The effectiveness of tutoring interventions in mathematics for disadvantaged students

This literature review was originally published 04 June 2015.

Image: The effectiveness of tutoring interventions in mathematics for disadvantaged students



Disadvantaged students from low socioeconomic status backgrounds often have poorer educational outcomes compared to their more advantaged counterparts. In Australia, this is particularly evident in mathematics achievement at the secondary school level. One commonly used approach to address disparities in student performance is through the implementation of remedial tutoring programs. This review provides a summary of the relevant empirical literature on the effectiveness of tutoring interventions in maths for disadvantaged students.

Overall, the evidence reviewed strongly suggests that tutoring in maths can have significant positive effects on performance among disadvantaged students. While the magnitude of tutoring effects vary markedly across studies (effect sizes range from approximately .04 to 1.17), the majority of the observed effects represent marked improvements in the performance of tutored students. The best-practice elements of tutoring that are often associated with the most significant performance gains include: designing quality instructional materials that closely reflect classroom content; having close co-ordination with classroom teachers; providing extensive initial and on-going training for tutors; having well-structured programs; providing careful monitoring of student performance; providing regular feedback and reinforcement of progress; and, scheduling tutoring sessions on a frequent and regular basis.

The effectiveness of tutoring in mathematics: An overview of the literature

As noted above, tutoring is not a uniform intervention with a single prescribed approach. Tutoring interventions can vary across a number of dimensions and this heterogeneity is reflected across the empirical literature. For example, the content and delivery of tutoring programs varies depending on the age and skill level of the tutee, the age or type of tutor employed (same-age or older peers, community volunteers, paid private tutors, teachers), the format of tutoring (one-to-one, small group), the frequency and duration of tutoring sessions, the instructional approach used, and perhaps most importantly, the subject, content or skill-set being tutored (Center for Prevention Research and Development, 2009). Despite this variability, tutoring programs typically share the common core goal of providing students with individualised instruction focusing on a particular subject area or goal. Gaustad (1992) has suggested that it is this individualised instruction that is responsive to students’ needs and the emotional and motivational benefits of this type of interaction that account for much of the improvement associated with tutoring (pp. 7-8). In addition, tutoring allows students to derive academic benefits from spending more time-on-task as well as having more opportunities to receive performance feedback and individualised monitoring (Bowman-Perrott et al., 2013; Ginsburg-Block & Fantuzzo, 1998; Greenwood, Carta & Hall, 1988; Topping, 2005).

While the inherent variability in how tutoring programs are implemented can present some challenges to distilling an overall picture of tutoring effectiveness (Center for Prevention Research and Development, 2009, p. 2), a vast amount of research has been conducted worldwide examining the effects of tutoring on student achievement. An often-cited early meta-analysis of tutoring conducted by Cohen, Kulik and Kulik (1982) synthesized the results from 65 independent evaluations of school-based tutoring programs. This analysis included studies with tutees of varying ages (class levels one- three, four-six and seven-nine) and focused predominately on outcomes in reading and maths. Results showed that tutoring programs had a significant positive effect on student academic performance with an overall effect size of 0.401. This equates to an increase in performance among tutored students of two-fifths of a standard deviation unit compared to non-tutored students (Cohen et al., 1982). While the evaluations analysed included a greater focus on interventions in reading compared to maths (30 studies versus 18 studies), when effects were examined separately by subject, results showed markedly larger impacts of maths tutoring on student achievement (0.60) than interventions focussing on reading instruction (0.29). There were no significant differences for tutees of different ages. The efficacy of maths tutoring specifically for disadvantaged students was further supported by a 1989 narrative review which concluded that peer tutoring was an effective method of improving maths outcomes for low achieving students, including those from socially disadvantaged backgrounds, and for students with mild disabilities (Britz, Dixon & McLaughlin, 1989). However, this study failed to report measures of effect size, making it difficult to estimate the magnitude of the reported effects.

A later meta-analysis examining the impact of school-based peer tutoring programs on the maths achievement of low-achieving students from kindergarten through year 12 revealed a moderately high mean effect size of 0.62 (Baker, Gersten & Lee, 2002). However, a literature review conducted by Robinson, Schofield and Steers-Wentzell (2005) focusing on the impact of peer and cross-age tutoring on maths achievement among minority students showed that while the effects of tutoring are mostly positive, the magnitude of the observed effects varies markedly from study to study (effect sizes from 0.30 to 1.17). The authors note that much of the variance across studies is likely associated with the types of tutors employed, the age of tutees and tutors, whether tutors received training, the length of the tutoring program, and the measures used to assess achievement (Robinson et al., 2005, p. 334). More recently, Bowman-Perrott et al. (2013) conducted a meta-analytic review of peer tutoring interventions for students in both elementary and secondary school, focussing largely on studies of poor performing students and those with disabilities, or at-risk for disabilities. The results from this analysis revealed a significant positive overall effect of peer tutoring on student performance (0.75), with slightly larger effects observed for secondary (0.74) compared to elementary students (0.69). When outcomes were analysed by content area, results showed relatively large effects for studies focusing on maths (0.86), and for those examining vocabulary based interventions (0.92). However, the authors note that these subject level effects should be interpreted with caution as the content area analysis was based on a relatively small number of studies in each discipline (Bowman-Perrott et al., 2013, p. 51).


The evidence summarised in this review strongly suggests that tutoring in maths for disadvantaged students can have significant positive effects on academic performance. While the types of tutoring programs implemented vary across the literature, it is clear that significant gains in academic performance can be made among the most disadvantaged or low achieving secondary school students. However, this evidence is largely drawn from research conducted overseas and although tutoring is a commonly employed method of remediating student performance in Australia, there is currently a dearth of rigorous, independent research examining the effectiveness of tutoring programs that specifically target maths achievement among disadvantaged secondary school students. This underscores the need for future research to examine whether interventions shown to be effective elsewhere can produce similar positive gains for disadvantaged students in an Australian context.

1 Effect size is a measure of the difference in performance of two groups. In the context of the current review, effect size refers to the difference in performance between tutored and non-tutored students. As a guide for interpretation, Cohen (1988) suggests that an effect size of 0.20 reflects a ‘small’ effect, 0.50 reflects a ‘medium’ effect and 0.80 reflects a ‘large’ effect.

Return to top of page Back to top