Conveying Probability

How to describe the likelihood something is true or will occur

Types of Probability

One of the most important parts of an intelligence product is the assessment or assessments it contains.

An assessment is a judgement of:

  • How likely a something is to be true; or

  • How likely an event is to occur

An assessment may be used by the customer to inform their decision on what their next action should be. There are three types of probability that can be used to qualify an assessment. These are (in descending order of accuracy):

  • Specific Probability;

  • Judged Probability; and

  • Unjudgeable Probability

Specific Probability

Specific probability is the most accurate type of probability, and should be used whenever the exact probability of something is known. For example:

  • “There is a 50% probability of this coin landing on heads”;

  • “The probability of this card being the Queen of Hearts is 1 in 52”; or

  • “It will get dark tonight”*

*Note: Stating something will definitely happen or is definitely true in an assessment should be avoided unless you are absolutely 100% certain of it - and even then, it is not advisable!

Judged Probability

If there is not enough evidence to arrive at a specific probability, but there is enough evidence to make a judgement, judged probability should be used.

When using judged probability, analysts should avoid using equivocal modal language and instead use terms from The Probability Yardstick.

Equivocal Language

If I were to say to you “It might rain this afternoon”, how likely do you think it is that it would rain in percentage terms? 10%? 20%? 70%? In truth, if I were to ask this same question to a group of 20 people, it is likely I would receive 20 different answers.

It is highly likely that your customer's interpretation of the level of probability that "might" represents is different to yours - for instance, they may thing it represents a 20% likelihood, while you may think it refers to a 40% likelihood. If this is the case, the accuracy of any assessments they receive from you that use the term "might" will reduce considerably.

The Probability Yardstick

To avoid the potential confusion that can be caused when using equivocal language, we can use The Probability Yardstick. This is a scale which applies relative value to probabilistic terms.

You will notice there are deliberate buffers of 5% between the percentages of probability assigned to each term. These are there to reduce ambiguity and confusion in the mind of the analyst and customer.

There are no terms which convey absolute certainty. This is deliberate - when judging probability the most concrete assessment you should offer is that something is "almost certain" or "highly unlikely".

If terms from the Probability Yardstick are used in a product, it should contain a tabular representation of it similar to the example above. Neither the terms you use or the relative values you apply to them have to be the same as in this example, but whichever terms and values you use must be clearly indicated in your tabular representation.

Examples of assessments using terms from the Probability Yardstick include:

  • ”It is unlikely the incumbent will win the election”;

  • ”There is a realistic possibility the enemy will deploy fighter jets”; and

  • “It is almost certain the business will make a profit this year”

Avoid Reintroducing Ambiguity!

When conveying probability using specific or judged probability, ensure you use an unequivocally certain modal verb after the term used to convey probability, for example:

“It is highly likely that Boko Haram will attempt to enter the village”

Using an equivocal modal verb after the term used to convey probability risks making the assessment just as ambiguous as it would have been had you not used a specific term or term from the probability yardstick in the first place, for example:

“It is highly likely that Boko Haram might attempt to enter the village”

Further examples include writing:

  • “It is likely the business is overvalued” instead of “It is likely the business may be overvalued”;

  • “There is a 50% chance the enemy will attempt to cross the river using Bridge X” instead of “There is a 50% chance the enemy might attempt to cross the river using Bridge X”;

  • “It is almost certain Suspect X has left the country” instead of “It is almost certain Suspect X could have left the country”.

Unjudgeable Probability

If there is not enough evidence to judge probability, it is permissible to use modal verbs which convey equivocal probability, such as “might”. In this case, the assessment must be caveated with words to the effect of:

“…but there is not enough evidence to judge the probability of this”

For instance:

  • “The opposition may call for a vote of no confidence, but there is not enough evidence to judge the probability of this”; or

  • "The river might be passable, but there is not enough evidence to judge the probability of this”.

Found this interesting? Feel free to follow me on LinkedIn for more intelligence content

Reply

or to participate.