Investigations of the effects of schools (or teachers) on student achievement focus on either (1) individual school effects, such as value-added analyses, or (2) school-type effects, such as comparisons of charter and public schools. Controlling for school composition by including student covariates is critical for valid estimation of either kind of school effect. Student covariates often have different effects between schools than within schools. Econometricians typically attribute such differences to a form of endogeneity, specifically, “Level-2 endogeneity,” or the confounding of student covariates with unobserved school characteristics, whereas education researchers primarily interpret the differences as contextual effects or the effects of collective peer attributes on individual student achievement. This article considers both and makes connections between the econometric and education research literatures. We show that the Hausman and Taylor approach from panel data econometrics can be used for valid estimation of individual school or school-type effects when there is only Level-2 endogeneity but can lead to bias when there are also contextual or peer effects. In contrast, contextual effects are typically estimated by including school means of student covariates in addition to the student-level covariates (equivalent to the Mundlak device), but this leads to biased school comparisons in the presence of Level-2 endogeneity. We interpret the estimates from these two competing estimators in terms of the “Type A” and “Type B” school effects defined by Raudenbush and Willms and show that both estimators are preferable to the common group-mean-centering approach.
AudS.HussarW.JohnsonF.KenaG.RothE.ManningE.…ZhangJ. (2012). The condition of education 2012 (NCES 2012-045). Washington, DC: U.S. Department of Education, National Center for Education Statistics. Retrieved fromhttp://nces.ed.gov/pubsearch
2.
BallouD.SandersW.WrightP. (2004). Controlling for student background variables in value-added assessment of teachers. Journal of Educational and Behavioral Statistics, 29, 37–65.
3.
BaltagiB. H. (2013). The econometrics of panel data (4th ed.). London, England: John Wiley.
4.
BatesD.MaechlerM.BolkerB. (2011). lme4: Linear mixed-effects models using S4 classes [Computer software manual]. Retrieved fromhttp://CRAN.R-project.org/package=lme4
5.
BingenheimerJ. B.RaudenbushS. W. (2004). Statistical and substantive inferences in public health: Issues in the application of multilevel models. Annual Review of Public Health, 25, 53–77.
6.
BoucherV.BramoulléY.DjebbariH.FortinB. (2012). Do peers affect student achievement? Evidence from Canada using group size variation. Journal of Applied Econometrics, 29, 91–109.
7.
BreuschT.WardM. B.NguyenH. T. M.KomposT. (2011). On the fixed-effects vector decomposition. Political Analysis, 19, 123–134.
BuddinR.ZamarroG. (2009). Teacher qualifications and student achievement in urban elementary schools. Journal of Urban Economics, 66, 103–115.
10.
BursteinL. (1980). The analysis of multilevel data in educational research and evaluation. Review of Research in Education, 8, 158–233.
11.
ChamberlainG. (1980). Analysis of covariance with qualitative data. Review of Economic Studies, 47, 225–238.
12.
CroissantY.MilloG. (2008). Panel data econometrics in R: The plm package. Journal of Statistical Software, 27, 1–43.
13.
CronbachL. J. (1976). Research on classrooms and schools: Formulation of questions, design, and analysis(Occasional Paper). Stanford Evaluation Consortium, Stanford, CA.
14.
DuncanC.JonesK.MoonG. (1998). Context, composition, and heterogeneity: Using multilevel models in health research. Social Science & Medicine, 46, 97–117.
15.
EbbesP.BöckenholtU.WedelM. (2004). Regressor and random-effects dependencies in multilevel models. Statistica Neerlandica, 58, 161–178.
16.
FullerW. A.BatteseG. E. (1973). Transformations for estimation of linear models with nested-error structure. Journal of the American Statistical Association, 68, 626–632.
17.
GoldhaberD. D.BrewerD. J. (1997). Why don’t schools and teachers seem to matter? Assessing the impact of unobservables on educational productivity. The Journal of Human Resources, 32, 505–523.
18.
GongG.SamaniegoF. J. (1981). Pseudo maximum likelihood estimation: Theory and applications. The Annals of Statistics, 9, 861–869.
19.
GourierouxC.MonfortA. (1995). Statistics and econometric models (Vol. 2). Cambridge, England: Cambridge University Press.
20.
GreeneW. (2011). Fixed-effects vector decomposition: A magical solution to the problem of time-invariant variables in fixed effects models?Political Analysis, 19, 135–146.
21.
HanchaneS.MostafaT. (2012). Solving endogeneity problems in multilevel estimation: An example using education production functions. Journal of Applied Statistics, 39, 1101–1114.
22.
HanushekE. A.KainJ. F.MarkmanJ. M.RivkinS. G. (2003). Does peer ability affect student achievement?Journal of Applied Econometrics, 18, 527–544.
23.
HauserR. M. (1970). Context and consex: A cautionary tale. American Journal of Sociology, 75, 645–664.
24.
HausmanJ. A.TaylorW. E. (1981). Panel data and unobservable individual effects. Econometrica, 49, 1377–1398.
25.
KennedyE.MandevilleG. (2000). Some methodological issues in school effectiveness research. In TeddieC.ReynoldsD. (Eds.), International handbook of school effectiveness research (pp. 189–205). New York, NY: Routledge.
26.
KimJ. S.FreesE. W. (2006). Omitted variables in multilevel models. Psychometrika, 71, 659–690.
27.
KimJ. S.FreesE. W. (2007). Multilevel modeling with correlated effects. Psychometrika, 72, 505–533.
28.
KimJ. S.SwobodaC. M. (2011). Handling omitted variable bias in multilevel models: Model specification tests and robust estimation. In HoxJ. J.RobertsJ. K. (Eds.), Handbook of advanced multilevel analysis (p. 197–217). New York, NY: Routledge.
29.
KoedelC.BettsJ. (2010). Value added to what? How a ceiling in the testing instrument influences value-added estimation. Education Finance and Policy, 5, 54–81.
30.
KreftI. G. G.de LeeuwJ.AikenL. S. (1995). The effect of different forms of centering in hierarchical linear models. Multivariate Behavioral Research, 30, 1–21.
31.
LehmannE. L. (1990). Model specification: The views of Fisher and Neyman, and later developments. Statistical Science, 5, 160–168.
32.
LütkeO.MarshH. W.RobitzschA.TrautweinU.AsparouhovT.MuthénB. (2008). The multilevel latent covariate model: A new, more reliable approach to group-level effects in contextual studies. Psychological Methods, 13, 203–229.
33.
MaddalaG. S. (1971). The use of variance components models in pooling cross-section and time series data. Econometrica, 39, 341–358.
34.
ManskiC. F. (1993). Identification of endogenous social effects: The reflection problem. Review of Economic Studies, 60, 531–542.
35.
MarshH. W.SeatonM.TrautweinU.LütkeO.HauK. T.O’MaraA. J.CravenR. G. (2008). The big-fish-little-pond-effect stands up to critical scrutiny: Implications for theory, methodology, and future research. Educational Psychology Review, 20, 319–350.
36.
McCaffreyD. F.LockwoodJ.KoretzD.LouisT. A.HamiltonL. (2004). Models for value-added modeling of teacher effects. Journal of Educational and Behavioral Statistics, 29, 67–101.
37.
MundlakY. (1978). On the pooling of time series and cross-sectional data. Econometrica, 46, 69–86.
38.
National Research Council & National Academy of Education. (2010). Getting value out of value-added: Report of a workshop(Committee on Value-Added Methodology for Instructional Improvement, Program Evaluation, and Educational Accountability, H. Braun, N. Chudowsky, & J. Koenig, Eds.). Washington, DC: National Academic Press.
39.
NyeB.KonstantopoulosS.HedgesL. V. (2004). How large are teacher effects?Educational Evaluation and Policy Analysis, 26, 237–257.
40.
PacagnellaO. (2006). Centering or not centering in multilevel models?Evaluation Review, 30, 66–85.
41.
PalardyG. J.RumbergerR. (2008). Differential school effects among low, middle, and high social class composition schools: A multiple group, multilevel latent growth curve analysis. School Effectiveness and School Improvement, 19, 21–49.
PlümperT.TroegerV. E. (2007). Efficient estimation of time-invariant and rarely changing variables in finite sample panel analyses with unit fixed effects. Political Analysis, 15, 124–139.
44.
R Development Core Team. (2012). R: A Language and Environment for Statistical Computing [computer software]. Vienna, Austria. Retrieved fromhttp://www.R-project.org/
45.
Rabe-HeskethS.SkrondalA. (2012). Multilevel and longitudinal modeling using Stata (3rd ed., Vol. 1: Continuous responses). College Station, TX: Stata Press.
46.
RaudenbushS. W. (1989). Centering predictors in multilevel analysis: Choices and consequences. Multilevel Modelling Newsletter , 1, 10–12.
47.
RaudenbushS. W.BrykA. S. (2002). Hierarchical linear models: Applications and data analysis methods (2nd ed.). Thousand Oaks, CA: Sage.
48.
RaudenbushS. W.WillmsJ. D. (1995). The estimation of school effects. Journal of Educational and Behavioral Statistics, 20, 307–335.
49.
RivkinS. G.HanushekE. A.KainJ. F. (2005). Teachers, schools, and academic achievement. Econometrica, 73, 417–458.
50.
RockoffJ. E.JacobB. A.KaneJ.StaigerD. O. (2011). Can you recognize an effective teacher when you recruit one?Association for Education Finance and Policy, 6, 43–74.
51.
SacerdoteB. (2011). Peer effects in education: How might they work, how big are they and how much do we know thus far? In HanushekE. A.MachinS.WoessmannL. (Eds.), Handbook of the economics of education (Vol. 3, pp. 250–277). Amsterdam, The Netherlands: North-Holland.
52.
ShinY.RaudenbushS. W. (2010). A latent cluster-mean approach to the contextual effects model with missing data. Journal of Educational and Behavioral Statistics, 35, 26–53.
53.
SkrondalA.Rabe-HeskethS. (2009). Prediction in multilevel generalized linear models. Journal of the Royal Statistical Society, Series A, 172, 659–687.
54.
SpanosA. (2006). Where do statistical models come from? Revisiting the problem of specification. IMS Lecture Notes—Monograph Series 2nd Lehmann Symposium—Optimality, 49, 98–119.
55.
StataCorp. (2013a). Stata 13 Longitudinal-data/panel-data Reference Manual [Computer software manual]. College Station, TX: Stata Press.
56.
StataCorp. (2013b). Stata statistical software: Release 13 [Computer software]. College Station, TX: Stata Press.
57.
SteedleJ. T. (2012). Selecting value-added models for postsecondary institutional assessment. Assessment & Evaluation in Higher Education, 37, 637–652.
58.
U.S. Department of Education. (2010). Overview information: Race to the top fund; notice inviting applications for new awards for fiscal year (FY) 2010 (Federal Register Vol. 75, No. 71).
59.
Von SeckerC. E.LissitzR. W. (1999). Estimating the impact of instructional practices on student achievement in science. Journal of Research in Science Teaching, 36, 1110–1126.
60.
WayneA. J.YoungsP. (2003). Teacher characteristics and student achievement gains: A review. Review of Educational Research, 73, 89–122.
61.
WileyD. E. (1975). Another hour, another day: Quantity for schooling, a potent path for policy. In HauserR. M.SewellW. H.AlwinD. (Eds.), Schooling and achievement in American society (pp. 225–265). New York, NY: Academic Press.
62.
WillmsJ. D. (1986). Social class segregation and its relationship to pupil’s examination results in Scotland. American Sociological Review, 51, 224–241.
63.
WooldridgeJ. M. (2010). Econometric analysis of cross-section and panel data. Cambridge, MA: The MIT Press.
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
