Tuesday, November 22, 2022

2014

The Productivity of Public Charter Schools

Gene V Glass
Arizona State University

I. Introduction

On July 26, 2014, the University of Arkansas Center for Education Reform released a
report addressing the relative productivity of charter schools as compared to traditional
public schools. The report is entitled The Productivity of Public Charter Schools. (Note 1) The
report includes a main section in which the cost effectiveness of charter schools (CS) is
compared to that of traditional public schools (TPS) for 21 states plus the District of
Columbia and three Appendices in which various methods and data sources are described.
The authors of the report are as follows: 1) Patrick Wolf, Distinguished Professor of
Education Policy and 21st Century Endowed Chair in School Choice in the Department of
Education Reform at the University of Arkansas in Fayetteville; 2) Albert Cheng, a
graduate student in the Department of Education Reform, referred to as a “Distinguished
Doctoral Fellow”; 3) Meagan Batdorff, founder of Progressive EdGroup and a Teach for
America alumna; 4) Larry Maloney, president of Aspire Consulting; 5) Jay F. May, founder
and senior consultant for EduAnalytics; and 6) Sheree T. Speakman, founder and CEO of
CIE Learning and former evaluation director for the Walton Family Foundation.

The report reviewed here is the third in a succession of reports on charter schools that
have generally claimed that in relation to traditional public schools (TPS), charter schools
(CS) are more effective in producing achievement (Note 2) and less costly per pupil. (Note 3) The current report attempts to combine these attributes into an analysis of the cost effectiveness of CS versus TPS. Results are reported in terms of National Assessment of Educational Progress (NAEP) scores per $1,000 expenditure at the state level for 21 states and the District of Columbia. The claim is made based on these analyses that charter schools, while spending far less per pupil than traditional public schools, generally produce achievement as good as or superior to that of traditional public schools.

The claims made in the report rest on shaky ground. The comparison of achievement
scores between the CS and TPS sectors suffers from multiple sources of invalidity. The
assessment of expenditures in the two sectors rests on questionable data. In the
examination of cost effectiveness analyses, weaknesses in these two areas leave little
evidence on which to base any valid conclusions. If one is calculating “bang for the buck,”
what is left if neither the bang nor the buck can be believed?


II. Findings and Conclusions of the Report

Comparing NAEP achievement obtained in charter schools versus that in traditional public
schools for 21 states and DC, the report concludes that the charter school sector delivers a
weighted average of an additional 17 NAEP points per $1000 invested in math,
representing a productivity advantage of 40% for charters. The report goes on:
· In reading, the charter sector delivers an additional 16 NAEP points per $1000
invested, representing a productivity advantage of 41% for charters;
· Percentage differences in cost effectiveness for charters compared to that for TPS in
terms of NAEP math score points per $1000 invested ranges from 7 percent
(Hawaii) to 109 percent (Washington DC);
· Percentage differences in cost effectiveness for charters compared to that for TPS i n
terms of NAEP reading score points per $1000 invested ranges from 7 percent
(Hawaii) to 122 percent (Washington
DC). (p. 7).

Curiously, in spite of report’s executive
summary touting the superior cost
effectiveness of charter schools versus
traditional public schools, the authors insert
the following caveat in a later section:

. . . our cost effectiveness calculation using NAEP scores has important limitations. Most importantly, it is merely descriptive, not causal, because charter schools might be reporting higher NAEP scores per $1000 invested than TPS because of the characteristics of students attracted to the charter school sector and not because they actually do a better job educating similar students and at a lower cost. (p. 21).

This caveat, which basically undercuts the conclusions and recommendations of the report,
is missing from the press releases and media coverage. (Note 4)


The report concludes with calculations of the superior lifetime earnings, labeled Return on
Investment (ROI), that accrue to pupils educated in charter schools when compared with
those of pupils in traditional public schools. An example of a conclusion from this analysis
follows:

The higher ROI [essentially lifetime earnings] for charters compared to TPS
ranges from +0.4 percent (New Mexico) to +4 percent (Washington DC)
assuming a single year of charter schooling and from 3 percent to 33 percent
assuming a student spends half of their K-12 years in charters. (p. 7).

III. The Report’s Rationale for Its Findings and Conclusions

The relative cost effectiveness of the two sectors, CS and TPS, is based on the examination
of National Assessment of Educational Progress test score averages from those 21 states
and the District of Columbia in which the average score is reported separately for the two
sectors. Associated with these achievement test data are estimates of the per pupil
expenditure in the two sectors for each state. The simple division of the average test score
by the average expenditure at the level of the state is said to produce a cost-effectiveness
ratio that can be compared between the two sectors. Such calculations are frequently
described in basic textbooks, (Note 5) but they are rarely applied in the area of education due to
the complexity of capturing both the outcomes of teaching and the expenditure of funds
for instruction. (Note 6) One might as well ask, “What is the return on expending one’s effort to improve one’s marriage?”


IV. The Report’s Use of Research Literature


Issues such as the relative costs and effectiveness of CS and TPS are hotly contested in the
research literature of the past two decades. (Note 7) By and large, researchers have taken positions
of advocacy and cited related work that supports their position while ignoring conflicting
evidence. The present report continues that pattern. Virtually absent from the report are
citations to works that dispute the position assumed by these authors, their academic
affiliations, or their sponsors. While they do cite Bruce Baker’s work that disputes their
recent claims regarding the inequity of charter school funding (Note 8), they do so only in the
context of what they regard as a refutation of its claims. That refutation is wanting and has
been responded to by Baker. (Note 9) Just as important, a wide-ranging literature disputing the
claim of superior effectiveness of charter schools is completely ignored. (Note 10)
In the current case, the failure to reconcile the reported findings with a large literature of
contrary evidence is particularly egregious. At this stage in the accumulation of research
evidence, those who claim positive effects for charter schools in comparison with
traditional public schools have a burden of proof to demonstrate that their research is not
only sound but that the findings of prior research are somehow invalid.


V. Review of the Report’s Methods


The main argument of the report hinges on the estimation of two things: 1) the relative
performance on achievement tests of CS and TPS, and 2) the cost of educating the average
pupil in CS versus TPS.


For the former estimate, the authors have chosen to use the average statewide scores in
math and reading for those states that report NAEP scores for both sectors, CS and TPS for
FY11. For the latter estimate, i.e., cost, the choice was made to use “revenues received” rather than “expenditures made.” Relying on data from an earlier report, (Note 11) the authors
concluded that the “main conclusion of our charter school revenue study was that, on
average, charter schools nationally received $3,814 less in revenue per-pupil than did
traditional public schools.” (p. 10). At this early point, the report’s analysis runs off the
rails. Revenues received and actual expenditures are quite different things. Revenues
received by traditional public schools frequently involve funds not even intended for
instruction. The report purports to compare “all revenues” received by “district schools”
and by “charter schools,” claiming that comparing expenditures would be too complex. The
problem is that revenues for public schools often fund programs not provided by charter
schools (special education, compensatory education, food, transportation, special
populations, capital costs, state mandated instructional activities, physical education, and
the like.) Charter funding is in most states and districts received by pass -through from
district funding, and districts often retain some or all responsibility for the provision of
services to charter school students—a reality that the report acknowledges but then does
nothing to correct for in the data. There are huge variations across states and within states.
This non-comparability problem alone invalidates the findings and conclusions of the
study.


A sensible comparison of cost-effectiveness between the two sectors would require at a
minimum a parsing of these expenditures that isolates funds spent directly and indirectly
on instruction. No such parsing of revenues was even attempted in the present report. The
report suggests the reader do this for their state(s) of interest.


By employing different spending definitions according to each state’s format (“State
system of record”), comparability across states and aggregation of data is rendered
meaningless. Nevertheless, the report both ranks and aggregates the non-comparable data
in arriving at its conclusions. In Appendix B (p. 39), the report lists several comparability
problems but these problems are ignored thereafter. The deficiencies in the work of the
Department of Educational Reform with respect to funding that were addressed by Baker (Note 12)
have not been corrected. The report proceeds as if to mention a deficiency in data renders
it inoperative.


The primary limitation on the availability of achievement data was whether a state had
reported NAEP averages for both sectors, CS and TPS, separately. The District of Columbia
and 21 states did so and were included in the analysis. NAEP data for grade 8 were
employed in the analyses. However, NAEP tests are also administered at grades 4 and 12,
though these data were ignored claiming that 4th grade NAEP scores would underestimate
effects and 12th grade scores would overestimate them. This rationale is unclear, and by
foregoing any analyses at the other two grades, the report passes on the opportunity to
explore the robustness of their findings.


A cost effectiveness ratio was calculated by dividing the average NAEP score for a sector
and a state by the average cost (per pupil revenue received in thousands of dollars) for a
sector and a state. For example, for the state of Illinois, the average NAEP math score for
TPS was 283 and the average per pupil revenue in thousands of dollars was $6.73 yielding a cost effectiveness ratio of CE = 283/6.73 = 42 NAEP points in math per $1,000 of revenue.


The gist of the report’s argument involved the comparison of cost effectiveness ratios
between sectors. At this point, the logic and the validity of the research run into
difficulties. The validity of the comparison of “effectiveness” (i.e., NAEP averages) depends
on the truth of the counterfactual that if CS pupils had attended TPS, they would have
scored where the TPS pupils scored on the NAEP test. (This is in fact the same as the logic
of any valid, controlled comparative experiment.) In a slightly different context, the
authors have presented data that argues against the truth of this counterfactual.
In attempting to refute claims by Miron (Note 13) and Baker (Note 14) that there are not big funding
inequities between the two sectors, the report presents data on the percentages of Free or
Reduced-Price Lunch (FRL) and Special Education (SP) pupils in each sector for each
state. Baker had argued previously that the differential in revenues between CS and TPS
was due in large part to the fact that the latter serve a greater percentage of pupils who are
poor or who require special services. In their attempt to refute this claim, the authors
assert:


The charter sectors in our study actually tend to enroll a higher percentage of
low-income students than the TPS sectors, regardless of whether one uses free
lunch or FRL as the poverty measure. The special education enrollment gap of
just 3 percentage points is far too small to explain much of the charter school
funding
gap, even if many of the additional special education students in the
TPS sector had the most severe, highest cost, disabilities imaginable. As our
revenue study concluded, a far more obvious explanation for the large charter
school funding gap is that state and local policies and practices deny public
charter schools access to some educational funding streams . . . (p. 11).


In support of the claim that the charter schools in the study enroll more poor pupils and
only slightly fewer special-needs pupils, the report presents Table 1 on page 12, which has
been reproduced below.


Of the 31 states in Table 1, only 22 were involved in the calculation of comparative cost
effectiveness ratios, due to availability of NAEP data for both sectors. Of those 22 used to
calculate cost-effectiveness ratios, 10 states had a higher proportion of poor (FL) pupils
enrolled in CS than in TPS. In 11 states, TPS enrolled a higher percentage of poor pupils
than did CS. (Hawaii had equal percentages in both sectors.) So for the purposes of
calculating the comparative effectiveness of TPS vs CS, it is noteworthy that there are
slightly more states in which TPS has a greater percentage of poor students than CS.

To further examine this bias, consider the relationship between the poverty differential between TPS and CS and the achievement (NAEP) differential. In Figure 1
below is presented the scatter plot relating the poverty differential to the achievement
differential.
(Note: NAEP scores are for math. Hawaii data excluded as an outlier. Correlation = -.72)

The greater the incidence of poverty in TPSs than in CSs, the greater the CS advantage over
the TPSs in achievement, as indicated by the strong negative relationship depicted in
Figure 1 for math. In fact, the correlation coefficient between POVERTY(TPS) –
POVERTY(CS) and NAEP(TPS) – NAEP(CS) is -.72 (excluding Hawaii which is an outlier
in the scatter diagram). Thus, the productivity differentials between CS and TPS which
constitute the numerator of the report’s cost effectiveness measure
are confounded with
differences in poverty between the two sectors, as one would expect. The same analysis for
reading produces almost identical results with a correlation coefficient equal to -.71.
The implication of the strong negative correlation between the poverty differential and the
NAEP score differential is that the effectiveness measure employed in the cost-
effectiveness calculation is really a measure of the differences in poverty at the state level between the charter school and the traditional public school sectors. It is well established
that poverty level is one of the strongest influences on standardized test scores, even
outweighing the influence of schooling itself in most instances. (Note 15)
Data on the percentage of pupils in Special Education similarly show differences that
would bias effectiveness estimates in favor of charter schools. Of 31 states and the District
of Columbia, 16 states show a larger percentage of pupils classified as special needs in TPS
than in CS; 4 show the reverse and 11 states did not have data available by sector. Of the 22
states employed in the report for cost effectiveness analysis, 12 showed higher percentages
of special education pupils in TPS than in CS; 3 showed the reverse; one showed no
difference and 6 have no data available by sector. The authors ignored the implication s of
these data to the validity of their comparisons of NAEP scores and instead attempted to
discredit the special education discrepancy between the sectors as an explanation for the
greater pupil revenues in TPS. (See p. 11 of the report.)


VI. Review of the Validity of the Findings and Conclusions


The very title of the report, The Productivity of Public Charter Schools, invites the
interpretation that charter schools produce greater academic achievement at lesser cost.
However, it is difficult and in many cases impossible to parse revenues into those that are
directed at promoting academic learning and those that serve other purposes (e.g.,
administration, guidance, special services to disabled children and the like). Such parsing
is particularly difficult in the charter school sector where accounting practices are often lax
and transparency of expenditures is sometimes completely lacking. Thus, to argue that a
simple arithmetic ratio of NAEP points and revenues describes a school’s “productivity” is
little more than a weak metaphor.


As noted above, the principal conclusions of the report are as follows:


· In reading, the charter sector delivers an additional 16 NAEP points per $1000
invested, representing a productivity advantage of 41% for charters;
· Percentage differences in cost effectiveness for charters compared to that for TPS
in terms of NAEP math score points per $1000 invested ranges from 7 percent
(Hawaii) to 109 percent (Washington DC);
· Percentage differences in cost effectiveness for charters compared to that for TPS
in terms of NAEP reading score points per $1000 invested ranges from 7 percent
(Hawaii) to 122 percent (Washington DC). (p. 7).

The validity of these conclusions rests in essential ways on the estimates of relative
effectiveness of TPS and CS as reflected in state average NAEP scores in math and reading.
But these estimates have been shown to be seriously biased in favor of CS due to the non -
comparability of the CS and TPS students (disadvantaging the TPS both in wealth and incidence of special needs pupils). Moreover, the conflating of “revenues received” and
“costs” to produce a given effect has reduced the reports exercise to little more than
political arithmetic. (Note 16)


It is astounding that the report offers a proviso concerning its findings while
simultaneously proceeding as though the proviso had never been mentioned:
. . . our cost effectiveness calculation using NAEP scores has important
limitations. Most importantly, it is merely descriptive, not causal, because
charter schools might be reporting higher NAEP scores per $1000 invested than
TPS because of the characteristics of students attracted to the charter school
sector and not because they actually do a better job educating similar students
and at a lower cost. (p. 21).


By any reasonable interpretation of the language employed in the report, the calculations
are put forward as a description of a causal claim. “Effectiveness” implies “effect” and in
both common and academic parlance, effects are the results of causes. The report speaks
out of both sides of its mouth, but only softly does it whisper limits while shouting the
alleged superior productivity of charter schools.


VII. Usefulness of the Report for Guidance of Policy and Practice


The report continues a program of advocacy research that will be cited by supporters of the
charter school movement. It can be expected to be mentioned frequently when arguments
are made that charter schools deserve higher levels of funding. The report will be cited
when attempts are made to refute research on the poor academic performance of charter
schools. Nothing in the report provides any guidance to educators in either sector, CS and
TPS, on how to improve the practice of education. Although the evidence reported
provides no credible foundation for evaluating either the costs or the effectiveness of
charter schools, it can be expected that the report will find frequent use by politicians and
companies managing charter schools as they pursue their reform agenda.

Notes and References

  1. Wolf, P.J., Cheng, A., Batdorff, M., Maloney, L., May, J.F. & Speakman, S.T. (2014). The productivity of public
    charter schools.
    Fayetteville, AR: Department of Education Reform University of Arkansas. Retrieved August
    5, 2014, from
    http://www.uaedreform.org/downloads/2014/07/the-productivity-of-public-charter-schools.pdf.
  2. Witte, J.F., Wolf, P.J., Carlson, D. & Dean, A. (2012). Milwaukee Independent Charter Schools Study: Final
    Report on Four-Year Achievement Gains. Fayetteville, AR: Department of Education Reform University of
    Arkansas. Retrieved August 5, 2014, from
    http://www.uaedreform.org/downloads/2012/02/report-31-milwaukee-independent-charter-schools-study-
    final-report-on-four-year-achievement-gains.pdf.
  3. Batdorff, M., Maloney, L., May, J., Speakman, S., Wolf, P. J., & Cheng, A. (2014). Charter school funding:
    Inequity increases. Fayetteville, AR: School Choice Demonstration Project, University of Arkansas. Retrieved
    August 5, 2014, from
    http://www.uaedreform.org/wp-content/uploads/charter-funding-inequity-expands.pdf.
  4. University of Arkansas (2014). Department of Education Reform in the News (Web page). Fayetteville, AR :
    Author. Retrieved August 7, 2014, from http://www.uaedreform.org/der-in-the-news/.
  5. Levin, H.M. and McEwan, P. (2000). Cost-Effectiveness Analysis: Methods and Applications. (2nd ed.)
    Thousand Oaks, Calif.: SAGE.
  6. Levin, H.M.; Glass, G.V. & Meister, G.R. (1985). Efficiencia de costes de cuatro intervenciones educativas.
    Revista de Educacion, 276, 61-102;
    Levin, H.M., Glass, G.V. & Meister, G.M. (1986). The political arithmetic of cost -effectiveness analysis.
    Kappan, 68 (1), 69-72.
  7. Center for Research on Education Outcomes (CREDO) (2013, June). National Charter School Study. Palo
    Alto: CREDO, Stanford University. Retrieved July 10, 2013, from
    http://credo.stanford.edu/research-reports.html.
  8. Baker, B.D. (2014). Review of “Charter Funding: Inequity Expands.” Boulder, CO: National Education Policy
    Center. Retrieved August 7, 2014, from
    http://nepc.colorado.edu/thinktank/review-charter-funding-inequity.
  9. Baker, B.D. (2014, July 22). UARK study shamelessly (& knowingly) uses bogus measures to make charter
    productivity claims (blog post). School Finance 101. Retrieved August 7, 2014, from
    http://schoolfinance101.wordpress.com/2014/07/22/uark-study-shamelessly-knowingly-uses-bogus-
    measures-to-make-charter-productivity-claims/.
  10. Yeh, S.S. (2013). A re-analysis of the effects of KIPP and the Harlem Promise Academies. Teachers College
    Record.
    Retrieved August 5, 2014, from
    http://www.tcrecord.org/content.asp?contentid=16917;
    Powers, J.M. and Glass, G.V (2014). When statistical significance hides more than it reveals. Teachers College
    Record.
    Retrieved August 5, 2014, from
    http://www.tcrecord.org/Content.asp?ContentID=17591;
    Briggs, D.C. (2009). Review of "Charter Schools in Eight States: Effects on Achievement, Attainment,
    Integration and Competition." Boulder, CO: National Education Policy Center. Retrieved August 5, 2014, from
    http://nepc.colorado.edu/thinktank/review-charter-schools-eight-states;

    Zimmer, R., Gill, B., Booker, K., Lavertu, S., Sass, T.R., & Witte, J. (2009). Charter schools in eight states:
    Effects on achievement, attainment, integration, and competition. Washington DC: RAND Corporation.
    Retrieved August 5, 2014, from http://www.rand.org/pubs/monographs/MG869.html.
  11. Batdorff, M., Maloney, L., May, J., Speakman, S., Wolf, P. J., & Cheng, A. (2014). Charter school funding:
    Inequity increases. School Choice Demonstration Project, University of Arkansas, Fayetteville, AR. Retrieved
    August 5, 2014, from
    http://www.uaedreform.org/wp-content/uploads/charter-funding-inequity-expands.pdf.
  12. Baker, B.D. (2014). Review of “Charter Funding: Inequity Expands.” Boulder, CO: National Education Policy
    Center. Retrieved August 7, 2014, from
    http://nepc.colorado.edu/thinktank/review-charter-funding-inequity.
  13. As quoted in Morones, A. (2014, May 1). Charter schools receive inequitable funding, says report ( Charters
    and Choice Blog). Education Week (online). Retrieved August 7, 2014, from
    http://blogs.edweek.org/edweek/charterschoice/2014/05/charter_schools_receive_inequitable
    _funding_says_report.html.
  14. Baker, B.D. (2014). Review of “Charter Funding: Inequity Expands.” Boulder, CO: National Education Policy
    Center. Retrieved August 7, 2014, from
    http://nepc.colorado.edu/thinktank/review-charter-funding-inequity.
  15. Berliner, D.C. (2013). Effects of inequality and poverty vs. teachers and schooling on America’s youth.
    Teachers College Record, 115(12). Retrieved August 8, 2014, from
    http://www.tcrecord.org/Content.asp?ContentId=16889.
  16. Levin, H.M., Glass, G.V. & Meister, G.M. (1986). The political arithmetic of cost -effectiveness analysis.
    Kappan, 68 (1), 69-72.

    DOCUMENT REVIEWED: The Productivity of Public Charter
    Schools
    AUTHORS: Patrick J. Wolf, Albert Cheng, Meagan
    Batdorff, Larry Maloney, Jay F. May,
    and Sheree T. Speakman.
    PUBLISHER/THINK TANK: Department of Education Reform,
    University of Arkansas

No comments:

Post a Comment

Politics of Teacher Evaluation

1993 Glass, G. V & Martinez, B. A. (1993, June 3). Politics of teacher evaluation. Proceedings of the CREATE Cross-Cutting Eval...