Translated by Mirjam Neelen & Paul A. Kirschner
This article was originally published in Dutch by Tim Surma, Paul Kirschner, and Daniel Muijs on the ExCEL website on 30 August 2020.
ExCEL is the Centre of Expertise for Effective Learning of Thomas More Hogeschool België, which helps to build good quality education. ExCEL conducts research in schools into what works in terms of didactics, learning strategies, curriculum and teaching materials, translates findings from the best international scientific research into feasible and concrete applications for the classroom. Last but not least, it offers tailor-made professionalization programs for schools, based on what (evidence on) works.
“In the corridors of the secondary school where I work as a teacher, it was never completely quiet during breaks. During the morning break, I saw colleagues explaining the subject matter to small groups of students. Sometimes this even happened just five minutes after class to explain one type of exercise a bit more. Often, a pupil who was absent due to illness was given remedial explanations by the specialist teacher after the seventh hour. This type of ‘additional instruction’, rarely involved a full lesson. After all, instruction in small groups or – even better – individual instruction is simply more effective and efficient than instruction in larger groups. What happens in my school during breaks is what is called – in scientific terms – tutoring.”
Tutoring: an overall positive effect but …
Tutoring is an example of an educational intervention which offers extra instruction, sometimes one-on-one, sometimes in a small group. It doesn’t replace lessons in a classroom; it’s organised as an addition to ‘regular teaching time’. In other words, it complements regular instruction, rather than replacing it. There’’s a lot of scientific evidence showing that tutoring is one of the most effective tools to support learning. Everyone who ever taught knows that tutoring works. We have known this in an empirical sense since Benjamin Bloom (yes, the one of taxonomy) wrote his article on the 2-sigma studies (Bloom, 1984).
Bloom describes the task of educational research (the search for the instructional approach that has the same 2-sigma effect in large class groups) as the 2-sigma problem, because in a one-to-one lesson the average performance of a student is 2 sigma (i.e., 2 standard deviations) higher than in an average class with thirty students. An average improvement of 2 sigma means that the average student who participated in the one-to-one lesson performs better than 98% of the students who took part in the conventional lesson. Note: we’re talking about education in the eighties here. (Kirschner, Claessens, & Raaijmakers, 2018, ‘Op de schouders van reuzen’ [On The Shoulders Of Giants], p. 13-14,).
 In “How learning happens, Kirschner & Hendrick write:
What is particularly striking is that the average tutored student score was above 98% of the students in the control class, meaning that the vast majority of students in this study have the capability to achieve a high level of learning under the right conditions. The question then posted by the Bloom is; can we develop a way of achieving the same kind of results from the tutoring conditions under group teaching conditions? Bloom refers to this dilemma as the “2 sigma problem”. (Kirschner & Hendrick, 2020, p. 129). In Benjamin Blooms’ (1984) own words: “The tutoring process demonstrates that most of the students do have the potential to reach this high level of learning. I believe an important task of research and instruction is to seek ways of accomplishing this under more practical and realistic conditions than the one-to-one tutoring, which is too costly for most societies to bear on a large scale. This is the “2 sigma” problem.” (p. 4)
In a recently published review study, tutoring is described as one of the most versatile and potentially transformative educational tools in use today (Nickow, Oreopoulos, & Quan, 2020). In the Teaching Toolkit of the Education Endowment Foundation, it’s estimated that structural tutoring projects provide students with a learning gain of 4 to 5 months (see here and here). The evidence that (partial) distance education slows down the learning process of large groups of learners is piling up. Identifying this problem is one thing, actively engaging in a solution is another. So, it becomes just as important to design strategies that meet the individual needs of the students. Tutoring thus seems like an obvious answer (Dietrichson et al, 2017). We’re not alone is this perspective: many blogs from Robert Slavin point towards the same solution: tutoring.
After all, many studies show that tutoring has a strong positive impact. It works best for young students in primary education, but it also has positive effects in secondary education. Tutoring has mainly been studied in the context of mathematics/arithmetic and language, but can of course also be applied to other subjects. Tutoring is most effective when taught by teachers or external professionals. With non-professional volunteers and parents, the effects are positive but smaller. Tutoring appears to be most effective when it’s organised at school, more so than after school. There’s a lot of variation in the possible ‘time tutored’, but effective tutoring rarely lasts more than an hour, is given one to four times a week, and rarely more than 20 consecutive weeks on one subject. Often, tutoring stops when the student has ‘caught up’. Ideally, the number of students in the tutor group is kept to one to three students and certainly doesn’t have more than seven students.
Below is a diagram that maps out a number of ‘input factors’: the elements that schools and teachers can consider when setting up tutoring. The impact, on the far right, is shown in an effect size, expressed in number of standard deviations difference.
What are standard deviation and effect size?
Standard deviation (sd or σ – the Greek letter sigma) is a number used to represent how measurements are spread out in relation to an average of a group or population. This can be heights, weights, achievement in school, etc. For IQ, for example, the population average is 100 and the standard deviation is 15. This means that an IQ of 85 or 115 differs 1 σ from the population and an IQ of 70 or 130 differs 2 σ. Also, 68.2% of the population falls within 1 σ of the average (thus between an IQ of 85 and 115) and 95.6% of the population within 2 σ (between 70 and 130; see Figure 13.1).
Effect size (d) is a measurement used to express the size of a particular effect, for example to quantify the size of the difference between two different teaching approaches. In general we speak of a small (d=0.2), medium (d=0.4), or large (d=0.6) effect size. John Hattie defined d=0.4 as the hinge point because in his studies he found that this effect size is approximately what one will see if we do nothing special but just measure achievement gains after one year of school (i.e., pure maturation). That means if an intervention doesn’t exceed this 0.4 you could have just as well done nothing and let the child age, and if it’s less than 0.4, then it actually hindered learning.
 How Learning Happens, p. 128
 This is especially the case in children.
For example, the learning gain for tutoring interventions is 0.37 SD. In plain language, this means that 64% of the group of students who received tutoring will score higher than the average of the group of students who did not receive tutoring.
Why does tutoring work?
There are various possible reasons why tutoring improves learning.
Reason 1: It helps students who have fallen behind by simply giving them more instruction time.
Reason 2: It provides the possibility to customise the learning process. Teaching at the appropriate level is easier when you have 3 students than when you have 26.
However, the individualised aspect cannot be the only explanation, because if that would be the case, digital, personalised learning environments could have the same effect as human tutoring and it doesn’t (see this work by Robert Slavin). Therefore…
Reason 3: The human connection generated by the tutor-student relationship (i.e., the mentorship relationship), providing personal attention and focus – among other things – impacts learning positively. In addition, teachers often spontaneously use effective instructional approaches and only phase out support when possible. They first model and explain, then let students try them themselves, next they offer independent practice opportunities and of course they give quick feedback in the meantime while looking over the shoulder of the student. In addition, one can respond flexibly to what happens in the moment. By the way, these approaches also work well in larger groups.
How do other countries tackle tutoring?
In England, large budgets (comparable with an investment in Flanders of €90 per student) have been set aside for organised, structural tutoring projects where external partners provide the tutors. For example, England has allocated £76 million (!) to Tuition Partners in order to give schools access to high-quality tutoring from an approved list of providers, all of whom have met a range of quality standards. For example, Teach First has been allocated £6.4 million to fund academic mentors, giving schools in the most disadvantaged areas access to a pool of trained graduates they can use to support their students. Each mentor receives training in how to provide optimal tutoring. These academic mentors will provide intensive learning support while teachers can focus on teaching to the whole group.
Whoa! That’s Expensive!
Because teachers have a lot on their plate, it might seem as if the extras, such as the occasional tutoring, come on ’top’ of everything else. Yet, we dare to say that providing small-scale (i.e., not structural) tutoring to students who have fallen behind is actually (or should be!) part of the core business of education. Sometimes only five minutes of extra instruction or attention is enough to do the trick. At times a student won’t understand something until the teacher has explained it a fourth time and if that’s the case, the teacher has the responsibility to indeed give that explanation four times. Of course, this is by no means a plea for teachers to give up all their breaks (after all, each person is entitled to a break) and to give loads of tutoring at any requested time. Fortunately, it doesn’t have to be ‘that big’. Many teachers engage in tutoring spontaneously, often without even realising that an additional brief explanation in the school hall can have more impact than one might suspect.
This is different from the student who needs two extra hours of tuition for a subject every week. This type of structural tutoring is more difficult to cook up and it wouldn’t be fair nor realistic to expect teachers to invest time in that type of tutoring. Structural tutoring in small groups is therefore an expensive affair, as illustrated by the British approach. It requires many teachers, whom we simply don’t have. Commitment to structural tutoring requires a well-thought-out vision at a school level (adapted to the needs of the school and the pupil audience) and at a policy level, for example when considering the channel through which external professionals (recent graduates in higher education or people from the private sector who want to support) are trained for tutoring projects.
Our hope is that this blog is a reminder of the power and impact of that small-scale intervention and a ‘boost’ for the teacher and school who invest their precious time: tutoring makes a big difference for students.
Dietrichson, J., Bøg, M., Filges, T., & Klint Jørgensen, A. M. (2017). Academic interventions for elementary and middle school students with low socioeconomic status: A systematic review and meta-analysis. Review of Educational Research, 87(2), 243-282.
Kirschner, P. A., & Hendrick, C. (2020). How learning happens: Seminal works in educational psychology and what they mean in practice. London, UK: Routledge.
Nickow, A. J., Oreopoulos, P., &Quan, V. (2020). The impressive effects of tutoring on preK-12 learning: A systematic review and meta-analysis of the experimental evidence. (EdWorkingPaper: 20-267). Retrieved from Annenberg Institute at Brown University: https://doi.org/10.26300/eh0c-pc52
Higgins, S., Kokotsaki, D., & Coe, R. (2012). The teaching and learning toolkit. Education Endowment Foundation and Sutton Trust. https://dro.dur.ac.uk/20987/