LOCUS – Using experimentation to assess the scale of use of CBRN weapons

Problem: The problem of when and whether chemical, biological, radiological or nuclear (CBRN) weapons will be used is a fundamental one in CBRN defence, since it determines the extent of the requirement for such a defensive system.  The fact that such weapons have only been used in a limited number of conflicts, even when states possessed the means to use them, implies that the decision is by no means a foregone conclusion.  It is apparent that in many cases the decision as to whether or not to use these weapons has been only partly a military one.  The profound psychological and political impact of the use of such weapons means that the decision to use them is traditionally made at the very highest political level.  In fact, the use of such weapons is usually the prerogative of the head of state or government.

In many studies carried out for the MOD to examine the likely use of these kinds of weapons, the assumption was made that political authorisation was available from the start.  The planning from that point on was carried out with that assumption having been made, and therefore the decisions as to where and when to use the weapons was based solely upon military requirements and the need for an optimal military plan.  This neglects the possibility that the use of these weapons would actually be authorised partially and at different points during the campaign, since usage of these weapons tends to flow from political rather than military motivations.

Approach: The difficulty in understanding the likelihood of use of CBRN weapons is therefore linked to the need to somehow represent the political leadership of a hostile regime.  If the motives for usage of such weapons are largely political, a simple military appreciation of the problem will not suffice.  To overcome this difficulty this study has combined two techniques, the use of detailed personality briefings (or ‘wrappers’) to enable role-players to represent the high level political leadership of states of interest, and the use of controlled card-based experimental decision gaming to represent the effect of different prompts upon the likelihood of political authorisation being given.

Methodology: The personality briefings set out details on the history and family background of the leader, the political situation both within the country and internationally, and the leader’s own political ideology, motivations and individual leadership style.  The personality aspects of this were based upon the work of Prof Margaret Hermann (Hermann, 1999), an expert on international relations, who categorised different foreign leaders according to their openness to information and their belief in their ability to influence events.

A personality instrument was also used to compare the risk-taking propensity of players prior to their exposure to the ‘wrappers’.   This allowed the study to determine whether the ‘wrappers’ or player’s underlying personalities were driving behaviour.  The results showed clearly that statistically significant differences were obtained between different leader ‘wrappers’ but no effect (in fact a statistically insignificant negative effect) was observed from the players’ underlying personalities.

The card-based decision games were based upon a technique previously developed to build models of decision-making using the presentation of a number of snippets of information on a series of information cards.  This method has been applied elsewhere and is more fully described in (Medhurst et al 2008, Medhurst, Stanton and Berry 2009).  It allows randomized, controlled and balanced experiments to be carried out on decision-making behaviour.

The method also allows quantitative models to be developed of the behaviour of decision makers.  It generates large amounts of information from a single day’s gaming, which in turn allows more powerful statistical techniques such as logistic regression to be used to analyse the results.  The structure of the games also allows a balanced trial design using Latin or Graeco-Latin squares to determine the order of play of individual serials and the presence or absence of particular conditions. Because all players play the same serials it is also possible to allow different leaderships to play what is essentially the same game, making direct comparisons between the responses of different leaderships possible.  This high level of control in an experimental situation was key to the ability to build testable statistical models of behaviour.

Outcome: The detailed results of the study are classified, but the approach was able to identify those kinds of events most likely to trigger use of CBRN weapons, and the kind of targets against which these weapons were used. It also allowed us to construct a quantitative statistical model of the way in which different types of events and external pressure would influence both the willingness of the hostile leadership to concede and their likelihood of using these weapons.  This model was useful in identifying optimal strategies for coercion and in comparing the effectiveness of diplomatic and military approaches.

Benefits: This study demonstrated that the card-based decision gaming technique could generate highly effective statistical models of behaviour even in situations involving complex, high-level political decisions.  The ‘wrapper’ method was invaluable in allowing players to represent the leaders of interest, and the use of the personality instrument allowed the study to demonstrate statistically that behaviour was driven by the wrappers rather than the variation in the players’ own personalities.

Contact: John Medhurst