Some tests turn a candidates RAW SCORE into a PERCENTILE

( sometimes known as a PERCENTAGE RANK or CENTILE ). This is not the same as a PERCENTAGE.


Let's say the maximum score you can get on a particular test in which there are right or wrong answers is 50 and PERSON A achieved 25. That means he got 50% of the questions right. This is his or her percentage of right answers.


A percentile tells you something different. You ( or the computer ) looks up that score of 25 in the test’s norm tables. This indicates that 70% of the people who took the test to create the norms got less than a score of 25. Person A’s PERCENTILE rank is therefore 7; he or she did better than 70 % of the sample.




Reasonably stable preferred ways of acting or thinking.


Personality tests are used very widely, particularly in development and in the recruitment of senior staff and for jobs where interpersonal interaction is crucial – sales people for instance. Increasingly, organisations are looking at how individual personalities work together – in teams, in departments and in  project groups, and how individual personalities fit with organisational culture.


Note the use of the word ‘preferred’ in our definition. Personality tests measure typical behaviour, but just because you prefer to act in a rule-based style doesn’t mean that you won’t suddenly break the rules in different circumstances. Common sense tells us that people constantly surprise us. Personality profile are not exhaustive strait jackets but predict how a person typically acts.


Age discrimination has highlighted that while personalities are reasonably stable, they do change over time and older people become “ more like themselves” and more different from each other as they get older.


There are a huge number of models of human personality and approaches to measuring it ranging from factor analytic scales ( which produce comparisons with other people on a set of factor scales ); type questionnaires ( where you are matched against certain types of behaviour ) and projective measures ( where you react to a stimulus, such as the famous inkblots )


What we can say is that personality is crucial in the successful carrying out of job roles .




The personal characteristics required to do a job. This should be used in recruitment to structure the recruitment process and to ensure that each characteristic is measured in the best possible way. The Person Specification is usually derived from the JOB ANALYSIS. See here for more information on this process.


The crucial issue here is that you can’t decide which test is relevant to your needs until you’ve done some sort of analysis of what success at a job entails and what sort of person you’re looking for to do that job.




A timed test  designed to see how many questions can answer correctly. Thus power tests have right and wrong answers ( as opposed to, say, personality tests where you’re often reacting to statements rather than answering questions)




This is the most important aspect of tests used in selection. It measures whether the test is predicting future performance – good or bad – in a job. If a test does do this then it saves huge amounts of wasted time, money and the widely experienced grief that everyone undergoes in the cases of obviously wrong recruitment decisions.


Unfortunately you need some time to work out whether a test ( or other method of selection ) is predicting accurately. In developing a test, researchers have to follow recruits, sometimes for years, to get the data to analyse predictive validity. Test users can set up exercises, or use external experts to evaluate and improve their selection processes if the techniques they’re using do not provide adequate data


Where predictive validity is reported it is often, for technical reasons, understated.


Where it’s not feasible to establish PREDICTIVE VALIDITY, test developers and users often use CONCURRENT VALIDITY.








In the selection, development and management of people we’re often trying to measure the intangible – thoughts, feelings, values, knowledge – and relate it to observable behaviour.


Psychometrics is the study of the measurement of individual psychological differences. It forms part of the discipline of psychology but increasingly draws on statistics, computer sciences and business theory among other academic areas.


Typically psychometrics looks at:


  • How you can create ITEMS that validly and fairly  measure an area ( including ensuring that they’re not measuring something different and that the items are not easy to manipulate by candidates )
  • Building models of what should and can be measure (for instance different models of personality )
  •  Hugely sophisticated statistical techniques to ensure that measurement is both accurate and useful.


Psychometrics is used in setting school and university tests, professional qualifications, medical assessments and tests for patients suffering from, for instance, Alzheimer’s, autism, schizophrenia and strokes.


You’ll usually see the word “ Psychometrics “ associated with the word “test” but any measure of human characteristics can be made psychometric by the use of the discipline’s techniques.  For instance increasing use of objective scoring methods, structured questions and trained interviewers have made structured interviews considerably more predictive than old-style unstructured interviews.






Another standard scoring system. Whereas the DECILE divides RAW SCORES into 10 categories each containing 10% of the scores, QUARTILES divide then into 4 groups each containing 25% of the scores.