Judgement and Decision Making in Construction
What is it, why you should be thinking about it, and how to get better at it.
TL;DR: Heuristics and emotional decision making definitely have a place in our lives. When you’re truly pressed and need to make a snap decision they can be life saving. Unfortunately, we sometimes fall back on them too readily and this can trip us up into making poor choices because they aren't necessarily optimal or rational. By using a systemised approach we can improve the quality and accuracy of our decision making. Measuring the outcomes of our decisions we can improve our future decision making. This can fundamentally transform the success of our endeavours. Define the true problem. Search for all options available. Use tested information from multiple sources. Plan. Measure. Learn.
What on earth does this have to do with digital engineering and BIM? Nothing… and everything. I would argue that all systems and processes we develop are to help us make effective judgements and make decisions. BIM, Prince2, Lean Thinking, Government Soft Landings, the RIBA Plan of Work…
I recently read the book ‘Noise: A flaw in human judgement’, and have previously read ‘Thinking, Fast and Slow’ (from the great Nobel Prize winner Daniel Kahneman) both have given me an interest in how decisions are made and how to improve my own approach to making them.
And so the below is a collection of my thoughts and those of others on the subject of decision making, which I’ve tried to convey through the eyes of construction.
I hope you find this useful, equally I hope if any points are incorrect or there is alternative thinking I can do then please point it out. I’m here to learn.
Introduction
Our judgements and decisions on a daily basis are core to our experience and outcomes in construction - as in an industry, and life. The better our decision making the better our outcomes - and I would argue our experience.
Yet how many of us consider the possible flaws in our judgement, truly evaluate how we are making the decision, fully consider the outcomes based on which decision we make, and then measure the effect of that decision?
I’m not saying that there needs to be a drawn out conundrum every time we need to make the stationary order but I would debate that we should be doing our best as individuals and organisations to consider how we can make consistently better decisions.
What effects our judgements and decisions?
Judgement and decision making as a field
There is whole field of science and thought around judgement and decision making. Let’s break down for you judgements and decision making effects with a construction lens.
In broad terms what do I mean by Judgement and Decision Making?
Judgement is the evaluation of information to make an informed decision. The ‘judge’ reviews the information available, balances the evidence, and then plots out possible outcomes based on the information available and the possible decisions to be made.
Decision Making is the process of selecting the course of action based upon the judgement made ending in a final choice. The decision maker assigns either formal or informal probability of success to the options available to them to make their selection.
In this article we’ll dig further into how this can go well or badly - and how we can help ourselves to do consistently better. Based on the above definitions there are three areas which can trip us up:
The quality of the information on which we judge - is it accurate, is it representative of the real world?
The quality of how the information is reviewed - is the ‘judge’ biased, is there ‘noise’ (more later), is the correct person (or group) making the judgement?
Completeness and correct assignment of outcome probability - are all possible outcomes considered, is the correct probability applied to each?
What is clear to me is that we will struggle to get this right for every decision but by applying the right processes we can raise the general quality of decisions taken - therefore, improving our results as a whole.
Daniel Kahneman (and Amos Tversky) - System 1 and System 2 thinking
Daniel Kahneman has spent a long time thinking about thinking. Awarded the Nobel Prize for his work in Behaviour Economics (with his colleague Amos Tversky), he may be best known for his book “Thinking, Fast and Slow”.
In this book Kahneman outlines his thesis on (as he calls it) System 1 and System 2 human modes of thought:
System 1 is our unconscious system, used to play out learnt patterns - it is fast, emotional, and automatic. Examples are;
Solve 2+2
Understand the sentence “My name is Neil”
Determine that an object is further away than another
Driving on an empty road (for experienced drivers)
System 2 is our conscious system, used to deliberately consider - it is slow, effortful, calculating, and logical. Examples are;
Find the value of z, if 2z + 2= 10
Try to recognise a sound
Count the number of E’s in the sentence “My name is Neil”
Drive on a busy roundabout
What does this mean?
Example sources of error specific to construction:
Here are some examples of errors in thinking that can be made due to the human condition. I’ve given a short description and then provided simple examples of that error at play:
Availability Heuristic (or Availability Bias)
This is a mental shortcut (System 1) which relies on immediate examples which come to mind when making a decision. If it can be recalled then it is important. This means that people tend to make judgements on more recent information, rather than all the information.
Example: John is a project manager and must decide on which supplier of drainage pipe to use who are equal on price and quality. Supplier A’s last delivery was on time and correct, Supplier B made a counting error on couplers on the last delivery. However, Supplier B has the overall best record of delivery. John chooses A due to the issues with B last time. Supplier B’s subpar performance last time was more ‘available’ and therefore ruled over actual long term performance difference between the two suppliers.
Anchoring
This is where a persons decision is influenced by a particular reference point or ‘anchor’. Once an anchor value is set this bases subsequent arguments and estimates around that set point. This means that people tend to be influenced by set reference point more than their own judgement of value.
Example: There has been an issue on a project and the client and contractor have entered into negotiation over compensation. The client opens by stating that they need £500,000 to compensate their loss. Though the contractor only assesses the loss to be around £100,000 they find it difficult to move too far away from the client’s initial ‘anchor’ as their assessment now ‘feels’ too low, too far away to reach consensus. They counter with £400,000.
Plan Continuation Bias (or Sunk Cost Fallacy)
The continuation of an existing plan even in the face of changing conditions. People become ‘stuck’ on the plan because of effort already given and/or money already spent. This means that changing conditions which effect the outcome of the plan are not countered and ultimately the outcome is less than optimal given these changed conditions.
Example: An energy company has began developing new gas production plant, they have reached funding, progressed design and land purchase, and have a supply chain in place. However, a global crisis is looming which is changing the demand for gas consumption in commercial and domestic use. As the company has already invested £25,000,000 and the remaining costs are ‘only’ another £75,000,000 they decide to proceed with construction.
Confirmation Bias
Confirmation bias is our tendency to search for, interpret, favour, and recall information in a way that supports our current beliefs or values. By selecting information that supports our views, ignoring contrary information, or when we interpret ambiguous evidence as supporting our existing attitudes - we neglect to think critically about our judgements. The effect is strongest for desired outcomes, for emotionally charged issues, and for deeply entrenched beliefs.
Example: Joan manages the Mechanical Design Team for a design consultant. The team are defining their delivery plan for their package of works. Joan feels that 3D modelling is a waste of time and money, but she consults her team for direction given her lack of hands-on experience. Katie comes to Joan with evidence from within the company that strongly supports 3D modelling methodologies as more effective at managing coordination internally and with other teams. Paul, however, states that 3D modelling is time consuming because it leads to more meetings. Joan goes with Peter and decides to complete the design in 2D to save all that wasted time she suspected from the start…
Causation vs Correlation. Performance vs Result. Regression to the Mean
Are you feeling lucky?
'Regression to the Mean' isn't thought about all that much in construction, but it plays a big part in the decision making and judgement of individuals, teams and organisations.
Basically:
Outcome = Skill + Luck
But our human traits see us often discounting luck as part of outcomes - we apply post outcome justification (story telling) to justify the outcome (hindsight bias). Our minds are wired to apply causation to things after the fact even when there is little evidence.
Our decisions suffer from this because we look at our outcomes and make future decisions based on the story we told ourselves about "the last time this happened" which is often based much more in intuition then on actual measurement of performance.
As an example; a contractor has a fantastic project and makes an exceptional % profit - this will often be deemed as excellent performance. The same contractor has a horror show on the next job and make a huge % loss. This will be deemed as poor performance. Can the performance have dropped off so much? In actual fact the performance on both can be exactly the same but the 'luck' changed - in fact it is statistically likely it will happen.
To counter this we must look to measure the quality of our decisions, and key metrics aside from a single outcome (in the example % profit). Doing so will allow us to understand the role of luck - which we can't influence - and see our true performance - which we can influence. By improving actual performance or 'skill' we can improve our mean and long term success.
Bias and Noise
Bias isn’t the only way our minds can work against us in decision making. ‘Noise’ is also an issue we need to be aware of - like “when was the last time you ate?”!
What is Noise?
If Bias is an individuals or groups leaning towards a certain behaviour or judgement, noise is the variability in judgement within yourself or the group for a single decision or over time. The below diagram illustrates this well:
Our state of mind at the time of making a decision will effect our decision, and this won’t be consistent over time. Ask yourself this - “how could my mood be effecting my decision making at the moment?”.
Examples of Noise
Education
A study looked at 682 real decisions by college admissions officers and found that the officers awarded the academic strengths of applicants more importance on cloudy days and, conversely, favoured non-academic strengths on sunny days.
Recruitment
A meta-analysis showed that a quarter of the time, two separate recruitment interviewers disagreed on which job candidate was the best fit for the job. This was despite the interviewers sitting on the same panel, thus having seen the candidates in the exact same circumstances.
Why think about decision making?
Impact of decision making - Construction’s Fat Tails
For this part I really must thank and recommend the work by Brian Potter. His article ‘Why it’s Hard to Innovate in Construction’ was a real head turner for me and is an excellent eye opener on why new ideas don’t stick too well in construction.
Brian discusses constructions ‘Fat Tails’. This means that in construction extreme events are more likely - statistically speaking. To use Brian’s words:
“What this means is what everyone who works in construction knows intuitively - a project that goes extremely well might come in 10 or 20% under budget. But a project that goes badly could be 100% over budget or more.”
Why? Well because of the very nature of construction - we prototype every time, to an extent, and on a fresh canvas (differing conditions). We are affected by the weather, a large and varied supply chain, economic forces, competition for workers, etc… And when the effects of the above occur they usually have knock on effects into other areas of our projects.
Put simply, construction is complex. Why does this matter in terms of decision making? In construction (as in many other settings) decisions have complex inputs, and their outputs have complex effects.
Marginal Gains or Swinging Big?
We make decisions every day, big and small. Therefore some may ask - can I just get the big decision’s right and relax on the small one’s?
Let’s think it through. Is it better to get a couple of big decisions right or get 100’s of small ones right? To answer this I turn to compound effects and marginal gains. Improving the situation by 1% isn’t particularly exciting, especially if that’s a 1% improvement on a small constituent part of a larger project. But - in the long run it is impactful. Given enough opportunities to change enough things by 1% the outcome of the larger piece can be hugely improved.
Big decisions do make a difference - which could be very large - so they should be given extra time. But my contention as we reach toward the end of this enourmous article is that we just need to apply a consistent approach to minimise bias, allow for noise, and consider the complexity for every decision we make.
How can you make better decisions?
Think Again
There are only so many routes into construction. Whilst there are many personalities every one of us have been conditioned by our training, our mentors, our roles, and our industry to an extent which effects our thinking and decision making.
Whilst this has upsides - one downside is that it can narrow our thinking and limit our potential. However you can, dear reader, and must find ways to pursue reconsideration and challenge your own beliefs, values, and biases.
If you’re looking to fill a position - maybe look to increase your recruiting pool to those outside of the industry. If you like to read, read widely and outside of your comfort zone. Don’t dismiss ideas out of hand, be humble and don’t stick doggedly to trenched thoughts, be open to the possibility that “the way we’ve always done it” may be the most dangerous words you mutter.
“Be open to the possibility that “the way we’ve always done it” may be the most dangerous words you mutter.”
DECIDE
To be consistent and apply structure to our decisions we can use frameworks. One of which is handily uses the acronym DECIDE.
D = Define the problem - what is the problem? Why should I do anything at all? What should or could be happening?
E = Establish the criteria - what would be desirable? what are we trying to achieve? what do we want to avoid?
C = Consider all alternatives - what are the options? what factors affect each alternative?
I = Identify the best option - based on experience, intuition, information available, and experimentation
D = Develop and implement the plan of action - how will I execute this option? what are the processes, resources required?
E = Evaluate and monitor the solution - troubleshoot the decision, carry out a pre-mortem, put a process in place to measure whether the decision has the desired affect, and take feedback when necessary to improve future decisions.
This looks like and could be a long process. However, this can take minutes once you’re in the habit of using it as a mental model. Try it an see how it can work for you.
It’s also worth looking at other decision frameworks such as OOPA Loops and the Plan-Do-Check-Act cycle to find the one that works best for you or the situation.
Automated Decision Making and Decision Support Systems
Sometimes it’s just easier for someone else to do it, and sometimes you should just let your computer - at least let a computation tell you the optimum option. Algorithmic decision making is not a panacea or the correct option every time. But once a workflow is understood and where inputs and outputs are digitised then computers can play a vital part in making decisions the way we want them to, every time.
Add in Machine Learning and the algorithm can self correct based on training data, and outcome analysis, through pattern recognition.
The benefits of this can be:
Leverage time and effort - build the system once, it will run in the background
Consistent result - no noise, consistent output (computers don’t get hangry!)
Audit trail - there will be a history of the optimum options, the formula used, and the data that drove them.
However, automation and computational decision support are not without there downsides which must always be considered and mitigated. Some of these are:
Built in bias - the way the algorithm is built to the training data used could build in inherent bias. These systems are better at removing ‘noise’ than 'bias’.
Shit in, shit out - data must be sufficient (in volume, quality, and representativeness) to ensure that the outcomes can be trusted.
Black box problem - unless you’re a programmer or data scientist you probably won’t know how the formula is working. Therefore, it may be difficult to understand why an output is what it is.
Decision Hygiene
As a decision comes clear there are methodologies to minimise the effect of ‘noise’. This will help to ensure decision variation between ‘judges’ (those making decisions) is reduced or eliminated. Some of these are:
Systemised approach - linked to automation or algorithmic methods - this is the strict adherence to a systemised approach to making the decision. Processes and procedures.
Take many people and then aggregate their judgements - basically take multiple people (as diverse as possible whilst retaining competence specific to the subject), give them the same problem and options, the take the popular outcome
The person or people to make the decision. Competence matters, competent people will be more accurate and probably less biased and noisy. Do remember, competence must be is a combination of training, skill, experience, and knowledge - be careful when assessing it.
Rank possible outcomes rather than scoring them. Working to a ranked scale turns out to be a lot less noise than giving scores.
Final Thoughts
Heuristics and emotion definitely have a place in the world. When you’re truly pressed and need to make a snap decision they can be life saving. Unfortunately, we sometimes fall back on them too much and can trip us up into making poor decisions.
By using a framework to make decisions - or take a systemised approach - then we can improve the quality and accuracy of our decision making. Further, by then measuring the outcomes of our decisions we can improve our future decision making.
Try to recognise in the situation you’re in whether the snap decision is needed or whether you have more time to consider a more optimal outcome.
Define the true problem. Search for all options available. Use tested information from multiple sources. Have a plan to deliver the optimum decision. Measure how it went.
With better decision making people, projects, teams, and organisations in construction can reap the reward of marginal gains - which can only help improve this vital industry.