Evaluating the impact of individual tutoring on GCSE attainment
Judith Ireson and Katie Rushforth
Institute of Education, University of London
Paper presented at the British Educational Research Association Annual Conference, University of Glamorgan, 14-17 September 2005
Abstract
Individual tutoring is generally thought to provide an ideal environment in which students can receive help tailored to their needs. Many middle class families employ tutors to help children with their schoolwork and these children are seen as gaining advantage in the education system. There is, however, very little research on the effectiveness of private tuition. Evidence on the effectiveness of individual and small group tutoring in school shows mixed results. Some tutoring programmes show strong positive effects on attainment, while others show little if any improvement. Similarly international research on the effects of private tuition presents a mixed picture. To date there has been very little research on the effectiveness of private tuition in the UK. This paper offers the first analysis of the impact of private tuition on GCSE attainment in England. A sample of 302, year 11 students in 7 mixed secondary schools completed questionnaires providing information on the nature and extent of tutoring received. Students also supplied information on socioeconomic status, family education, gender and ethnicity. Key Stage 3 and GCSE results of 282 of these students were also obtained. During years 10 and 11, 48 students in the sample received private tuition in mathematics and 20 in English, while a few were tutored in science and other subjects. The most common pattern was for students to have tutors for one or two terms. The GCSE grades of students who did and did not receive tutoring during years 10 and 11were compared, taking account of Key Stage 3 test results. On average, students with tutors for mathematics achieved just under half a grade higher than students who did not have tutors, however girls did not appear to benefit as much as boys. The implications of these findings are discussed in relation to parents’ and students’ evaluations of private tuition, government proposals to increase personalised learning in schools and the potential impact of private tuition on schools’ performance in league tables.
Introduction
Many families employ private tutors to help children with their schoolwork, thus participating in a ‘shadow education’ system that supplements normal schooling. (Stevenson & Baker, 1992). Private tuition takes a variety of forms in different countries (Bray & Kwok, 2003) and includes additional teaching in large classes. In countries such as Japan and Greece students go from school to private establishments where there may be taught in large groups. In Mauritius students remain in school and pay for additional lessons often from the class teacher who taught them earlier in the day (Foondun, 2002). In contrast, tutoring in England is generally one-to-one and often takes place in the home of the tutor or the student. These differences are likely to influence the effectiveness of tutoring.
Evidence on the effectiveness of private tutoring is mixed (Ireson, 2004). An international comparison of private tuition in mathematics reported no effects of private tuition on 8th grade pupils’ attainment (Baker, Akiba, LeTendre & Wiseman, 2001). This research used survey data collected for the Third International Mathematics and Science Survey, which provided limited information on the tutoring received. It may be that uncontrolled variables, such as whether tuition was individual or in large classes may mediate or interact with the effects of tutoring. In contrast, an experimental study demonstrated that students who received small group tuition after school in private institutions in Germany made greater gains in mathematics, English, French and Latin than pupils of similar attainment who did not receive tutoring (Mischo & Haag, 2002). The experimental group received tutoring in small groups of 4 for 90 minutes a day, four days a week from student teachers or unemployed teachers. This carefully designed study shows that private tuition in small groups can be very effective.
Individual and small group tutoring in school may also be very effective(Bloom, 1984; Cohen, Kulik & Kulik, 1982; Ellson, 1976; Rosenshine & Furst, 1969; Wasik & Slavin 1993). Bloom (1984) reported that the average attainment of students who received tutoring in place of normal teaching was about two standard deviations above the average of students in control groups taught with conventional methods. This means that the average tutored student outperformed 98 percent of the students in the control class. Moreover, variability between students was greatly reduced with tutoring. Bloom's claim about the effectiveness of tutoring is based on research undertaken by two doctoral students, only one of whom subsequently published their findings in an academic journal. Although this research was well designed, the evaluation of the tutoring programme, which was developed by the author, was on a small scale (Anania, 1983). Before generalising the findings, the research should be replicated in other contexts and with other samples of students. Yet Bloom's claim that tutoring offers gains of two standard deviations received considerable attention. In a review of the effects of alterable variables on student achievement, Walberg (1984) identified tutorial instruction as having the greatest influence on attainment. He compared Bloom's findings of effect size of 2.0 with other variables such as mastery learning, student time on task and improved study skills, all of which have reported effect sizes of 1.0. A more recent meta-analysis of individualisation indicates only a small effect (Hattie, 1999).
In schools, individual tutoring is rarely provided except in the teaching of reading, where children receive individual help if they make slow progress in learning to read. There is now a substantial literature on the quality of tutoring and its effectiveness in raising children's reading attainment, including several reviews and meta-analyses (Elbaum, Vaughn, Hughes & Moody, 2000; Shanahan, 1998; Wasik & Slavin, 1993). Elbaum et al (2000) identified 29 studies that compared elementary students who received one-to-one tutoring with students whose reading attainment was comparable but who did not receive individual tutoring. The majority of students were identified as at risk of reading failure and some had learning disabilities. The 29 studies contained data on 42 samples of students who received one-to-one tutoring from trained teachers, college students or community volunteers. On average, students who received one-to-one tutoring performed at a level two-thirds of a standard deviation higher that the average level of the comparison group. This translates into a rise from the 50th to the 60th percentile on a standardised measure, which would enable students to keep up with classroom instruction. Although substantial, this is lower than the effect sizes reported by Bloom (1984) and Walberg (1984) and suggests that students with reading difficulties make smaller gains.
The effectiveness of tutoring also relates to the content and structure of the programme followed. In the teaching of reading, tutorial programmes are based on different theoretical approaches and differ substantially in content, some placing much greater emphasis on phonics while others emphasise reading books from an early stage. Programmes that devote more time to reading and writing extended text are more effective than those with a narrower focus on letters and words (Pinnell, Lyons, DeFord, Bryk & Seltzer, 1994; Wasik & Slavin, 1993). It is important to note, however, that tutoring is not equally effective for all students and that some programmes such as Reading Recovery do not meet the needs of children who experience the greatest difficulties with phonological aspects of reading (Elbaum, Vaughn, Hughes & Moody, 2000). The content of a programme needs to be tailored to the individual child's needs.
To date there has been no research on the effectiveness of coaching for GCSE examinations. This paper examines the effects of private tuition on students’ attainment in GCSE English and mathematics. It forms part of a larger project mapping the nature and extent of private tuition, exploring reasons for participating in the shadow education system and evaluating its effectiveness.
Method
Design of the research
Student questionnaires provided information on the nature and extent of private tuition received, gender, ethnicity and parental education and occupation. Information on pupils’ level of attainment was collected from DfES.
Samples
The analysis of effects on attainment is based on a sample of 296 year 11 students in 7 schools who completed questionnaires and sat GCSEs during 2003. Of the 294 students who identified their sex 48% were male and 52% female. The majority of these students (90%) identified themselves as white or white European and 9% indicated that their first language was not English.
Results
The analysis was undertaken in two stages. The first examined the effect of having a private tutor in any subject on overall achievement. The second examined the more specific effects of tutoring in maths on maths GCSE and tutoring in English on English GCSE.
For the first analysis the outcome measure was the students’ average GCSE score in English, maths and science. The main explanatory variable of interest was whether the students had received any private tuition during Key Stage 4 (years 10 and 11). In all analyses, the KS3 test results were adjusted for, so that the analyses examined the effect of tuition upon the progress in attainment since the beginning of the Key Stage. In addition, all results were adjusted for the school that the pupils went to, as it may be that pupils in some schools make more progress than pupils in other schools simply due to the school effects.
Other pupil characteristics included in the analyses were gender and eligibility for free school meals. Parental occupation and education were also included. Ethnic group was not included due to small numbers. To check for differential effects relating to level of ability the students were split into three groups based on their prior attainment. The 25% of students with the lowest KS3 scores were put into one group, the 25% with the highest scores were placed into a second category, and the middle 50% were put into a third group.
Initially the individual effect of each variable upon the outcome was examined and also the interaction between each variable and private tuition. Subsequently a multi-variate analysis was performed to examine the joint effect of each variable upon the outcome variable. A backwards selection procedure was used to retain only the statistically significant variables. This involves removing non-significant variables one at a time, until only the significant variables remain in the model. All analysis was performed using linear regression. Multilevel models were not used for the analysis due to the small number of schools providing data, although the analysis adjusted for the school attended, as described above.
The first step in the analysis was to examine the effects of each explanatory variable individually. Gender and first language not English were found to significantly influence the progress made from KS3 to KS4. Females made more progress than males whilst pupils with a first language other than English performed better than those whose first language was English. Pupils in a higher socio-economic group did slightly better, although this was not quite statistically significant. There was no effect of SEN status, eligibility for free school meals, father’s job or private tuition before Year 10 upon the progress made during KS4.
The second stage in the analysis was to examine the joint effect of the explanatory variables together in a multivariate analysis. The final regression model is summarised in Table 1. The results indicated that after adjusting for the effects of the other explanatory variables, there was still a significant effect of both gender and first language not English upon KS4 attainment, but no effect of socio-economic status.
Variable / Group / Effect (95% CI) / P-valueGender / Male / 0
Female / 0.23 (0.08, 0.38) / 0.002
First language / English / 0
Not Eng. / 0.47 (0.03, 0.90) / 0.04
Private tuition / No / 0
Yes / 0.22 (0.05, 0.38) / 0.009
Table 1. Estimates for effects on average English, mathematics and science grades.
After adjusting for gender and first language, there was a significant overall effect of tuition, with pupils receiving tuition having scores that were, on average, 0.2 units higher than pupils not having tuition. This shows that private tuition had a small positive effect on average GCSE attainment.
The second set of analyses examined the effect of private tuition in mathematics on GCSE maths grades. Forty-eight students (16% of the total responses given) had received tutoring in mathematics during years 10 and 11. Attainment data was available for 42 students, 22 males and 20 females. Most of these students (27) had achieved intermediate levels 5 and 6 at Key Stage 3, 9 were at level 7, 5 at level 4 and 1 at level 3. Although the profile was similar for males and females a few more females had low KS3 grades 3 and 4, whereas more a few more males had KS3 level 7 (Table 2).
Key stage 3 maths / Males / Females / TotalLevel 3 / 1 / 0 / 1
Level 4 / 1 / 4 / 5
Level 5 / 5 / 5 / 10
Level 6 / 9 / 8 / 17
Level 7 / 6 / 3 / 9
22 / 20 / 42
Table 2. Key Stage 3 mathematics levels of students with tutors, by gender
Table 3 displays the GCSE grades achieved by these students. The one male student who achieved KS3 level 3 did not have a recorded GCSE grade. This table shows that the distribution of GCSE grades is lower for females than for males.
Key Stage 3 mathslevel / GCSE mathematics grade
F / E / D / C / B / A / A*
4 / M / 1
F / 1 / 2 / 1
5 / M / 3 / 2
F / 1 / 1 / 3
6 / M / 3 / 5 / 1
F / 1 / 3 / 4
7 / M / 4 / 1 / 1
F / 3
Total / M / 3 / 6 / 9 / 2 / 1
F / 1 / 3 / 3 / 6 / 7
Table 3. GCSE results for students with maths tutors, by Key Stage 3 level and gender
The effects of private tuition on GCSE results were then examined using regression analysis, which adjusted for Key Stage 3 attainment in the relevant subject. Independent effects of each variable were examined first and showed that there was a significant effect of private tuition on GCSE mathematics grades. In addition there were significant interactions between private tuition and both gender and father’s occupation, suggesting that the effect of private tuition was greater for males than females and for students whose fathers were in non-professional occupations. When the joint effects of these variables were examined together in a multi-variate analysis, an interaction with gender remained, however there was no effect of father’s occupation. Estimates for the final model are displayed in Table 4 and indicate that there was a significant effect of private tuition for males, who gained on average 0.7 of a GCSE grade, while the grades of females were not significantly affected by having private tuition. As the results indicated that the effect of tutoring was different for the two genders, some caution must be exercised when combining these groups, however if all students are considered together, pupils receiving private tuition had maths scores that were on average 0.4 of a grade higher than students who did not have tuition.
It is also of interest to note that the final models indicated that there was a significant difference between schools and that the adjusted R-squared value was 0.77. This means that the model explains a high proportion of the variance in GCSE grades.
Variable / Subgroup / Group / Effect (95% CI) / P-valueGender / No tuition / Male / 0
Female / 0.39 (0.17, 0.61) / 0.001
Tuition / Male / 0
Female / -0.31 (-0.84, 0.21) / 0.24
Private tuition / Male / No / 0
Yes / 0.70 (0.31, 1.09) / 0.001
Female / No / 0
Yes / 0.00 (-0.42, 0.41) / 0.99
Private tuition / All pupils / No / 0
Combined / Yes / 0.39 (0.08, 0.66) / 0.01
Table 4. Estimates for effects on mathematics GCSE grades.
A similar analysis was undertaken for English, however only 20 students (7%) had received tuition in English during years 10 and 11. The initial analysis of individual effects of variables indicated that only gender had a significant effect on GCSE attainment. There was also some evidence of a tendency for students eligible for free school meals to achieve lower grades and for students whose first language was not English to achieve higher grades. The joint effect of the explanatory variables was examined in a multivariate analysis using backwards selection and the final model is given in Table 5.
Variable / Group / Effect (95% CI) / P-valuePrivate tuition / No / 0
Yes / -0.06 (-0.47, 0.34) / 0.77
Gender / Male / 0
Female / 0.27 (0.07, 0.46) / 0.008
First language / English / 0
Not English / 0.31 (-0.05, 0.67) / 0.09
SES / - / 0.04 (0.00, 0.08) / 0.09
Table 5. Estimates for effects on English GCSE grades.
The results show that females performed significantly better than males. Also, there was some indication that students whose first language was not English and students not taking free school meals made more progress. After adjusting for the effects of these variables, there was no difference between pupils receiving and not receiving private tuition.
Taken together these results indicate that students who have extra help in mathematics during years 10 and 11 achieve higher grades, however this boost appears to be for males only, not for females. There was no evidence of an effect of private tuition on GCSE English grades.
Discussion and conclusions
The aim of this paper was to explore the effects of private tuition on attainment. We analysed the effects of tutoring during years 10 and 11 on GCSE attainment, controlling statistically for students’ Key Stage 3 levels. Our findings suggest that private tuition does have a positive impact on attainment, particularly in mathematics where students gained, on average, 0.4 of a grade higher if they had a tutor. Similarly there was a general, but smaller improvement in grades resulting from private tuition in any subject.
In mathematics, boys benefited more than girls from private tuition, the average gain for boys being 0.7 of a grade while girls showed no significant improvement. This finding is unexpected and we can only speculate on possible reasons for the difference. A first possibility is that females who had tutors would have made good progress anyway but lacked confidence in their ability. Evidence from several large-scale studies of students’ self-concepts indicates that during adolescence males give higher ratings than girls of their achievement, interest and enjoyment in mathematics and science (Ireson, Hallam and Plewis, 2001; Ireson & Hallam, in preparation; Marsh). Responses to the parent questionnaire also suggested that an important reason for employing a tutor was to improve their child’s confidence (Ireson & Rushforth, 2005). More direct evidence was available from students’ responses to an item on the student questionnaire where they were asked to indicate the benefits of private tuition, one of which was ‘Having a tutor makes me feel more confident’. Very few of the students who had maths tutors during Key Stage 4 selected this option and there was no difference between males and females. The main perceived benefit, indicated by 40% of these students, was ‘I get one-to-one help’. Students also indicated that their tutors’ explanation of work was beneficial, something that is possible in the one-to-one situation. Problems with understanding a subject may affect students’ interest, enjoyment and their perceptions of achievement. As yet we have no evidence that there are gender differences in this respect.