Growth mindset fails to increase grades or non-cognitive skills. What now?

Growth mindset is a popular classroom intervention which attempts to change children’s views towards studying and class work. The core principle is teaching students their brain has the potential to grow by exerting mental effort during challenging tasks. This will increase the pupil’s willingness to persevere with work, and will subsequently improve their grades (Dweck, 2006). It has been adopted in schools in a large number of countries, from Hungary to Norway to India (O’Hara, 2019). Many teachers and students value the intervention (Yettick, Harwin, Riemer, & Swanson, 2016) and there have been several studies finding positive results e.g. Yeager et al. (2016)[note]This study should be commended for its use of preregistration and data collection and cleaning by an independent firm.[/note] and Bettinger et al. (2018).

The Education Endowment Foundation (EEF) is a charity which funds high quality research into education interventions with the goal of improving practices in classrooms. They recently published a randomised control trial (RCT) examining the effect of growth mindset on literacy and numeracy grades, as well as non-cognitive skills like test anxiety (Foliano, Rolfe, Buzzeo, Runge, Wilkinson, 2019). This was a high quality test of the efficacy of growth mindset on factors previously shown to be impacted by it.

Grand designs

4584 Year 6 students across the whole of England took part. There was a mixture of high, medium, and low performing schools[note]Determined by Key Stage (KS) 1 results.[/note] randomly assigned to either the intervention or control. The intervention consisted of 2 types of interventions which had previously shown promise. Pupils were trained across 8 sessions using a structured series of learning resources and activities on intelligence, how to deal with mistakes and emotions, and encouraging perseverance. Teachers went to a 1 day training session on mindset theory, the evidence for it, and received pointers on how to use the approach in their classrooms. This was supplemented by giving them materials to teach a 8-week programme on growth mindset, which formed the pupil teaching arm of the intervention. 8 months after the first training session all students sat their KS2 exams. The control group did not receive any training prior to data collection[note]A ‘waitlist’ control group.[/note].

Beyond measure

The primary outcomes was the pupils’ scores in their reading, numeracy, and grammar, punctuation, and spelling (GPS) KS2 exams. The secondary outcomes were their scores[note]From 1 (‘strongly disagree’) to 7 (‘strongly agree’).[/note] on a series of questions exploring non-cognitive skills. These skills were: self-efficacy[note]How much a pupil believes in their ability to succeed at a task.[/note], intrinsic value[note]The extent to which pupils’ participation in an academic task is the goal, rather than being a means to an end.[/note], test anxiety, and self-regulation[note]How well can students control their cognitive strategies to succeed in an academic task.[/note]. They were tested before and after the intervention.

Prior to data collection, the researchers ran a power analysis to determine how many participants they would need to calculate an effect size of 0.2[note]With a variety of assumptions e.g. 0.8 power, 0.05 significance level.[/note]. Their minimum detectable effect size (MDES) reduced to 0.09 after participant recruit, thus they had a good chance of detecting a very small effect.

What did the study find?

The difference between the control group and the intervention group on all 3 primary outcomes was 0. The 95% confidence intervals (CI) were -0.04 to 0.01 for maths, -0.02 to 0.02 for reading, and -0.03 to 0.03 for GPS. The difference between the groups for all 4 secondary outcomes was also 0. The 95% CI were -0.10 to 0.28 for intrinsic value, -0.21 to 0.10 for self-efficacy, -0.09 to 0.06 for test anxiety, and -0.11 to 0.21 for self-regulation. The CI were wider for the non-cognitive measures because they had a much larger % of missing data and a correspondingly smaller sample size. For secondary measures, the % of missing data was over 50%. This reduced both the informative value and the statistical power of the secondary results. The missing data depended on the observed data and followed a pattern[note]Specifically, missingness could be predicted by Free School Meal (FSM) eligibility, ethnicity, and prior achievement.[/note], which is called missing at random. However, analyses incorporating these factors into the model suggested the missing data did not affect the overall results for the secondary measures.

Some participants reported short-term changes in pupils’ enthusiasm, perseverance, and attitudes. But these were self-reported by participants at 6 treatment schools interviewed post intervention. Thus, it is not known whether this was found across a majority of the schools.

Everyone has issues

The report details some of the reasons why they may have not found evidence of efficacy. One of them was the possibility of cross group contamination: a significant portion of the control schools were already using growth mindset to some degree. Schools that had already delivered a school-wide growth mindset programme were excluded, but 2/3’s of teachers in the control group reported having knowledge of growth mindset. Most teachers reported not using growth mindset techniques prior, but they may have incorporated some of it into their practice without explicitly choosing to.

Another identified issue centred around the age of the pupils and the longevity of the intervention. 11-year-old students may be too young to utilise the skills taught using growth mindset interventions. For example, young students may not be able to direct their learning in the same way as older students. Pupils may also take more time than roughly 8 months to fully learn the skills taught. Therefore, the intervention may have collected data before a growth mindset could have helped the students[note]The EEF share their data so it may be possible to run a follow-up examination to see if there were any long-term effects.[/note].

The last issue relates to whether the schools kept delivering the training as taught. Schools were asked to report back weekly how long their sessions lasted. By the final week, only 9 of the schools sent this data back (49 sent data in the first week). However, this does not mean the schools didn’t deliver all the sessions[note]Built into the intervention was flexibility as to how long the sessions would last and for how many weeks the schools would deliver the sessions. Schools could therefore deliver all the training without using the full time allocation of 8 weeks.[/note]; it just means they didn’t tell the researchers they were.

Candace Lapan wrote a very good tweet thread critically looking at the RCT, I recommend reading it:

Does growth mindset have a positive impact on students?

This RCT was a highly powered test of the efficacy of growth mindset in a real-world environment across a wide range of schools in the England. The fact none of the primary or secondary outcomes were distinguishable from 0 raises serious questions as to the efficacy of growth mindset for Year 6 students. This RCT comes off the back of a meta-analysis by Sisk, Burgoyne, Sun, Butler, Macnamara (2018), which found only slightly more promising results. They ran two meta-analyses looking at: the relationship between growth mindset and academic outcomes, and the effect of mindset interventions on academic achievement. They found weak evidence for both research questions (r = .10, 95% CI = 0.08 to 0.13, and d = 0.08, 95% CI = 0.02 to 0.14, respectively).

In response, Carol Dweck provided a defence of growth mindset interventions (Dweck, 2018) which I recommend reading. She highlighted that the literature predicts certain groups would benefit from growth mindset more than others. This was found in the meta-analyses. Those with high academic risk e.g. students who had previously failed courses, had an effect size of d = 0.19 (95% CI = 0.02 to 0.36). Those of lower socio-economic status had an effect size of d = 0.34 (95% CI = 0.07 to 0.62). However, the very wide CI reduces our ability to know how effective growth mindset is.

What’s a teacher to do?

In my humble opinion, the evidence for growth mindset is currently underwhelming at best. Whilst there are positive studies, these typically fall into two categories. The first are higher quality studies which find effect sizes of a similar magnitude to Sisk, Burgoyne, Sun, Butler, Macnamara (2018) e.g. d = 0.10 for those with lower prior attainment (Yeager et al., 2016). The second are studies which seem to employ questionable research practices[note]Probably not intentionally, as many researchers don’t realise these lead to false inferences (Simmons, Nelson, Simonsohn, 2011).[/note] to find significance e.g. Yeager et al. (2014) finding significance in the garden of forking paths (Gelman and Loken, 2013)[note]The garden of forking paths is a metaphor for researcher degrees of freedom in data collection and analysis. By making seemingly reasonable decisions e.g. dividing groups into smaller groups and reanalysing after the first set of analyses, you inflate the chance of a false positive.[/note].

So should schools abandon growth mindset as a dead end? In the RCT, all of the treatment schools stated they planned on encouraging a growth mindset in the future. The intervention costs little for schools (approximately £4 per pupil), which is always appealing to schools looking to save money due to funding cuts (Coughlan, 2019). But if it has almost no real impact, can even a small amount be justified? And it does take up time. Arguably not a lot, but time is precious in schools. Time could certainly be spent on more valuable and effective things. Even having that time as a break from lessons could be valuable, as break and lunch times are important for children’s development and there have been large time cuts to them (Baines and Blatchford, 2019).

Given the evidence so far, it is unrealistic to expect growth mindset to have large and/or wide-scale impact. But if there is suitable time given to develop these mindsets and it is focused more on specific groups, there may be value to using this tool. Ultimately though, it is up to schools and teachers to decide whether to use growth mindset or not.


[zotpress items=”{5421944:6R5N4NSX}” style=”apa”]

[zotpress items=”{5421944:XWX73JYU}” style=”apa”]

[zotpress items=”{5421944:6TIWX58U}” style=”apa”]

[zotpress items=”{5421944:QNB83R4C}” style=”apa”]

[zotpress items=”{5421944:DLBR27CE}” style=”apa”]

[zotpress items=”{5421944:2LUDKHXB}” style=”apa”]

[zotpress items=”{5421944:24YMEF5U}” style=”apa”]

[zotpress items=”{5421944:XD2SWENX}” style=”apa”]

[zotpress items=”{5421944:LBJMSMWC}” style=”apa”]

[zotpress items=”{5421944:RBQ5SCRM}” style=”apa”]

[zotpress items=”{5421944:SG9T76R4}” style=”apa”]

[zotpress items=”{5421944:RZU23GRS}” style=”apa”]

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: