A good test should produce consistently similar results when administered to the same group of people over different time periods. For example, the reliability coefficients for the tests that we use and recommend range from .86 to .92. In every day language this means an 86% to 92% test to re-test accuracy. This is considered to be a very high correlation between test and re-test results.

First, remember, tests are a product. Just like other products there are branded, generic and own-label tests. New ones are being introduced all the time - some become dated, some fall out of fashion and, like other products, some just never make it. Just because a test has worked well elsewhere there is no guarantee that it will be of value to your organisation. So don't base your decision solely on a sales presentation. Do your own market research and be an informed consumer.

We recommend that, if possible, you validate all tests within your own organisation. You can do this by conducting what is described as a Concurrent Validation Study. Conducting a Concurrent Validation Study will also provide you with localised ‘norms' – benchmarks against which future applicants can be considered. A Concurrent Validation Study is carried out by administering tests to a cross section of a workforce who have, as objectively as possible, been ranked in order of job competency or against specific criteria. Results are then analysed to identify the relationship between test results and performance levels. By way of example, one recent Validation Study conducted by one of Britain's largest assurance companies, established that certain personality traits correlated highly with success or non-success as a Financial Adviser. Test results indicating that an individual would be in the top half of the sales force - were 88% correct. Test results indicating that an individual would be in the lower half of the sales force - were 95% correct. These results are similar, in terms of predictive validity, to many other Concurrent Validation Studies. Please remember – we vigorously stress that no selection, development or training decisions should be made solely on the basis of a test result.

This takes longer but is probably even more effective than a Concurrent Validation Study. However, you can only do this if you are hiring a reasonably large number of people – to do the same job – over a short period of time. In a Predictive Validation you administer tests as part of your normal selection procedure but you do not disclose the results to anyone involved in the selection, training or day to day management of those who are hired. Over a period of time you compare test results against actual job performance criteria. This will identify any positive/negative correlation and so assess the predictive value of the test questionnaire. One slight disadvantage of this method is that you cannot give chosen candidates ‘feedback' until you have completed your predictive validation study. All authors and sellers of tests should be willing to help you conduct your concurrent or predictive validation study. They will normally do this for a reduced cost. They may even consider doing it for free if you are a big enough fish! Remember there is nothing mystical about tests, they are a ‘product' like every thing else.

These are the Control Groups against which you can compare people. Norms are sets of scores derived from a clearly defined sample. The importance of norms is that it gives a test score (raw score) a meaning, making accurate comparisons possible. Remember, your test score (raw score) is a constant. Your norm score will vary depending upon the norm group against which you are being compared. For example: if you are a 6' 2" male adult you would have a high norm score for height when compared against a cross section of the UK/USA adult male or female population... However, you would only have an average norm score for height if you were compared against African Watusi male adults (who average around 7'!) We recommend that you create your own norms wherever possible. However, if you do not have sufficient numbers to create your own norm - or if you are in the process of change and are looking for people who are different from your current workforce - the authors of this Guide can recommend an appropriate occupational norm group. These norm groups include: (This range is constantly expanding to include other occupational groups.) UK general population norms. Sub-groups – demographic, socio-economic, working/non-working, age range and sex, etc. Occupational norms – primarily white collar. These include senior, middle and junior management and supervisory. Financial services (high net worth sales, direct sales, home business, IFA and tied agents). Call centre – both reactive and pro-active. Retail – white goods, fashion, food, video rental, etc. Catering – managerial, waiting and production staff.

What if you spend money on a validation study and the results are inconclusive or establish that there is no correlation between areas measured and job performance? Isn't that money wasted? No, it will probably be the best money you have ever spent ! Its better to invest in a validation study than to simply accept ‘sales speak' and use a test that is valueless. For example, we know of one major supermarket chain that wasted much money and time testing checkout assistants for numeracy skills. Think about it! They did that because someone did not conduct a validation study. After taking our advice, they conducted a concurrent validation study and now invest money on visual comprehension, dexterity and colour recognition testing.

Beware of the sales person who offers you the ‘Sales Prospect Validation Study', aka the ‘Barnum Bait Validation Study'. This is where they offer you one free test to do personally. They then try to get you to base your decision to buy their tests on the accuracy of your personal test report. You can imagine how good that report is likely to be! Barnum