Cognitive bias in school-based research

In developing a culture of research, I have been thinking about how we can best support our staff team in spotting bias in their own research and in the research of others.

Cognitive biasHeuristic – providing a simple and often imperfect answer to a difficult question (Kahneman, 2011)

I love the work of Daniel Kahneman and if you haven’t read his book, Thinking Fast, Thinking Slow it is a worthy read for anyone engaging in reviewing literature or indeed working in school leadership. Here is a link to a previous blog I wrote on slow leadership.

Research is riddled with heuristic. While the Cambridge English Dictionary defines heuristic as a method of learning or solving problems that allows people to discover things for themselves. There is a wider caution of heuristic in school-based research. The heuristic is described by Kahneman as providing ‘a simple and often imperfect answer to a complicated question’. An example of this would be to respond to the question posed by a work colleague, ‘how are you?’with the standard and benign response ‘Fine.’. If we are to use some slow thinking, we would delve, and ask a question that would proffer a more fulsome response. We could encourage the respondent to dig deeper into their response by asking ‘are you equally happy about all aspects of your work life?’. This may ask the responder to go beyond their initial heuristic of ‘Fine’ and they may then be prompted to speak more widely about their job role, their relationships with their team leader, their frustrations about appraisal, their joy of teaching music to a group of pupils and so on. Challenging our school-based researchers to go beyond the heuristic is key in broadening their understanding of their field of study.

In his book Thinking Fast, Thinking Slow, Kahneman talks of lazy thinking. His book is truly worthy of a read for any researcher considering the effects of bias on their own research and the research of others. He outlines two systems of thinking. Genuinely, I’m not making money on sales of his book. The first system depends on heuristic. This lazy thinking allows someone to answer a complex question swiftly without engaging in deep thought.  It is in this system of thinking that our cognitive bias runs riot and we come to decisions based on our current thinking or little thought at all.  Just like the respondent above who answered ‘fine’ when asked ‘how are you?’. In becoming increasingly aware of cognitive bias, and in particular, recognising bias in your own and in others, your researcher will become increasingly deep thinking.  They will move their thinking from lazy system 1 thinking into deep system 2 thought where the multifaceted threads of their field of study will start to present themselves.

I share these cognitive biases with my early career teacher research group to help them to develop an awareness of their own bias and to spot bias with increasing accuracy in the research they encounter. Knowledge of cognitive bias is also invaluable as a school leader as a shared understanding of cognitive bias in a senior team opens out the breadth of discussion you can have with one another.   I will now draw on some of the cognitive biases outlined Kahneman’s book.

Availability Bias – WYSiATi
Availability bias is rooted in our natural response to answering a difficult question. It operates on the assumption that we already know all there is on the topic being discussed. Put in simple terms by Kahnaman, he uses the acronym WYSiATi to represent ‘What You See is All There is’. This sums up the availability bias in us all that we often assume that we know all there is to be known on a given subject when forming our response.

An example of availability bias operating in a school can be seen in this scenario. One senior teacher, observing the behaviours in a teacher’s class states, ‘ Jane’s teaching seems to be deteriorating – she has a tough class – she clearly isn’t coping well with the pupils.’. Jane’s team leader is trapped by some lazy thinking and their availability bias is blocking them from thinking more deeply about what is not known. When questioned, Jane’s work colleague states, ‘Jane has changed her teaching strategies recently and is off curriculum as she needs to plug concept gaps for the students who have now entered Year 10 after a Year 9 where six supply teachers led to poor progression for the students.’ Further to this, when Jane is questioned about the issue the leader finds that Jane has noted the students have poor collaborative skills so has mixed up previously established groupings to help build resilience in the students prior to their accelerated programme for GCSE. While this has unsettled the students in the short term, causing low level disruption, it is a strategy to support progress in the long term.

As you can see, in the assumption that what you know is all there is, the understanding of Jane’s rationale for her classroom management would go unseen. By working through this cognitive bias, the senior leader is in far better position to support Jane to get the very best from her students.  This also works in the field of research, and in particular when forming the research question and engaging in reviewing the evidence relating to your researcher’s field of study. We need our researchers to be aware of availability bias in both themselves as researchers and in the research and ideas presented by others. We do this by asking if  what you see is all there is ?

Confirmation Bias

Confirmation bias is where we seek out only that which confirms what we already know or think. This is seen in the associations we unconsciously make. For example, if you buy a new car, without warning the road is full of drivers in that same car. Our brains are hard-wired to recognise and make sense of patterns and connections. Because of this, we often subconsciously ignore anything that does not support our view of the world. Confirmation bias is important to recognise in the research of others as well as in our own research and our professional lives.

So, how does confirmation bias appear in schools and in research? Let us consider the following statement, ‘Mrs Jones gets great results in her maths GCSE by grouping students by gender’. Being aware of confirmation bias, what are the assumptions in the statement? One assumption would be that gender segregation has resulted in the great results. A second assumption could be that all students in Mrs Jones’ class achieve these great results.  A third assumption may link to the quality of teaching delivered by Mrs Jones, who may have a range of interpersonal skills that lead to high engagement for all her pupils. I could go on.  Without thinking more deeply about the wider factors, the researcher could use this statement to simply confirm what they already think about gender grouping, asserting that gender segregation in the teaching of maths is the key factor for the results achieved by students.

Sunk cost fallacy

Sunk cost fallacy describes a bias where you continue blindly doing something because you have invested time or money in it. You see this in someone who goes to the theatre to see a play and finds it to be the most tedious experience ever. They reach the interval but head back to their seat after because they have paid for the ticket and although they are not enjoying the experience, continue as they don’t want to waste the money spent. Taking this bias into the school environment, we may hear a school leader say when questioned by their staff on the value of the digital assessment system in place, ‘we have been using this assessment system for five years. We have invested time and money in the system and it produces super graphs of students’ progress.’. 
Then a keen member of the senior team states, ‘Is there a better solution to manage the time our staff are spending entering data into the system?’. With sunk cost fallacy engrained unconsciously in the school leader’s mind, they respond by saying,  ‘We have been using this assessment system for five years. We have invested time and money in the system and it produces super graphs of students’ progress.’. 
‘But,’ retorts the keen teacher, in a desperate attempt for the school leader to see the folly of their thinking, ‘The current assessments on the system are no longer in line with the new exam board’s expectations.’. The senior leader then responds (yes you’ve guessed it) ‘Seriously, do you know how much time it took us to train our teachers to use this system and what the set up costs were? We have been using this assessment system for five years. We have invested time and money in the system and it produces super graphs of students’ progress.’. The school leader is resolute, even though the wider evidence suggests the current system is no longer fit for purpose.

Sunk cost fallacy is an unconscious bias in us all and you will find this also in research articles, books and rooted in the practice of colleagues. You need to alert your researcher to listen for evidence of sunk loss fallacy as this is often used as a key argument for opposing organisational change in schools. As such, research can sometimes be wrongly used to validate the status quo and obstruct the evolution of new practice within our schools.

Group Bias – if all around you agree, it must be true.

Group bias is an unconscious bias that relies on the social convention that if all around you agree, it must be true. An example of group bias is seen in the story of the Emperor’s New Clothes where the Emperor is convinced by his aides that while his new robes are invisible to him they are visibly the finest and most luxurious stately robes  to all around him. As his subjects were all too scared to admit to the Emperor was in fact naked, the Emperor himself believed what was blatantly not true. Group bias is a powerful and convincing bias and has caused atrocities to take place across the world as groups of people and nations hold biased, yet unfounded viewpoints.

In the field of research, group bias can be seen where groups of people gather with similar viewpoints. This is often the case on social media where groups of like-minded or like-opinionated people meet. Forming a bubble of understanding. In order to dispel this bias, a lone voice in the group needs to ask a challenging question, ‘ but what would others who may disagree with our perspective say and on what information would their opinions be founded?’ This is an important bias for your researcher to challenge when undertaking their evidence gathering relating to their research question.

Over confidence Bias
‘In the modern world the stupid are certain while the intelligent are full of doubt.’ Bertrand Russel

Find someone who has an absolute view on an issue and therein you are likely to find over confidence is a bias. This is also linked to the Dunning Kruger Effect, that stupid people are too stupid to recognise their own stupidity. (Kahneman, 2011). It is over confidence bias that can lead a selection panel to appoint the most enthusiastic candidate and then find they are not capable in the roles and responsibilities of the job. To counter over confidence bias, we need to help our researchers to understand the simple principle that the more we know, the more we know we don’t know. Also, to watch out for over confidence in the evidence they use to deepen their understanding of their field of study.

Anchoring effectYour researcher also needs to be alert to the dangers of being blinded by big numbers. The anchoring effect (Kahneman, 2011) is a cognitive bias that connects a false validity to research outcomes because the number appears significant. Remember the importance of the p level or Cohen’s level in academic research which clearly quantifies the significance of the research. An example of the anchoring effect is seen when shopping and you spot a pair of shoes for £120. You then spot a similar pair for £60 and you immediately feel you have found a bargain. In using your initial reference point of £120, you naturally assume that this is all the information you need and come to the conclusion that £60 is a great price to pay. However, there is a danger of referencing your conclusion to the first or limited piece of information you have, as it may well be that the shoe at £60 is also over-priced and a wider search of other shops give a much broader picture of the value of these shoes.
Another example of the anchoring effect is seen in the staff room when introducing a new school initiative with colleagues. It is often the voice of caution or dissention that will speak out first in a discussion. The response usually starts with the word ‘but…’. The anchoring effect comes into play and it is then very hard to argue against the first comment made, particularly if this is a negative comment presented with conviction. The negative comment then becomes the ‘anchor’ to all other thinking on this issue and the conversation and wider possibilities of a deeper discussion considering all the perspectives available is hindered. Your researcher needs to be alert to the anchoring effect as they may come across a convincing and compelling argument from a member of staff when investigating their field of study. If they are convinced by the first strong opinion they meet, they may not deepen their understanding of the wider perspectives of others.

Halo Effect

The halo effect is similar in ways to the over confidence bias. It depends on an unconscious bias that we are drawn to believe someone who is deeply enthusiastic about their viewpoint or has a track record of performing well. Teachers are more likely to grade a student’s essay favourably if the student has a track record of writing well. David Didau puts it in this way when writing about the achievement of boys in English exams, ‘Do we expect girls to be more compliant and achieve better than boys in school? Are boys and girls treated differently in school and wider society? We expect girls to be made of sugar and spice and all things nice while boys are unwashed louts. Might we be making it easier for girls to achieve in schools because of the expectation we have of them?’ (Didau, 2015).

Be on your guard

As we strengthen our researcher’s capacity to review evidence relating to their research question, we need to draw their attention to the powerful influence of unconscious cognitive bias. In learning more about bias in other’s work, they become increasingly erudite at spotting this bias in themselves and as such, strengthening their objectivity when engaging in their own research.

References:
Didau, D (2015) What if everything you knew about teaching is wrong? Crown House Publishing 
Kahneman, D (2011) Thinking, fast and slow. Penguin

When thinking of research in schools, the key to a purposeful research based enquiry is to lead the researcher to a helpful question for their research project. This process is one that can often be hampered by what Daniel Kahnaman would call ‘fast thinking’. Jumping at the first question that springs to mind can often lead to difficulties in the research journey as the question may be too broad, overly vague, biased or just plain boring. The question needs to have purpose and the researcher needs to ensure the question leads them towards the desired outcome of their research. The researcher needs to have the end point, or at least a presumption of what the end point may look like, in view. If the researcher settles on the wrong question, there is a real danger that the research that is undertaken leads them to an end point that lacks purpose or relevance to their practice.

Gary Thomas 2017, in his book How To Do Your Research Project,  suggests there are four types of research questions and defines the as:

1.       What’s the situation?

2.       What’s going on here?

3.       What happens when?

4.       What is related to what?

These help the researcher to begin to define a purposeful question.

What’s the situation?

The first of Thomas’ questions asks the researcher to simply describe what they are observing.  This is a helpful starting point for a researcher’s first steps into research in schools. This may be seen in a teaching assistant asking the question, ‘what are the different strategies for supporting children with dyslexia?’. At first sight this is a helpful question to allow the teaching assistant who may be new to post explore the different strategies for supporting a pupil with dyslexia within the school. The question may, however require further clarification.  It is the skill of the coach to make sure the researcher a refines their question to ensure it aligns to the desired outcome of their research.Taking this example, the coach needs to ask the researcher to be more specific about the first research question. The coach could ask, ‘is there a particular subject area that would be of interest to your research question’ or ‘is there a specific age group, gender, ability, level of deprivation you are interested in in order to make your research question more specific?’ The coach is trying, though these questions, to help the researcher to narrow their field of study in order to make their research both increasingly manageable and the outcome more impactful. With a narrower focus on this descriptive question, the researcher has greater chance in finding out something of interest that will impact positively on their practice and potentially the practice of others.

Thomas draws an analogy between this style of question and painting a picture. He states that the researcher should, through their research question and subsequent research, be able to paint a picture of what is happening through words.  As with any painting, however, it is provided through the lens of the artist, and this is also the case in research.  The researcher can put their own interpretation on the research, making it unique to their setting and their own interests.  This is not something to shy away from, but to embrace.  The researcher, provided they name their own perspective in this research and their own potential bias, can share their findings to a well constructed situation question freely.

What’s going on here?

The ‘what’s going on here?’ style of question asks the researcher to question deeper. Thomas suggests this is likened to shining a spotlight on a darkened area of understanding. The spotlight, metaphorically the research question, allows the researcher to shine light, or understanding, on a key part of their practice that is currently not fully understood. Thomas puts this well as he writes:

‘First, it means that you are expecting to see something that you couldn’t see before. Second, it implies also that you will be able to see because you are looking in a way that you weren’t able to previously. Third, it implies you are giving time and energy to looking hard 9i.e. shining the light) and using your own self – your intelligence and experience – to make sense of the subject under study.’   Thomas, 2017

The researcher asks about an element of their practice, forming the question, ‘Why do children in my intervention group struggle to stay on task in maths lessons?’. This question poses a problem to the researcher who now needs to try to get into the heads of the children in this group.  An impossible task. However, the question then asks the researcher to be subjective in their research. Being subjective is sometimes seen as a weaker style of research compared to a more objective approach. Subjective research is perfectly valid as a research process, provided the researcher acknowledges this and states their potential bias or viewpoint.  

What happens when?

This approach asks the researcher to ask about possibilities. This may be linked to a question that asks ‘what happens to pupil’s confidence when a teacher limits their praise?’. The question relies on the researcher making a hypothesis that something will happen as a result of an action. This action may be an existing element of practice or, more typically, when a new element of practice is introduced. In this case the shift from furnishing pupils with praise to a more limited offer of praise.

The ‘what happens when?’ question can be helpful in the researcher developing an element of their own practice and being able to use the research to validate their rationale behind the practice. In this example, the researcher asking about the impact of praise on confidence in their pupils leads to some further considerations. Firstly, would the research question involve a limited group of pupils? Would the pupil group be those that are currently confident, or target those who are less confident? Would the research group be limited to a particular subject, time of day, period in the school year? The coach needs to unpack all these possibilities with the e researcher in order to make sure the research question is refined and the research that leads from it is helpful to their practice.
Another example of the what happens when question was posed by a member of my early years team. The researcher asked, ‘What happens to the levels of concentration for my group of autistic children when adults walk through the room?’. The context of this question was rooted in the early years classroom being used as a walk through room when the hall was in use. With three autistic children in the class, the researcher wanted to research the impact of these perceived distractions on the children. This question was helpful in forming a basis to this research project but came with further questions for the coach. The coach asked:

Why is this an important field of research for you?

How do you hope this will impact on your practice?How do you hope this will impact on the pupils?
What assumptions are you making here?Can you refine your question to make sure you really target what you are trying to find out?
Can you spot any bias in your research question?

The ‘what happens when?’ question therefore tries to define whether one particular factor causes a reaction, or as Thomas puts it ‘does this cause that’ or ‘does x cause y?’. This can be a really helpful question style for staff introducing new initiatives in school and a well-constructed research question can unpack the impact of the new initiative. 

An interesting, but challenging, research question was raised by a member of the teaching team in our school as they asked ‘does research in school impact positively on practice?’. When I first read this question, I felt this teacher was questioning the whole approach to research informed practice in our school and trust. At first sight, this question appeared subversive and one that was set to challenge the approach to research across our entire staff body.  I considered the personal bias this member of the team was bringing to their research and that they possibly hoped to devalue research informed practice. I then, with Kahnaman’s slow thinking , stepped back from my initial reaction and recognised that I was bringing my own bias to this reaction. I had let my bias, in this case what Kahnaman would call the ‘halo effect’, cloud my openness to the light that research can cast on your own practice. I had allowed my passion and excitement for the impact of research informed practice to block the possibility that research itself could suggest that research did not impact positively on the organisation. I mention this only to alert the issue that bias is not only rooted in the researcher, but also in the coach and needs to be recognised and named where it arises. 

I am , however, more than relieved to state that the outcome of the research question by this teacher led to a fascinating study on the impact of research.  The research highlighted the affirming power of research on the individual and the organisation. The research also highlighted the importance of recognising the workload research places on some members of the team and through this being exposed through a well-crafted research question, our approach to supporting staff in understanding research methodology and providing time to all staff to undertake research days was introduced.

What is related to what?This style of question asks the researcher to consider how two or more factors interact to cause an outcome. This may be embedded in a question like the one used by a member of staff who asked ‘How does attendance, deprivation and special educational needs affect the outcomes in phonics for year 1 pupils?’. This question leads the researcher to consider the connectedness of differing factors on their research.  It leads the researcher to pose questions about the links between differing factors to help them draw an answer from this connectedness to find a new level of understanding.
As with the previous three forms of questions, the ‘what is related to what?’ question is open to cognitive bias. The researcher inevitably brings an assumption to this question. In the example above, the researcher may have made the assumption that attendance, deprivation and special educational needs have an adverse effect on outcomes in phonics for year 1 pupils. The research may therefore lead to the study of a group of pupils whose attainment in phonics may be limited by another factor beyond the study, such as lack of verbal engagement in the family home in the formative years. It is once again the role of the coach to carefully unpack these potential assumptions and biases through skilful questioning in order to define a helpful research question.

The hypothesis and null-hypothesis

‘Hypothesis – an idea or explanation of something that is based on a few known facts but that has not yet been proved to be true or correct. ‘ (Oxford, 2020)

Research questions can sometimes test out a hypothesis. Thomas refers to the hypothesis as the fifth style of question a researcher could choose.  The hypothesis presents a difficult choice of research question as it often makes an assumption of truth based on limited facts and reading. An example of the hypothesis is used in current school inspection framework in England where the lead inspector will consider the evidence gathered during the first day of a two day inspection and formulate an hypothesis to test out on the second day. An example hypothesis may be related to the leadership seen in three subjects seen where in each subject seen during the first day of the inspection a particular group of pupils appear to perform differently when compared to their peers. The following hypothesis is therefore formed by the inspector, ‘In a wide range of subjects, disadvantaged pupils make stronger progress compared with their peers as a result of the additional support they receive.’. This hypothesis, as a research tool, has difficulties. One key difficulty is that the hypothesis makes a presumption based on limited knowledge, this has the potential to put the metaphorical blinkers on the researcher. In setting sights on a hypothesis, there is a tendency for the researcher to endeavour, sometimes at all costs, to prove their hypothesis right. This creates a strength of bias that can lead to a foregone conclusion, limiting the validity of the research itself. While the hypothesis style approach to forming a research question is a valid process, the researcher must be cautious that they are not charging into a research project that is restricted by their bias.

As an undergraduate scientist in the early 1990s, I was introduced to the concept of the null-hypothesis. The null-hypothesis flips the bias created by the hypothesis on its head by inverting the research focus to try to disprove the researcher’s hypothesis or idea by suggesting that there is no relationship between two or more variables in a study. As an example, if a researcher was trying to prove the hypothesis above that ‘in a wide range of subjects, disadvantaged pupils make stronger progress compared with their peers as a result of the additional support they receive.’ The null hypothesis would turn this on its head. The null-hypothesis in this instance would read ‘in a wide range of subjects, disadvantaged pupils make less strong progress compared with their peers as a result of the additional support they receive.’ Or alternatively taking the additional support as the variable here we could consider the following null-hypothesis of ‘additional support for disadvantaged pupils has no effect on the progress made when compared to their peers in a wide range of subjects.’ By flipping the hypothesis, the researcher is encouraged to disprove their own hypothesis, and therefore be increasingly open to challenging their own cognitive bias.

To ask or not to ask, that is the question…
In formulating a research question, the researcher must therefore think carefully about what style of question will best suit their research. The coach plays a vital role in helping the researcher to refine their research question to ensure it is purposeful, manageable and relevant to the researcher. The research question is, however, still in its infancy. The question is open to growth as there is still much more that the researcher will find out. As the researcher moves through step three in the research cycle and reviews what is known about the issue, the question has a chance to grow. In addition to this, the researcher’s own understanding and awareness of their bias and the bias of others in their organisation will heighten. These factors will help the research question grow and mature into a meaningful question upon which the research enquiry can rest.

All the best in formulating your own research question.