PROBABILITY & RISK
See also pages on Judgement and Judgementalism ; Risk Dimensions ; Outcome formulation and Decision-Making.
We have added this page because if there is one subject in social care which is beset by confusion and misunderstanding it is risk and, even more so, the probability component of risk. Never has there been a subject so mishandled, bungled and confused in social care thinking. Speaking personally, I find much of social care writing on risk very confused and confusing.
There are many different theories of probability. There are at least 11 theories of which I am aware. You would never know this from reading social work risk textbooks. However, what should be clearly distinguished is that the (standard) mathematical theory is not the same as interpretations of the meaning of probability thinking. The standard mathematical theory lives in its own perfect Platonic world where if you were to roll a six-sided dice (die) you only ever have one chance in six of getting a six. There is no such thing as a biased die in Plato land. The real world is not like that. Hence very different interpretations (theories) of probability have evolved. These broadly fall into two major groups:
 |
Objective Theories |
|
Subjective Theories |
And there is a world of difference between them. And we use BOTH types in social care.
An account of some types of probability theories (interpretations of the meaning of probability) is to be found here. , (Wikipedia) or alternatively here. (Stanford Encyclopedia)
The Section below is taken from the Osiris Error Theory handout
Confusion regarding the concept of probability and risk: I lay the blame for this on the Colleges and the way they do not teach risk theory. The typical confusion centres on how the concept of objective probability is used within empirical research to build checklists (correspondences or cues) of factors that represent heightened levels of risk (e.g. dual diagnosis – mental ill health coupled with substance abuse can represent up to 33% higher risk of violence compared with the norm). This is then confused (transferred) to situations regarding the overall assessment of risk (e.g. Case Conferences) where, in reality, an entirely different model of probability holds sway – the subjective theory (degrees of belief). The subjective theory is about making judgements on risk which derive from a coherent analysis of strengths (positive risks) and weaknesses (negative risks) within the situation. Correspondences inform this analysis but cannot be used to make the overall judgement – this is a synthesis of all that is considered.
The
following is a quote from Dr
(now Professor) Roger Bullock (then) of the Dartington Unit taken from a Research into
Practice website article on: Child Protection: everybody's business
http://www.rip.org.uk/publications/documents/audioseries/tape7script.htm
(this link no longer works - quoted for reference only)
“Supposing
there are five of the sort of things I have mentioned, i.e. a violent partner in
the house; poor parental care; previous histories of child abuse. If you apply
those to a general population of all children you will find that probably 1 in
10 families show those factors and therefore do pose a risk to children but the
probability of an incident actually occurring is very much less and what people
get confused about is that you can be high risk of something but still low
probability. So I am at high risk of having an accident in my car in the next
half an hour because I have to drive home, whereas someone who hasn't got to
drive home isn't. So, I am at high risk of having an accident, but the
probability of having an accident is actually very, very small. Something that
probably only happens to us once in ten years, and then hopefully isn't very
serious. So we apply these factors to a general population and we find that
5-10% of families show these characteristics, whereas the number of families
that seriously abuse children or cause them injury or death is probably only 100
a year across the country when you look at a population of 55 million or 11
million children.”
In the above quote, Bullock does help illustrate the difference between objective and subjective probability theory. When Bullock talks of “high risk” he is referencing research on the objective (frequency) probability of car accidents near home. When he is talking of the probability of him actually having an accident on his way home, he is talking of what is called in risk theory the “point probability” (what subjective degree of belief can be reasonably considered as justified) on the prospect of his mooted accident. Bullock is quite wrong when he tries to divorce point probability from the concept of risk. This is just total nonsense, but he is completely right when he says that objective probabilities cannot uniquely predict (at least in child care or car accidents) point probabilities but only offer (at best) an advisory guide to likelihood.
The root of this difficulty goes deeper than this and concerns our use of different theories of truth (correspondence theory versus coherence theory) in differing modes of explanation. The problem is that in making judgements about complex areas of life (e.g. human interaction) we are constrained to use complex models of analysis (The Framework for Assessment is such a model) which depend on coherent analysis – as described above – rather than simplistic correspondence checklists of risk factors (see Hammond 1996). The scope for Judgement Error is therefore huge and is often (usually) compounded or even initiated by Organisational Errors which conspire to facilitate judgement error.
For example, the Caleb Ness Inquiry (2003) noted that mother Shirley’s lengthy record of failure to care for her existing two children was dismissed by workers as “historical”. Note that evidence of “probability” has now been reframed to be of no importance! And yet the thorough reading of ALL previous case records is a key requirement in trying to estimate future probabilities. This is an exercise in using objective probability (frequency of what has happened) to inform judgement on the likelihood of what will happen. The end judgement is itself an exercise in subjective probability - degree of belief about what will happen in the future.
So is "Risk" an objective or subjective concept?
The consideration of Risk is always subjective. If you stop for a moment to think about it you will see that when we consider something a risk we are entertaining thoughts of an outcome in a future world - a world in which the outcome we are considering has not yet happened. There cannot exist "past risk". The concept is nonsensical and mixes up the concept of "risk" with the concept of Induction and methods for the determination of event frequency (various objective probability theories)
How is it that people can get so confused about Probability?
What is the answer to the question: Will there be a sea battle tomorrow? How would you go about answering a future contingent like this? Suppose it true that there will be a sea battle. If so, it is false that there will not be a sea battle. But either there will be a sea battle or there won't be. And if it is true that there will be a sea battle it is always true. The philosopher Diodorus Cronus used the argument of the sea battle in the 4th century B.C. to help illustrate his so-called "Master Argument" for the truth of Fatalism. The philosopher Aristotle, in his De Interpretatione, wrestled with Diodorus' paradox and invented his famous Theory Of Potentiality as a result. But what Aristotle lacked was the concept of probability. The ancient Greeks simply did not have one.
Today we would argue that given the presence of two warring Nations' naval battle fleets in proximity of a day's sail of each other, there is a probability that the two will clash in a sea battle. Notice that our feeling for the probability of the sea battle is fed by our knowledge of the circumstances. We have two Nations at war with each other. This is not a good sign. If they were at peace there is still a possibility (a very small probability) that there will be a sea battle but we would consider it highly unlikely. Other knowledge factors will feed our estimates on the probability of battle. Suppose one fleet is small compared to the other and is poorly equipped in armament. The fleet Admiral will likely do everything possible to avoid battle. However, the Admiral of the stronger fleet will do everything possible to engage in battle. The question of whether there will be a sea battle tomorrow becomes, as in the famous Battle of Midway, a game of strategy and tactics and each Commander will try to estimate the probabilities of the moves of each fleet. It is a technical game (as in "Game Theory") albeit with severe consequences.
But what have sea battles got to do with social work? Everything is the answer! Future contingents abound in social work. Your job is about future contingents. Games also abound in social work but mostly they are of a different type to the sea battle. The sea battle is a zero-sum game. One side wins and the other loses. In social work we want both "sides" to win. We make every effort we can to promote cooperation and mutual working toward goals. This type is of game is called a non zero sum game.
For a further discussion on Game Theory in social work practice see: Alan Cooper (2004) Game Theory and Partnership in Child Protection Practice: Vol 16, No 3, September BASW Practice journal. However, this piece was written partly as a polemic against the pre-2006 practice of Child Protection Case Conferences operating two opposing and contradictory decision models which finally ended in April 2008 with the removal of the Child Protection Register. I had first pointed out this contradiction in 2003 in my earlier work on "Risk and the Framework for Assessment" (See Bibliography). Nevertheless, the piece shows how certain types of "Game Pattern" dominate in social work practice.
Future contingents are Outcomes. The Outcomes which you see right now (baseline outcomes) are the outcomes which, as in a sea battle which has occurred, were once future but are now present. Maybe, 3 years ago you perceived that a child was under-stimulated, not shown emotional warmth and no parental interest was shown in terms of explanation and guidance. And now the child is aged 5 it is hard to handle and is showing distinct signs of "Oppositional Defiance Disorder". So, given what you see right now in terms of the child's cognitive ability, its displays of negative emotion and behaviour, its lack of understanding of itself and others, its negative responses to guidance and boundaries - what do you think will be the outcome in 5 years time if nothing is done? And what are you and the family going to do about it if that outcome seems unacceptable?
Notice the demand on knowledge and skills in asking these questions. Not only do you need a very good grasp of child development and theoretical knowledge (you can't predict the future from facts alone -see Attachment Theory) but you also need a very good knowledge of intervention methods and techniques whether or not it is actually going to be you that does the interventions. You need to know this information in order to be able to select and commission the right intervention(s) (Attachment again) and evaluate results.
The button "UP" will take you back to the Risk Dimensions page and Judgement will take you to a page where we consider the meaning of Judgement, Opinion and Judgementalism.
See also pages on Judgement and Judgementalism ; Risk Dimensions ; Outcome formulation and Decision-Making.
