence that supports the idea that teacher experience results in greater student achievement.  For example, teachers with less than three years experience are typically less effective than more senior teachers (Darling-Hammond, 1999) .  For senior teachers, their benefit to student achievement appears to level off after five years experience; however, this applies mostly to teachers who do not continue to grow and learn (e.g., pursue graduate degrees or other professional development opportunities).  Corroborating this finding was a study of 5-year teaching programs where teachers continued their education beyond their Bachelor of Arts (B.A.) degree into graduate studies.  These teachers showed effectiveness similar to that of senior teachers (Darling-Hammond, 1999) .

The findings regarding the influence of teacher experience reinforces several production functions completed on this topic, specifically that teacher experience has a strong positive correlation to student achievement.  The policy implications of this study to a school district are clear.  Districts should seek the most experienced teachers available, particularly those with a commitment to life-long self-development.   As was demonstrated in this study, estimates of the achievement gains associated with expenditures on teacher education were found to be the most productive investment for schools.

Teacher Education and Student Achievement

Teacher education is another metric reported in every Illinois School Report Card  (ISBE, 2007) .  The data is represented as percentage of teachers with Bachelor of Arts (B.A.) and percentage of teachers with Master of Arts (M.A.) degrees and higher.

Studies focusing on teacher education and their relation to student performance have been conducted for decades.  Most of the studies in the past three decades have produced indeterminate results.  Some studies have produced positive correlations, while others have produced negative correlations.

The spring 2003 edition of “Review of Education Research” (Wayne & Youngs, 2003) contained a report comparing teacher characteristics to student achievement.  The report acknowledged that much of the preceding research was indeterminate regarding teacher education and student performance.  It attributed these findings to incomplete data with respect to teacher education.  In most of the previous research, specific data on teacher coursework or the teachers' B.A. or M.A. major was not taken into account.  When research considers a teacher's coursework or the relevancy of his or her undergraduate or graduate degree, statistical significance becomes apparent.

This report produced two important findings with respect to teacher education and student achievement.  First, high school students taught by teachers with bachelor's degrees in mathematics learned more than students whose teachers had bachelor's degrees in non-mathematics subjects.  Second, when high school students were taught by teachers with master's degrees in mathematics, they had higher achievement gains compared to students whose teachers held advanced degrees in non-mathematics subjects, including general education master's degrees, or no advanced degree at all.  The contributions of these two indicators of teacher education were made independent of other teacher characteristics, such as years of experience and math certification (Wayne & Youngs, 2003) .

Darling-Hammond (1999) also examined teacher preparation in her report “Teacher Quality and Student Achievement”.  The study reported that teachers with full certification and a major in the subject area they teach produce greater student achievement results than, for example, teachers with general education master's degrees.  One explanation offered for this result is that many general education master's degree programs include a wide range of content that is not specifically beneficial to helping students perform better on standardized testing.  Master's content such as administration, leadership, legal and financial courses has little to do with improving student achievement.

The findings of this research are strong with respect to policymaking, especially as it pertains to the classroom environment.  They suggest that the student achievement gains resulting from teachers who are subject matter experts and also high quality deliverers far outweigh other environmental changes in the classroom, including the gains occasioned by smaller classes.  This is corroborated by the Tennessee experiment (Mosteller, 1995) where smaller class sizes had a significant impact on student achievement when they were accompanied by the employment of highly qualified teachers.  The opposite of this finding is supported with recent data.  Using California’s recent class size reduction initiative, it was found that large-scale hiring of unqualified teachers to reduce class sizes offset any achievement gains that could be realized by employing qualified teachers as in the Tennessee experiment (Bohrnstedt & Stecher, 1999) .

In “Effects of Teachers' Mathematical Knowledge for Teaching on Student Achievement”, researchers arrived at similar findings.  Teachers have to be highly effective in using the knowledge they possess.  For example, highly proficient math teachers can only increase student achievement if they are able to use their own knowledge to perform the tasks they must enact as teachers.  Effective teachers must be able to listen to students, select and make good use of assignments, manage discussions, and guide work on skills building.  These results are consistent with the education production functions that focus on teacher education as a factor for improving student achievement.  A direct measure of teachers' content knowledge and teaching ability outweighs measures such as courses taken or experience.  A teacher's knowledge should be at least content specific or, even better, specific to the knowledge used in teaching the students (Hill et al., 2005) .

In another study conducted by Goldhaber and Brewer (1999) for the Fordham Foundation, three categories of math teachers were compared.  The study used data from the National Educational Longitudinal Study of 1988.  The first category of math teachers was teachers with certifications in areas other than math and no B.A. or M.A. in math.  The second category of math teacher was teachers who held B.A.’s and M.A.’s in math but were not certified in math.  The third category of math teachers was teachers who were certified in math but did not have B.A.’s or M.A.’s in math.  The study did not include teacher with general education master's degrees that were or were not certified in math.

The results of the study concluded that having a B.A. in education only with no math certification had a statistically significant negative impact on scores in math for the tenth-graders tested.  The students of teachers who held B.A.’s and M.A.’s in math but did not hold math certifications, on average, tested one point higher, the equivalent of more than a third of a year of schooling beyond the aforementioned group.  And the students of the teachers who did not hold B.A.’s or M.A.’s in math but were math certified, on average, tested two points higher, which represents more than three-quarters of a year of schooling beyond the first group.  In summary the best scenario for maximizing student achievement with respect to teacher education would be to have certified math teachers who also hold B.A.’s and M.A.’s in math; after this are the certified math teachers regardless of B.A. or M.A. specialties, followed by the teachers with B.A.’s and M.A.’s in math.  The least favorable teachers for helping students improve achievement in math are teachers with general education certifications and no degrees in math (Goldhaber & Brewer, 1999) .

In summary, the effect of a teacher’s education on student achievement is well documented.  Undergraduate degrees and graduate degrees in mathematics have a statistically significant impact on student achievement math assessments, provided the holders of these degrees are able to use their knowledge to teach effectively.  Math certification appears to be a legitimate means of gauging a teacher’s ability to effectively teach math.  The least desirable situation for student achievement is to have an instructor who is neither certified nor holds undergraduate or graduate degrees in the subject.  The most desirable situation for student achievement exists when a teacher holds undergraduate and graduate degrees in the subject and is also state certified in the same subject.

Teacher Gender and Student Achievement

Another metric reported in every Illinois School Report Card is the percentage of female and male teachers  (ISBE, 2007) .  In 2007 the Illinois Education Research Council collected specific information regarding teacher gender for the state of Illinois report.  The following is a summary of this information.

The data for this report was collected between the years 1987 and 2006.  For this period, females constituted 84% of all teachers in elementary and middle school grades.  For the same period, females made up 52% of all teachers in high schools.  Overall the gender mix for the state of Illinois is 75% female and 25% male (DeAngelis and Presley, 2007) .

The question of whether teacher gender has an impact on student achievement was addressed in part in a report published in “Industrial and Labor Relations Review” (Ehrenberg, Goldhaber, and Brewer, 1995) .  This report used data collected for the National Education Longitudinal Study of 1988 (Department of Education, 1988).  One of the overall findings was that teacher gender did not affect student achievement from 8th grade to 10th grade in the areas of reading, social studies, math and science.  The teachers' gender did, however, sometimes influence the teachers' subjective evaluation of the student for the 8^th to 10^th graders.  For example, the study compared the effects of white female and white male teachers on math and science test scores among white female students.  The study concluded that white female teachers did not increase test scores of white female students.  The white male teachers did appear to increase the test scores of the white female students.  In addition the study found that these same white female teachers gave higher subjective scores to their female white students than did the white male teachers.

The findings from this report can be interpreted in two conflicting ways.  In the first case, it can be argued that if the amount of student learning is important, then the gender of the teacher does not matter.  In the second case, it can be argued that if teachers' subjective evaluations of students influence and encourage the students to maintain the commitments to their aspirations, then the results suggest that, in some cases, the gender of the teacher does matter.

Other research on teacher gender and student achievement is similarly inconclusive.  For example, Saha (1983) concluded that there was no meaningful difference in general student achievement based on whether a teacher is male or female.  He also concluded that when it came to math and science, students with male teachers had greater achievement in these curricular areas than those with female teachers.  His conclusions were based on careful examination of the effects of teacher gender on students' achievement in 21 less developed countries.

Supporting Saha's findings, Warwich and Jatoi (1994) analyzed the data from the first national survey of primary schools in Pakistan.  They concluded that, in Pakistan, teacher gender is a much greater predictor of the students' math achievement than the gender of the students.  Students who had male teachers scored significantly higher on math tests than students of female teachers.

Contrary to Saha, Warwich and Jatoi, Mwamwenda and Mwamwenda (1989) concluded that primary education students in Botswana who had female teachers had significantly higher achievement scores in math and other subjects than students of male teachers.  These results agreed with research completed by Lee and Lockheed (1990), which suggested that primary education female students in Nigeria who had female math teachers had much more positive attitudes towards math than female students taught by male teachers.

Brophy (1985) concluded that, overall, female and male teachers are much more similar to each other than different from one another.  There are, however, differences in the way they interact with students.  For example, female teachers tend to be more student-centered, indirect and supportive of students than male teachers.  Female teachers appeared to use more classroom discussion and encourage collaboration and affective learning techniques.  Female teachers are more likely to promote more collaborative learning environments, compared to their male counterparts (Li, 1999) .

Class Size and Student Achievement

Another metric reported in every Illinois School Report Card is class size, or the pupil-to-student ratio  (ISBE, 2007) .  Class size is one of the classroom environment statistics that receives significant attention from educators, administration, parents, and the media.  Although it would seem reasonable that a teacher's time shared between fewer students would result in greater student achievement, there is research that suggests this is not always the case.  This portion of the literature review presents findings that smaller class sizes and also well-managed larger class sizes lead to greater student achievement.

McGee (2004) used data collected from Illinois' Golden Spike schools to determine if class size matters.  Illinois Golden Spike schools are institutions that have a sustained record of closing the achievement gap.  Because the Golden Spike schools cut across all socio-economic grains, McGee was able to illustrate the differences in academic performance between low-income and children and their peers; between minority children and their classmates; and between schools that serve mostly low-income families and schools that have small percentages of low-income students.  McGee used quantitative and qualitative methods and discovered commonalities among Golden Spike schools including leadership, literacy level, teacher qualities, and community involvement.  He concluded that characteristics such as school size, alignment of curriculum to state standards, and class size made little, if any difference towards closing the students’ achievement gap.

In still another study, Hoxby (2000) used naturally-occurring class size variations from a set of 649 elementary schools.  Hoxby concluded that the size of the class had no effect on student achievement.  Similarly, Johnson (2000) used National Assessment of Educational Progress reading scores to study the effects of class size.  Johnson also concluded that the class size had no effect on student achievement.

A contrary conclusion was reached by Ferguson (1991) in an analysis of Texas schools.  Ferguson used data from more than 800 districts containing more than 2.4 million students and concluded there is a significant relationship among teacher quality, class size and student achievement.  Specifically, for first through seventh grades, he found that district student achievement fell for every additional student beyond an optimal student-to-teacher ratio of eighteen-to-one.  Wenglinsky (1997a) also used the Texas data to study class size effects on fourth graders in more than 200 districts and eighth graders in 182 districts.  Wenglinsky found that smaller class size positively affected math scores for fourth graders and improved the social environment for eighth graders, which in turn produced higher achievement. These effects were greatest for lower socio-economic students.

Instruction Time and Student Achievement

The amount of instructional time students are receiving in certain curricular areas, especially those areas subject to NCLB testing, is another metric reported in every Illinois School Report Card (ISBE, 2007) .  For example, eighth grade students in Illinois receive an average of 51 minutes per day of instruction in Math, 44 minutes per day of instruction in Science, and 93 minutes per day of instruction in English/Language Arts.  The majority of research comparing instructional time to student achievement concludes that more instructional time produces greater student achievement.  However some research finds to the contrary, unless specific expectations are established, specific methods are employed, and specific educational settings are developed and maintained.

According to the research report “Time for Change” (Massachusetts 2020, 2005) , additional time in school is correlated with an increase in proficiency and higher achievement on standardized tests.  If more content needs to be taught, it stands to reason that there must be more time to teach it.  This study found that extended learning time opportunities in the schools showed a substantial increase in student achievement scores.

No discussion about instructional time can made without reference to the landmark article written by John B. Carroll in 1963 titled "A Model of School Learning" (Carroll, 1963) . The findings in this article have been cited nearly 700 times since its release.  In this report, Carroll defined the degree of learning as time actually spent in learning divided by time needed for learning.  In the Carroll model, the time needed for a given student to learn a given concept depends upon five factors:

The findings by Carroll are widely regarded as the beginning of modern inquiry into the effects of time factors in the learning process.

b.      Appropriate to students' ability and maturity levels;

c.       Assigned regularly i.e., on a daily basis;

d.      Assigned in reasonable amounts i.e., 30 minutes per subject per day;

e.       Well explained and motivational;

f.       Collected and reviewed during class time;

g.      Used as an occasion for giving feedback to students; and

h.      Supported by parents.

5. Increasing allocated or engaged time is more beneficial to lower ability students than higher ability students.  Additionally, increased time on task is more beneficial to highly structured subjects such as math and foreign language.

6. The time allocated for education is not a good barometer of student achievement.  Proposals to extend the school day or school year would fall far short of their desired outcomes for student achievement and would be extremely costly.  What works is specialized and focused additional time devoted to the specific subject areas that students are lacking in.  This research also finds that the tools used in re-teaching must include different materials, examples, and demonstrations than those used during initial instruction.

Administrator Compensation and Student Achievement

Average administrator salary is another metric reported in every Illinois School Report Card  (ISBE, 2007) .  In Illinois, for 2006, the statewide mean administrator salary was $100,396.  Despite this attractive salary, which exceeded the average teachers' salary by about $43,000 in 2006, there is a severe shortage of administrators.  For example, the shortage of qualified applicants for principal vacancies in all states and across all geographical areas has been well documented (e.g., Adams, 1999 ; Educational Research Service, 1998 , 2000 ; McAdams, 1998 ).  The later report, issued by the Educational Research Service, confirmed that the principal applicant shortage continues to be prevalent.

Three trends have had a significant influence on the present shortage of principals.  First, many principals of the "baby-boom" generation are reaching eligibility for retirement.  A greater percentage of these principals is deciding to exercise their option to retire.  Second, many experienced and practicing principals are choosing to leave the profession and return to classroom teaching.  Finally, one of the significant sources of principals is declining.  Teachers who pursue principal certification credentials have always been a major source of new principals.  Research shows that fewer teachers who earn principal certification are deciding to pursue careers in administration.  A poll of over 400 superintendents cited inadequate salary, excessive time requirements, and the potential for job-related stress as the reason why fewer teachers are pursuing careers as principals Educational Research Service, 1998 , 2000 .

A person who aspires to be a principal, especially someone who has been a classroom teacher, must be prepared for the commitment of being a principal.  A national survey of superintendents found that when job responsibility, job stress, and hours worked are factored in, the principals' pay is far from adequate Educational Research Service, 1998 , 2000.  Principals typically work more hours per day and more days per year.  When one considers that the difference between a principal’s salary and that of a mid-career teacher can be as little as $10,000, it is not hard to see why teachers with administrative credentials would be reluctant to pursue careers as principals (Newton, Giesen, Freeman, Bishop, & Zeitoun, 2003) .

Newton's (et al.) research finds that there is a significant shortage of administrators, particularly principals, in America's schools.  Any profession that suffers from a shortage of qualified people will ultimately experience quality issues as it attempts to fill critical positions.  The role of administrators, and principals in particular, is very important to the achievement of the students they are responsible for.

Recently, a study was conducted of Arkansas Public Schools to try to measure the effects of administrative salaries on student education (Rainey & Murova, 2003) .  This study characterized student education using the term “technical efficiency”.  Technical efficiency was defined as the ability of the school district to maximize its ability to increase student achievement given its resources.  In short, this study developed an education production function comparing administrative salaries (as a percentage of expenditures per student) to student achievement.  This study employed stochastic frontier analysis performed with a Cobb-Douglas or trans-log parametric function.  For the purposes of this study, schools districts were divided into three sized categories.  Small school districts were defined as districts with less than 500 students.  Medium school districts were defined as districts with 500 to 1,250 students.  And large school districts were defined as districts with more than 1,250 students.  It is also worth noting that expenditures per student was the indicator used to determine how much money was available to attract quality teachers who directly impacted student achievement.

Not surprisingly, small and medium size school districts were hit hardest when the ratio of administrative salaries to per student expenditures was increased.  The higher ratio would lead to increased inefficiencies (lower student achievement) because there was less money available to attract quality teachers or provide other beneficial achievement enhancing services.

Summarizing, Rainey and Murova (2003) found that about 40% of all schools (regardless of size) registered efficiencies below the mean for their respective groups.  Medium and large school districts will not benefit from a general increase in student expenditures.  In order to increase efficiencies and have a positive affect on student achievement, increases in expenditures per student must be focused on increasing average teachers' salaries.  Simply increasing general expenditures to reduce the ratio of administrative salaries will not affect a positive change in student achievement.  The increase must be mandated to increase average teacher salaries to produce a positive effect on student achievement.

            The findings for small school district are slightly different.  Small school districts will also benefit academically from a smaller ratio of administrative salaries to student expenditures.  However, the additional money from increased expenditures per student for smaller schools needs to be used differently than it is for medium and large school districts.  Small school districts struggle more with operation and maintenance costs.  These costs tend to be fixed, independent of the number of students in the school.  Small school districts need to be able to cover these fixed costs before they can allocate remaining funds to attracting high-quality teachers or providing other services known to enhance achievement (Rainey & Murova, 2003) .  This is not to say that small school districts are not interested in attracting high quality teachers.  It is merely the reality of operating and maintaining the infrastructure.  The results imply that an increase in student population for small school districts would provide additional capital available to attract quality teachers and increase student achievement.

A similar research study (Strauss, 2003) was conducted to explore the effects that administrators have on student achievement in Pennsylvania.  The study spanned the years 1984 to 1999.  This report was based on the administrative records of individuals and not their responses to sample survey questionnaires.  Additionally, the population for this report was all school administrators in the state of Pennsylvania.

Part of the preliminary research covered the comparison of administrator salaries to teacher salaries, taking into account the inflation rate.  This part of the research found that teachers' salaries in Pennsylvania rose faster than the inflation rate.  The ratio of lowest to highest paid teacher stayed the same (1:2) over the course of the study.  In other words, highest paid teachers made about twice as much as lowest paid teachers but all increased at roughly the same rate, in excess of the inflation rate.

By contrast, the average administrators' salary declined compared to the inflation adjusted teachers' salaries.  This was most pronounced in the case of elementary school principals.  At the same time, the responsibilities of administrators increased as evidenced by the increase in average hours worked.

Since overall real compensation for administrators has decreased, the question arose as to the source of the positive gains in student achievement and what effect administration had on the gains.  Examining the background of the administrators shed light on their ability to positively influence student achievement.  Having previous experience as a K-12 guidance counselor was correlated to higher math and reading scores.  Having previous experience as an assistant superintendent for instruction raised students’ reading scores for elementary principals.  Administrators who held special education certifications vastly improved secondary math scores.  And the more years of experience as a principal of a middle school the administrator held, the better the middle school math scores.  Administrators in these examples had specialty training beyond general education certifications.  This may partly explain their high pay and the higher performance of the students who they administrator (Strauss, 2003) .  If true, this finding is not at all surprising.  If one accepts that teachers with advanced specialty training will produce students with higher achievement in their respective areas, then why would this not be true of administrators who rose from the ranks of specialist teachers?

Research conducted by Brewer (1993) demonstrated a positive correlation between increased administrative pay and higher student performance.  Brewer examined whether or not high school principals’ characteristics affected test score data collected as part of the “High School and Beyond” program.  This program measured student achievement gains between 1980 and 1982 in 2,070 sophomore and subsequent senior test scores, while holding constant community and student socio-economic characteristics.  Brewer found that, in schools where principals emphasized academic excellence, student performance was higher.  The effects were quite large and amplified by the percentage of faculty appointed by the principal.  Brewer found that higher principal salaries (when compared to the mean teacher salaries in the school) produced greater gains in test scores.  Specifically, when the relative salary of a principal was increased by 5%, the mean gain test score rose by 20%.

Student Expenditures and Student Achievement

The last metric considered in this literature review is the amount invested per student for instructional purposes.  Expenditure per pupil is reported in every Illinois School Report Card  (ISBE, 2007) ; however, the reported figure lags the report date by one school year.  For example, the 2006 average Illinois instructional expenditure per student was $5,366 as reported for the 2004-2005 school year.  The lag is the result of the tax assessment procedures.  Illinois residents pay property tax for the assessed value of their home for the previous year.  Despite the lag, this is a legitimate means of tracking the amount invested per student for instruction on a district-by-district basis.

Wenglinsky (1997b) studied the effects of school district spending on academic achievement.  Wenglinsky holds that the socio-economic status of the school is the prime contributor to the social environment of the school.  The social environment of the school in turn has a strong influence on achievement.  Thus, the focus of his study was the impact that district spending had on improving the school's social environment leading to higher achievement.  Wenglinsky argued that schools with lower socio-economic status have more difficult social and learning environments.  This makes learning more difficult and impacts student achievement.  The basis of his solution was to reduce the students-per-teacher ratio by hiring more teachers.

Wenglinsky found that spending patterns affect student achievement in two ways.  First, Wenglinsky found that expenditures-per-student for instruction strongly influences teacher hiring practices.  This reduces the student-teacher ratio, improves the school's social environment, and raises student achievement, particularly in mathematics.  Second, Wenglinsky found that non-instructional expenditures per students used for central office administration also leads to the hiring of more teachers with the same positive consequences as instructional spending.

Wenglinsky found two areas where additional spending did not result in higher student achievement.  When additional money was spent on administration at the school level, there was no significant increase in student achievement.  Also, when additional money was spent on more highly qualified teachers (e.g., teachers with advanced degrees), there was no significant student achievement gain.  This later finding however did not qualify the nature of the teachers' advanced degree.  Earlier findings suggest that subject matter advanced degrees do strongly influence student achievement where general Masters of Education degrees do not.

A strong opponent to increased education funding as a means of increasing student achievement is Hanushek (1981, 1996).  Hanushek looked at the expenditures per student over the last 100 years and compared it to student achievement.  Expenditures per student have consistently increased an average of 3.5% per year over this period.  During this period his research shows flat or declining student achievement.

In his arguments, Hanushek concludes that there is no relationship between the student expenditures and student achievement (Hanushek, 1997) .  Specifically, he feels that improving student achievement by spending more money to hire more teachers and reducing class sizes is a myth.  He feels that school districts are inherently inefficient organizations that will not become more efficient under their power or the pressure of the citizenry who support them.  Hanushek’s ultimate solution is to establish a system of performance incentives for teachers and administration based on student achievement.

Krueger (2000) took Hanushek to task over his widely publicized findings.  Krueger claims that Hanushek’s research is based on selective evidence.   Krueger argues that Hanushek’s research is based on a single study that is disproportionately weighted to support his position.  Further, Krueger feels Hanushek’s single supporting study includes a large number of estimates (Krueger, 2000) .  Krueger also points out several instances in Hanushek's research wherein data gathering and analysis techniques were sub par for a researcher.  Additionally, Krueger backed up his position with several well-known studies ( McGee, 2004 ; Hoxby, 2000 ; Krueger and Whitmore, 2000 ) linking smaller class sizes to higher student achievement.  Paraphrasing Krueger, if instructional expenditures aimed at reducing class sizes did not have a positive effect on student achievement, then how could school districts and state legislatures continue to use this connection as a major rallying point for increasing student achievement?

Summary of Influences on Student Achievement

            The purpose of this literature review was to introduce the production function model, describe its applicability to education, and identify specific controllable items of the education environment that are reported in every Illinois School Report Card and may influence student achievement.

            The literature review began with an examination of the production function model, in general, and its application to education, in specific.  The controllable items of the education environment that are reported in every Illinois School Report Card were identified.  Research findings for each of the controllable items and their influence on student achievement were gathered and presented.  Whenever available, both sides of the discussion were provided.  Findings supporting and refuting the influence of specific controllable factors and whether they influence student achievement were reported.

            The research in this literature review does not conclusively rule out any of the controllable factors as not having sufficient influence on student achievement.  Observationally, more research existed for some controllable factors than for others.  For example, there is a large amount of research available comparing teacher compensation, teacher experience, class size, or student expenditures to student achievement.  Less research is available comparing teacher education, teacher gender, instructional time, or administrative compensation to student achievement.  This is not to say that the later group has less influence on student achievement than the former group.  The balance of this research report will study the affects of these controllable items on eighth grade ISAT math scores gathered from schools in Cook (excluding Chicago Public Schools), DuPage, Kane, Lake, McHenry and Will counties and reported in the Illinois School Report Cards for 2006 and published on the Illinois State Board of Education website.