The effect of school resources on pupil attainment: A multilevel simultaneous equation modelling approach

University of Bristol, Bristol, England, United Kingdom
Journal of the Royal Statistical Society Series A (Statistics in Society) (Impact Factor: 1.64). 07/2007; 170(3):801-824. DOI: 10.1111/j.1467-985X.2007.00476.x
Source: RePEc


Improving educational achievement in UK schools is a priority, and of particular concern is the low achievement of specific groups, such as those from lower socio-economic backgrounds. An obvious question is whether we should be improving the outcomes of these pupils by spending more on their education. The literature on the effect of educational spending on the achievement of pupils has some methodological difficulties, in particular the endogeneity of school resource levels, and the intraschool correlations in pupils' responses. We adopt a multi-level simultaneous equation modelling approach to assess the effect of school resources on pupil attainment at age 14 years. The paper is the first to apply a simultaneous equation model to estimate the effect of school resources on pupils' achievement, using the newly available national pupil database and pupil level annual school census. Copyright 2007 Royal Statistical Society.

Download full-text


Available from: Anna Vignoles,
  • Source
    • "The effect of is separated into two associations, one at each level, which are interesting, interpretable, and relevant to the researcher (Enders and Tofighi, 2007 p130). Second, by 17 If covariates remain correlated with residuals (for example as a result of simultaneity, or other omitted variables), they can potentially be dealt with within this RE framework through other means, such as instrumental variable methods (Heckman and Vytlacil, 1998) using simultaneous equations (Steele et al., 2007), assuming of course that appropriate instruments can be found. Whilst all heterogeneity bias of lowerlevel variables has been dealt with, a variant of the Hausman-Taylor IV estimator (Greene, 2012 p434, Hausman and Taylor, 1981) can be used to deal with correlated time-invariant variables (Chatelain and Ralf, 2010). "
    [Show abstract] [Hide abstract]
    ABSTRACT: This article challenges Fixed Effects (FE) modeling as the ‘default’ for time-series-cross-sectional and panel data. Understanding different within and between effects is crucial when choosing modeling strategies. The downside of Random Effects (RE) modeling— correlated lower-level covariates and higher-level residuals—is omitted-variable bias, solvable with Mundlak’s (1978a) formulation. Consequently, RE can provide everything that FE promises and more, as confirmed by Monte-Carlo simulations, which additionally show problems with Pluemper and Troeger’s FE Vector Decomposition method when data are unbalanced. As well as incorporating time-invariant variables, RE models are readily extendable, with random coefficients, cross-level interactions and complex variance functions. In the appendix, an empirical example shows that disregarding these extensions can produce misleading results. We argue not simply for technical solutions to endogeneity, but for the substantive importance of context/heterogeneity, modeled using RE. The implications extend beyond political science to all multilevel datasets. However, omitted variables could still bias estimated higher-level variable effects; as with any model, care is required in interpretation.
    01/2015; 3(1):133-153. DOI:10.1017/psrm.2014.7
  • Source
    • "Their results were slightly smaller but broadly comparable to those found by Krueger. Two UK studies have reported associations between higher attainment in mathematics and lower pupil–teacher ratios for secondary school students (Bradley et al., 2001; Steele et al., 2007). Krueger and Whitmore (2001) found that estimates of long-term effects of class sizes in early elementary school were considerably smaller than short-term estimates. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Education policy in England has explicitly aimed to remodel state schools in the image of independent, private, schools. However, the body of research evidence on the operation of private schools is very small. Critics have frequently argued that, in contrast to state schools, private schools use resources efficiently because their autonomy gives them freedom to make efficient choices in response to market pressures. We investigated the resource allocation decisions of income private schools using data from their financial accounts. We report the associations of private school spending and performance and teacher and non-teacher numbers, teacher salaries and regional earnings. Schools with higher fees had more teachers and non-teachers per pupil, but similar average staff wages to schools with lower income. Yet we found little evidence that achievement was associated with resource use in English private schools. That is, we found no evidence to support the view that financial autonomy enabled flexibility in using resources to promote higher achievement.
    British Educational Research Journal 06/2014; 40(3). DOI:10.1002/berj.3107 · 1.50 Impact Factor
  • Source
    • "In doing so we take account of several econometric issues. Given that we focus on private unaided schools, we depart from the issue of allocation of school funds to government schools by local education authority (Steele et al, 2007). As such we try minimising estimation bias arising from two possible sources of omitted variables – one arising from parental potential demand for better performing schools and the other relating to schools' (by type) efforts to perform better and thereby investing/recruiting teachers/students of highest ability. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Abstract We study school choice and school efficiency in terms of secondary school completion test scores by utilizing a unique database from Nepal. There are two novel features of our model: firstly we allow for heterogeneity among private schools, by distingusihing socially motivated trust-run schools from profit-motivated company-run schools, and secondly, we include school's expenditure as a determinant of its efficiency per unit of cost. We find that when expenditure is not included, the trust-run school comes on top, slightly but distinctly, ahead of the profit-motivated school. But if expenditure is included, the trust-run school's position becomes sensitive to the level of expenditure, as it is the only school to exhibit sensitivity between expenditure and test score. In the urban area, the public school is always at the bottom, and between the two types of the private school the trust-run school ranks first (second) at high (low) levels of expenditure. However, in the rural area it is a three way race, with the trust school coming on top again at high expenditure, but falling to bottom at low levels of expenditure. This picture is fairly robust to considerations of subject fixed effets and to inclusion or exclusion of private tuition. We show both theoretically and empirically that socially motivated schools can be efficient and outperform profit-motivated schools.
    SSRN Electronic Journal 05/2014; DOI:10.2139/ssrn.2395944
Show more