Posts Tagged ‘variables’

The Latent Variable

In 1904 Charles Spearman published a research paper that set the foundation of latent variable modelling. Introducing the concept of general intelligence or ‘g’ Spearman statistical contribution showed how to measure a latent variable which could not be observed.

Latent variables refer to a small number of hidden variables that impact and are the cause of variables that can be observed and recorded. Human behaviour is extremely complex and governed by more than one latent variable. For example, job performance can be linked to g, but a latent variable of social avoidance can lead to social anxiety.

Latent variable models rescale the variables enabling them to be interpreted in regards to correlations between what is observed and what can’t be observed. However care needs to be taken when interpreting the relations between individual observed variables and what psychological meaning can be attributed to those factors.

In statistical analysis of Latent Variables, there are a number of methods for estimating the (alpha) values, the most suitable is called maximum likelihood, but other methods have their purposes and produce similar results. However, most computer methods of statistical analysis, such as SPSS load the output of values in tables or loading matrix.

The main reasons that people conduct latent variable analysis is firstly to understand the relationship between different observed variable and the other is in regards to statistical procedures to calculate factor scores, for example using the regression method.

In organisational life cognitive ability is measured using pen and paper tests (or online equivalents) because of the importance of g in many areas of everyday functioning. For example, the ability to achieve academically, the likelihood of divorce and of involvement in criminal behaviour have all been linked to cognitive ability. In many circles it is believed that in measuring future job performance cognitive ability is the best predictor of job performance. However, g measures the maximum potential that a person may have and there is no agreed definition for intelligence.

However, in recent years, research has disputed this fact, and there has been an increase in the use of aptitude testing which seek out latent variables which could point to as yet unrecognised or untapped potential within an individual. In additional personality tests, which measure typical behaviour offer a glimpse in to behavioural traits, which have more relevance in regards to cultural fit, motivation, attitude and the ability to adapt and flex which are essential in todays market place.

Many Personality and Aptitude test specifically measure latent variables that are necessary in order for someone to be able to perform a job well. Research and statistical evidence which you can read online and on UK TV with a VPN  has produced a number of tests that are able to recognise a latent variable for being able to learn languages proficiently; even in people who have had no language training. Other tests measure whether someone is a super responder, that is they have the ability to recognise faces, a skill which was used to great affect in identifying the often hoodie wearing perpetrators of the London riots in 2011.

Very little is understood about what latent aptitude variables are necessary to perform specific job tasks within the organisation. Organisations build talent management programmes on ‘leadership’ talent often employing personality and cognitive ability tests to identify high potential individuals in their organisation. But these programmes very often miss 80% – 90% of the employee population.

Does this mean that these people lack talent? The challenge, and future research on talent needs to explore latent variables in order to understand what aptitudes led to favourable job performance. The need is for organisations to find an efficient way to identify the latent talent that exists in their entire employee population rather than restricting ‘talent’ to those whose talent can be measured by the current raft of tests and statistical validity today.

Citations

Helen Caldwell – Watching BBC iPlayer in Spain

facebooktwittergoogle_plusredditpinterestlinkedinmailby feather