Decision-making, Predictability and Scientific Validity: New trends for new opportunities

Decision-making, Predictability and Scientific Validity: New trends for new opportunities

In this permanent necessity to have to make the right decisions, to be able to anticipate behaviours and to be assisted by secure solutions, the HR world is more than ever open to innovation. Research does not cease to multiply, nor do solutions. Among all these new trends, which ones really contribute to our professions and through which opportunities can we optimise our approaches and more broadly the support we offer to organisations and their HR managers? PerformanSe gives you the point of view of recognised specialists and experts, in order to better grasp the drivers on which we can act today and those which will make tomorrow.

Discover the opinions of :

  • Jacques JUHEL,Director of the Psychology, Cognition and Communication Research Centre at the University of Rennes 2, Member of the PerformanSe scientific committee

  • Jean PRALONG, “New Careers” Chairholder, NEOMA Business School

  • Alexandra Didry, Doctor in the Social Psychology of Work and Organisations,PerformanSe R&D Head

The opinion of Jacques Juhel

Taking decisions becomes more and more difficult to grasp in our world of instantaneity, to you, what are the best levers to not get it wrong?

“The scientific theories of individual decision-making (for example in a context of staff recruitment, as a managerial activity in a specific organisational context etc.) have broadly evolved over the last century, with a still very wide gap between what cognitivist research has taught us on the decision-making process and decision-making in a professional context, adapted to the requirements of the field. Decision-making was firstly understood as an ideally rational activity enabling one to reach set goals, in a given context and on the basis of clearly explicit rational principles. This form of rationality, sometimes referred to as ‘substantive’ was succeeded by a rationality ‘limited’ by the environment. Developed in the forties by Herbert Simon, this concept of limited rationality makes rational decisionmaking an approximation under constraints, as a function of time and the information the decision-maker has as well as of his cognitive resources which limit the analysis of information and lead to the selection of a satisfactory choice instead of an optimal choice. These are many elements with an influence which varies according to the contexts and it is essential to understand that their action would therefore disrupt rational decision-making. In a recruitment situation for example, too superficial an analysis of information about the job, about the organisational environment and about its values, insufficiently explicit selection criteria, an assessment of skills hardly linked to the functional skills of the job or with crossdisciplinary skills which are likely to be employed in new activities or responsibilities, the failure to take into consideration certain cognitive biases etc. increase the level of uncertainty inherent to limited rationality. Furthermore, a lot of research carried out over the past thirty years has taught us that individuals, when faced with urgency and reduced available information, may take apparently irrational decisions in the moment. Antonio Damasio’s research for example has demonstrated the existence of a very close connection between emotion and decision-making. Daniel Kahneman’s research, which distinguishes between two systems of thought, one quick and intuitive, the other slow and rational, also make one think that intuition (heuristics, judgement shortcuts etc.) plays an important role in the decision-making process. Better understanding of ‘intuitive’ decision-making and the mechanisms which underlie it is therefore necessary in order to better appraise and potentially better control the effects in decision-making practices.”

The term predictability seems to be in fashion. How would you define it, what level of reliability can we give to this notion and what precautions must we take?

“The meaning the psychologist immediately thinks of when the question is predictability is that of digital forecasting, i.e. forecasting the future state of a system based on a mathematical modelling of its behaviour. What can be mathematically modelled and the evolution of which over time can be analysed would therefore be predictive. For example, if one supposes that the attribute measured in an individual when administering a test is stable in time and that one has an estimate of the reliability of measurements made with the test, the individual’s score when the test is administered a second time (his ranking within the set of individuals who are comparable to him) may be predicted based on the score observed the first time the test was administered. As the aim of recruitment is to anticipate an applicant’s ability to fulfil a future task, the predictive validity of a selection procedure may be assessed by estimating the relationship between the results obtained by applicants and certain criteria measured later on in those who have been recruited (for example their effectiveness at work, their organisational commitment etc.).

It seems that the term predictive recruitment used today by consultants in reference to a Big Data type approach and to the new tools which accompany it, exceeds the traditional definition of the predictive validity of a recruitment method. Predictive recruitment consists in identifying in the considerable volume of computerised data to which companies may have access, significant predictors, both from the point of view of goals pursued and in the statistical sense of the term, of the accomplishment of certain events linked to organisational operation, for example desired behaviours and what one wants to promote within the organisations. The benefit of this approach, when it is part of a collective rationale, may be understood. The analysis of massive data from many sources contributes to a scientific management of human resources based on multiple criteria and therefore better informed. One can nonetheless wonder whether this approach, when it is part of an individual rationale, does not present the dual risk of schematic and unnatural modelling of performance at work and the progressive replacement of psychological practices by decision-making algorithms.”

The scientific validity of tests and their theoretical models is essential. Do you see any recent research which could improve what already exists?

“The systems developed to measure cognitive and behavioural skills must satisfy certain psychometric as well as economic, practical and legal requirements. Concerning systems for gathering observations, the progress made over recent years owes a lot to the development of IT and new technologies. We use the computer more and more to gather questionnaire data or to administer psychological tests, the organisation’s intranet to implement approaches in an integrated system (pre-screening, computerised testing etc.) and tactile tablets to gather intensive longitudinal data (multiple times a day, sometimes over several weeks). The exponential increase in computer calculation possibilities enables us in parallel to employ sophisticated psychometric and statistical modelling techniques in the field with results which can be rapidly exploited. In the cognitive field ‘adapted test-taking’ procedures enable us to reduce the administration length of a test by very quickly targeting the individual’s level of efficiency. Questionnaire assessment can benefit from so-called ‘functional’ methods enabling identification of the individual’s response strategy conditionally to the attributes measured, and his potential inconsistencies during the taking of the test. Some of these new systems bear witness to a very accentuated orientation towards dynamic assessment and the multilevel observation of conducts and their course, thereby improving the ecological validity of the observations made. This is the case in particular with daily diary and Ecological Momentary Assessment methods, or the Experience Sampling Method which are starting to be used in work psychology and organisations to study for example satisfaction and commitment to work and the dynamic of the effects of the organisational context on behaviour at work or of interactions within teams. We should certainly expect important progress in these different fields of research and application.”

Jacques JUHEL,

Director of the Psychology, Cognition and Communication Research Centre at the University of Rennes 2, Member of the PerformanSe scientific committee

The opinion of Jean Pralong

Taking decisions becomes more and more difficult to grasp in our world of instantaneity, to you, what are the best levers to not get it wrong?

“There are two important elements. Decision-making must be underpinned by reliable data obviously, but above all must be based on a reliable theoretical model, i.e. a scientifically validated model. The importance of the model is all the more topical with the appearance of the notion of ‘big data’: it is not because we compile thousands of items of data that we are going to bring out more reliable results. The truth does not lie in the data alone, but in the way the data is interpreted. This is why we need a strong theoretical model, which will give meaning to the relationships between data.”

The term predictability seems to be in fashion. How would you define it, what level of reliability can we give to this notion and what precautions must we take?

“The notions of ‘big data’ and ‘predictability’ are often confused. Big data is exploratory, we base ourselves on thousands of items of data to find statistical links. These correlations exist but can have no meaning: for example, the correlation between eye colour and baccalaureate results is strong but there is no causality link between this data. Predictive recruitment, on the contrary, is confirmatory. It is based on validated theoretical models, which data confirms. We do not look for correlations but causality links, and we will be able to explain why people who have such and such characteristic will, for example, perform better than others. In the use of prediction, the real problem concerns the contextualisation of the model. There is no single success model, which would be the only way to perform well in a job. The future of prediction will therefore be to process data by typological analysis, to find different forms of success via different behaviour, and to thereby predict success profiles. Today, in recruitment, HR staff bases themselves on skills and personality tests, but very little on ability tests. Yes these tests have an important predictive value in performance at work. It would be interesting to incorporate this type of data in the analysis. And just as there are behavioural profiles which are more or less adapted to types of success, to see if there are more or less adapted forms of intelligence which are complementary, and therefore of benefit to have in one’s teams.”

The scientific validity of tests and their theoretical models is essential. Do you see any recent research which could improve what already exists?

“I think that the current personality approach is maybe not the most relevant and would deserve to be renewed, but thinking about alternative models… The aim is doubtless not to throw away everything that we call ‘personality’ but to maybe look at it differently. For example, to review the way in which we conceive of predictive recruitment or the analysis of potential by a behavioural approach, based on this renewed idea of what personality could be, by measuring it and theorising it differently.”

Jean Pralong,

“New Careers” Chairholder, NEOMA Business School

The opinion of Alexandra Didry

“‘There is nothing more practical than a good theory’… This quote from Kurt Lewin is more than 60 years old and has never been more topical. It enables us to illustrate a strong position: fundamental science and applied science are linked and walk on the same path!

Theoreticians have the duty to create theories dedicated to the resolution of social and practical problems. Practitioners and researchers in applied psychology must use this scientific theory. It’s this complementarity which we must bring to life between theoretical models and scientific research in order to progress. But not only, it is important to connect the use of scientific theory with a beam of reliable proof. The idea is to bring more methodology and rigour to practices by linking them to facts and to tangible evidence. We can then use the principle of evidence-based practice, i.e. encourage practitioners, HR staff and consultants to base themselves on the best evidence from scientific research and field experience, taking into account preferences and societal trends.

The question of data and its proper use then arises. By basing ourselves on the best research, models and conceptualisations, we must properly frame what we are looking for to be able to give useful meaning to data and to thereby create close interactions with fundamental research. We must however remain watchful. We can all make data talk to make predictions, but to avoid any deviance in interpretation or methodology, rigour is required, simply that of science and research. The most revealing example is the current trend around ‘big data’. In our professions, wouldn’t it be more relevant to make our small data talk, as it carries lots of information, instead of launching into big data, where the data is on the other hand less rich with information (in essence the definition of big data). We therefore very naturally reach questions about predictability, which is very much in fashion today. Lots of solutions are being launched on the market, doubtless announcing too much to sell. Therefore let us remain pragmatic and base ourselves on precise data, evidence and models because what we must avoid is rough modelling at an individual level, which would bias any result therefore any prediction. For example, when it comes to measuring performance, the results can be relevant in collective terms, from an advanced HR practices and knowledge point of view, but not necessarily in individual terms, which are much more complex to grasp. So yes to HR scientific management, to the dynamic integration of fundamental science in practice / experience and to the opening of new mathematical methods. But let us avoid being seduced by solutions which are marketed all too well on paper but which are hardly tried and tested in field reality, and let us focus our energy on sources, methods and experience, while having a single stake, the human stake.”

Alexandra Didry,  Doctor in the Social Psychology of Work and Organisations, PerformanSe R&D Head

>> Watch the online version of our magazine ECHANGE
to access all of these items


Share This