It is a simple fact that people in positions of authority over employees have to make decisions: about whom to hire, who to fire, whom to promote. But it is a much more complex matter to determine what constitutes making a decision of this sort non-arbitrarily and fairly. Because of the complexities involved, serious and responsible decision-makers seek further information in making these personnel decisions, either to supplement their own judgment base upon letters of recommendation or actual observation of the employee, or entirely to supplant these. They use tests which promise objectivity and fairness, or “experts” who promise the same thing. On the other hand, irresponsible decision-makers tend to “pass the buck” by using the same sources in order not to have to take responsibility for difficult decisions. Either way, there is an increasing tendency for those making personnel decisions to rely on psychological tests either administered by themselves, or by outside experts.
The use of these tests raises the concern about the rights to privacy of employees who are subjected to these tests either as a condition for being hired in the first place or for remaining in their jobs or being promoted. Unfortunately, given the complexity of this issue, we are not in a position to make any blanket public policy pronouncements on this issue stronger than advising employers to exercise restraint in the use of such tests. We can, however, advise employers that tests they buy off the shelf, and administer themselves, are not very likely to give the information they desire, and this puts us in the position to recommend against their use. Given the employees’ rights to privacy that must be weighted against the employer’s rights to obtain information about the employees, the argument below will suggest that the balance tips in favour of the employees’ rights at least in the case where these tests are administered by employers who do not know very much about psychometry. For, as we argue below, the employer simply will not be getting information that is reliable under these conditions.
We are certainly not in a position to recommend any legislation on this issue But our lack of a simple, elegant, easily-applicable position on this issue is not due to lack of clarity about the civil liberties issues involved. These are already articulated in positions the BCCLA has taken on other issues of a similar nature such as AIDS testing, drug testing in the workplace, and the use of such questionable devices or methods such as the polygraph or graphology/graphoanalysis in the workplace. This brief will merely apply, in Section I, the principles this Association has worked out on these issues to the question of psychological testing.
The difficulties we face come, first, from the wide range of things that coming under the heading psychological tests. These items range from things even more unreliable than the polygraph to things that might be even more reliable than good old common sense coupled with keen observation. So, Section II will skim over this area. Second, even a reliable tool is downright dangerous in the hands of the incompetent; and therefore we have to consider the range of competence of the testers. And the range here is at least as wide as it is amongst tests. So, Section III will have something to say about this.
Having waved our arms at the empirical questions, we get down to the important business in Section IV. What this brief contains is a series of recommendations on how to deal, on a case-by-case basis, with those employees who approach the Association for advice or help when confronted with the request or demand that they submit themselves for such testing.
I. We have two prima facie reasons to support anyone concerned about being compelled to take a psychological test in the workplace. The first concern has to do with privacy. The second concern has to do with the fairness of a decision concerning the employee’s future career which is based on psychological tests.
Our organization has taken the position that an employee has a right to privacy in at least the respect of not being coerced to reveal information about him- or herself that is not directly relevant to the carrying out of his or her job. Thus, the right to privacy is not indefeasible; and can be outweighed by three broad types of considerations: having to do with (a) the responsibility of the employer to employees other than the one(s) being tested; (b) the employer’s responsibility to the public at large; and (c) the employer’s right to select the most competent person for the job in order to maximize his or her own profits.
With respect to (a) and (b), if we maintain that the employer has the right, or even the duty in some cases, to monitor an employee’s performance in these matters, then we could have no objection to the employer getting relevant evidence before an employee begins the job about how well an employee is likely to perform. In fact, if it were possible to get this evidence, we might think that, e.g., the employer has a duty to determine in advance whether an employee is likely to be prejudiced against minority groups before hiring that person as a personnel manager. Or, a prospective foreman who is likely to be sexist should not be foisted upon an all-female assembly line. Or, third, a person who has difficulties in explaining simple material would not be the best candidate for the job as receptionist in a large office where members of the public will be asking for directions to various departments. Thus, certain personality tests and attitude scales may seem, on the face of it, appropriate for hiring, transfer or promotion decisions.
Regarding (c), an employer who wishes to maximize his profits from his employees has the right to determine whether an employee possesses the requisite skills and capacities to be effective in the job. Thus, vocational preference tests, aptitude tests, and perhaps some personality profiles may be relevant to determining these questions. Again, if the employer has the right to monitor the effectiveness of an employee currently holding a job, we cannot see a principled reason to object to the employer seeking to gain information on this in advance.
These three categories of cases provide examples where the employee’s prima facie rights to privacy can be outweighed in those circumstances where the information sought by the test is important and directly relevant to the job. But this does not settle the matter. The question then turns on whether there are psychological tests that are capable of providing this information. An employee ought not to be expected to give up his or her right to privacy when the test in question really does not provide the relevant information anyway. This latter question will be taken up in Sections II and III.
But there is still the second consideration mentioned at the beginning of this section, that of the fairness of the evaluation procedure(s) that an employer may use to make these decisions. The case has been made that psychological tests with a high degree of reliability and validity, administered and scored by competent people, may in fact be fairer than the subjective evaluations of performance given by bosses or personnel officers. An employee may find it to be in his or her interests, then, to sit a test rather than, or in addition to, relying solely on such subjective impressions. And, since these other evaluations also involve loss of privacy, the employee is better off with a test than relying solely on these other evaluations. This argument has some weight, when the conclusion of it is phrased hypothetically: If the test is in fact a better measurement than the alternatives, then an employee’s interests may be best served by agreeing to take the tests. However, the burden in now on determining whether the antecedent of this conditional can be fulfilled. This is the question that we return to in Sections II and III.
But first we should clear away five bad arguments for extreme positions on the psychological testing question: three that purport to show that psychological testing should never be allowed, and two that purport to show that the BCCLA ought never to intervene on behalf of an employee who does not wish to take such a test.
The arguments for a blanket rejection of psychological testing in the workplace take three related forms. The first holds that such testing is discriminatory against minority groups and women. But we must be careful with the word discriminatory here. In one sense of the term, of course test are discriminatory. That is their function. The important question is whether they unfairly or unjustly discriminate. They do, in our view, if they assign a lower ranking simply in virtue of the subject being a woman or member of a minority group, when the stated purpose of the test is to measure something else. Now, first of all, whereas some tests are biased in this sense, not all are; so a blanket condemnation of testing cannot be justified on these grounds. As will be argued in Section IV, we will need the expertise of a psychometrician to determine which tests are biased in this way before moving against them. What we must get expert advice on is the more general question of what degree of reliability and validity the test has for that which it purports to measure.
Second, it may be held that there are some traits which are correlated with sex or minority group status, but which are conceivably also related to legitimate job-related concerns. Whether this is the case, and what position we ought to take on it if it is, is a much wider question than can be considered here. Suffice it to say that this is not a problem with testing for such factors, but whether these factors should be considered in the first place. So we are glad to leave this question aside for present purposes. Again, in Section II we will return to what is relevant in this line, the question about the validity and reliability of the particular test under consideration.
The third objection is intertwined with the second. The worry is that there may be certain relevant traits which, because of environmental factors less euphemistically, social prejudices are indeed correlated with minority group or sex status. Lack of educational or other social advantages are the sort of thing we have in mind here. But some people mistakenly hold that these traits are innately determined rather than the result of social factors. And to hold that these traits are innate invites one to draw the conclusion that they are fixed and not changeable; and therefore a person who scores badly on a test measuring these traits cannot learn on the job, as he or she could if he or she merely lacked them due to an impoverished environment. Therefore, the use of testing in the workplace serves as a reinforcement of false racial or sexual stereotypes. The short answer to this complex problem is that such views are not entailed by the test results themselves, but the interpretation of them. This is one reason why we advocate in Section III an examination of the credentials of the particular testers involved in a particular case, rather than a blanket condemnation of the process which sometimes leads to false and unfair judgements.
We now turn to two arguments suggesting that we move to the other extreme, and do not concern ourselves with the question of psychological testing in the workplace at all.
It might be objected by some that there is an infringement on the rights to individuals to make bargains with one another if the Association were to aid a person who does not want to be subjected to a psychological test. According to this view, justice is best served when individuals, either corporate or human, are left free to bargain all terms of employment without outside interference, including whether or not they will be subjected to psychological tests. But this objection is misguided; since the fact that one of the parties to the (prospective) bargain, in coming to the Association for assistance, is already indicating an unwillingness to agree to it. We are not proposing here any restriction on anyone’s rights to volunteer for such a test, or to ask someone else to take it. We are merely considering aiding someone in the bargaining process which defenders of this view hold in such high esteem.
Associated with the above position is the view that there is no need for the Association to actively take a stand on this matter, since the marketplace looks after such things on its own. Those employers who make what are thought to be unreasonable intrusions on their employees’ privacy will be dealt a slap from the invisible hand of the marketplace. But first of all, there does not seem to be a lot of empirical evidence to back up this view, and what evidence there is presents itself in the best light to those looking at it from the perspective of an armchair. Secondly, there are two moral arguments against leaving things in these invisible hands, even if they did work. While a competent employee who was denied a job unfairly may derive some satisfaction from seeing the company that refused to hire him go belly up fifteen years later, this does nothing to resolve the immediate issue of the injustice done to him now. As well, even if the employer deserves this fate, it is unclear that her employees do. While we may admire the ‘right stuff’ of the captain who goes down with her ship, the decision to take the rest of the crew along with her is far less admirable. So, if it is the case that the marketplace exerts these pressures on an employer to practice fair testing, the kinds of pressures seem to be either too little and too late to ensure justice, and where it exerts enough of a penalty on the unjust, it seems to be, like God, indiscriminate: exercising equally stringent penalties on the just as well. Therefore, we have reason to prefer the remedies of conscious, informed agents such as we will propose later, to the invisible hand of the marketplace.
II. The first thing to note about psychological tests is the plethora of them. For our purposes, we may divide the kinds of tests into five basic types:
(a) Achievement tests
purport to measure some particular knowledge or skill the subject has learned in the past.
(b) Aptitude tests
purport to measure the same things as (a), but with the added assumption that the skill in question is a long-term disposition, and not transitory knowledge of the sort measured in (a). In the history of psychological testing, assumptions about heredity were often made to explain the enduring nature of the trait being measured; but this is not often the case any more. Without such assumptions, however, (b) tends to shade off into (a).
(c) Intelligence tests
or, as they are usually called now, scholastic aptitude tests, attempt to measure a particular kind of aptitude, viz., that related to academic work. Obviously, these have been the battleground for the nature-nurture controversy; and one might conclude from the change in name a concession on the part of the test manufacturers of hereditarist assumptions that there is a fixed, genetically-determined, single factor being measured. There are a number of such tests with a high degree of validity and reliability for some purposes, and our question here centres around the use in the employment situation to which the test is to be put.
(d) Personality tests
run the gamut from tests to determine psychopathy to ones that work just as well on animals as humans to determine traits as poorly defined as friendliness and parsimony. Their use in business is to determine whether an employee has the personality traits thought useful in the job. But the biggest problem with most tests is that there is no way of determining whether the trait being tested for bears any resemblance to the one called by the same name that the employer is searching for.
(e) Vocational preference tests
as the name indicates, attempt to indicate to the subject what his or her preferences in fact are. For this reason, they can be considered the least intrusive for the subject, and the most beneficial as well. There is a lot to be said for asking an employee where he would most like to work in the company, and a few of these tests give more precise results than this. As well, with the not unreasonable assumption that an employee is likely to do best at what he likes doing, it is in the employer’s interests and perhaps in customers’ or fellow employees’ too to try to gain this information. The question left hanging at the end of Section I was how accurately psychological tests measure the various traits and dispositions mentioned above. In order to phrase this question in the right way for our purposes, we will need to look at the two criteria psychologists use as standards for accuracy:
This notion pertains to the question, “What trait, disposition, etc., does the test measure?”. But validity is a quantitative notion, and so we wish to talk about the degree of validity. We are thus answering the question, “To what extent does the test measure what it purports to measure?”. Thus, in what follows, the expression greater degree of validity or its cognates means, roughly, more accurate a measure of the trait in question.
The second criterion has to do with the consistency of the test. Does the same person score roughly the same on the test on two separate occasions? Would two different psychologists assign the same score to the same individual taking the test? Subject to the qualifications below, we should be looking for an answer of yes to these questions when we are looking for a fair test. For, if the employee could be reasonably expected to get a totally different score if the test were administered on another day, or by another tester, we have no reason to think that the particular test result is indicative of the degree to which he or she really possesses the trait being tested for.
These concepts are straightforward enough in the abstract, but knowledge of psychometry and the tests themselves is essential to apply them for our purposes. Two qualifications are essential. First, validity is not a simple property possessed by a test without qualification. It depends upon the particular types of trait which the test is designed to measure. And this will have to be compared with the traits being looked for in the particular testing situation. The traits which are relevant to a particular job may be quite specific: e.g., the sense of gregariousness which is relevant to being a good salesman of fine china religious ornaments to Presbyterians might not be very closely related to a test that identifies this trait by questions such as “When at a party, do you enjoy dancing with a lampshade on your head?”.
Second, reliability is a notoriously difficult concept, in part because some traits do vary in a given individual over time, and one would expect in this case that a reliable test would give that individual different scores at different times.
This raises the problem about estimating the reliability and validity of the kinds of tests listed above. We can make only generalizations here, that admit of many exceptions. In general, personality tests are as a class the least reliable, since the traits they measure are not often clearly defined in the first place. Those that give clear cut operational definitions to the traits measured will also be the most sensitive to the fact that defined traits will vary with the individual as her or she encounters different circumstances. Tests with the highest degrees of reliability and validity, such as the Minnesota Multi-Phasic Personality Inventory (MMPI), are ones that have been validated against the largest number of subjects, and require the greatest amount of skill in interpreting. They are not terribly useful in the workplace. On the other hand, the 16-PF test, which is easy to score and administer, is notoriously vague in its definition of the 16 personality types it purports to identify. And there are thousands of such tests whose reliability and validity coefficients are too low to be measured accurately.
On the other hand, IQ tests such as the Wechsler (WIS) have respectable reliability and validity coefficients, as do some aptitude tests for very specific aptitudes. But there are hundreds of competitors that fail to measure up to these standards, but have a corresponding price reduction and are easier to administer and hence are more widely used.
Vocational preference tests suffer the same problem. The standard for many years, the Kuder, has thousands of imitators. In general, such a test has the promise of validity, given that people’s preferences can be independently checked. But with some tests, they often aren’t, with anything approaching scientific respectability.
As well as the difficulties with the tests themselves, there is the major problem of using a test which has validity and reliability when used for a specific purpose for another purpose in the job market. But this is often what has to be done. If the employer has a need to identify a particular trait relevant to the particular job, it is unlikely that there will be a test already prepared which has been previously validated against a large number of subjects displaying this particular trait. Therefore, the tester has to interpret these test results for the specific situation. And this is where we move on to the examination of the competence of the people doing this.
III. Broadly speaking, we can divide those who administer tests into three groups. There are those psychologists who specialize in testing who have graduate degrees in psychometry and are the most reliable. In B.C., it is unlikely that companies any smaller than B.C. Hydro would have permanent staff meeting these criteria. Thus, when tests are administered in smaller firms, they are either administered by tests in the second and third groups below, or testers belonging to this first group are brought in an consultants. Second, there are those with some kind of degree in psychology, but not a specialty in this area. As well, in this group we find people with degrees in Organizational Behaviour from a business school. These people may be compared, not completely unfairly, to psychologists in the same way we compare graduates of a two year Engineering Technology programme at BCIT to a professional engineer. The more educated personnel managers in large (for B.C.) companies would fall into this group. Third, we have those people whose background in psychology was picked up by thumbing through titles in the self-help section at the suburban shopping mall bookstore. Anyone’s grandmother, provided that she has not been exposed to this literature, would be more reliable.
Unfortunately, in B.C., personnel managers of smaller corporations or bosses, are most likely to fall into this category. The advantages of a tester from the first category are primarily threefold: First, the problem of selecting a test that is the closest thing to one having validity for the particular traits looked for in the specific job situation is best handled by a psychometrician. Second, even if the test is valid for these specific traits, the way the test is administered is crucial. Now, one might suppose that any psychologist or personnel manager would have the sense to ask the subject whether the lights are dimmed too low, or to shut to door to the room next door where the printer is going steadily; but there is more involved than this. The original test was validated by subjects taking the test under certain protocols, and if the subject being tested now is to be tested fairly, these same protocols must be observed. So, the tester must have familiarity with the particular test. Third, the interpretation of the test is crucial, and someone in the first category is the one most likely to have respect for the issues raised above, and be the most cautious in emphasizing the limitations of particular tests. Also, they will be able to command a large enough fee that employers are more likely to take seriously their cautionary advice about how much or little weight to place on the test results.
As a general rule, not without exceptions noted below, a test administered by a tester in the first category should not be one to raise too many suspicions on the part of the BCCLA; but a test administered by anyone in the second or third groups requires closer scrutiny. However, there is one important caveat to take note of. If we are faced with a client wishing our assistance in this type of case, it is not sufficient to determine that the firm in question has experts of the first group on staff, or that they were brought in as outside consultants. We must further determine that these experts were or are to be the ones actually administering the test and scoring it; and that these tasks are not to be left to people in the second and third categories to do without close supervision.
Related to this is the point that the larger, more ethical publishers of the more reliable tests make some effort to restrict distribution of their tests to qualified practitioners. In B.C., this means at least a qualified psychologist a member of the B.C. Psychological Association (BCPA) or a member of a university psychology department. But the security of this falls short of what is promised for the SDI programme for several reasons. First, the BCPA does not recognize psychometry as a specialty; and very few psychologists have in fact specialized in this area in graduate school. Of course, many psychologists have picked up the knowledge and skills required over the years, but test distributors have no way of determining that this is so. Second, not all publishers have a very tight security system; some are as lax as warning prospective buyers on the order form that lies can result in a pimple on your tongue. Third, even where publishers are more careful than this, it is virtually impossible for them to prevent qualified members purchasing the tests and assigning the administration or interpretation of the tests to a work-study student. Or, in the day of photocopy machines, we have no guarantee that, because a tester has the test, she has been vetted by the publisher or the BCPA. This problem is even worse now that many tests are published on computer disk, and therefore can be administered, scored and interpreted by anyone with enough background in psychology to pirate a disk and read the manual.
IV. Given the complexities already raised, it is clear that we cannot formulate a simple general policy on psychological testing. The best we can do is to examine each complaint received on a case-by-case basis. But what we need to know about each case is the following. First, we need to know what the relevance of the traits being tested for are to the job the applicant is being tested for. The straightforward balancing of the rights to privacy against the rights of the employer to have this information can be settled by civil libertarians at this state. There will be cases where the information is irrelevant to the job, and this is all we need to know in order to proceed with the case.
But if a civil liberties issue fails to arise at this stage, we must still consider whether the test in question is a fair way to obtain that information. At this point, we need the expertise of a psychologist specializing in psychometry and familiar with the test in question. We should consult with an academic psychologist at one of the universities, or a practicing specialist in psychometry at a hospital such as Riverview before discussing the matter with the employer who wishes to administer the test. The following is a list of questions that would be helpful to have answered before consulting with the psychologist, since the answers to these are necessary for the psychologist to give an informed opinion on whether the test is likely to yield the information that is being sought by the employer:
- What is the degree of coercion exerted on the employee to get him/her to take the test (implicit/explicit promise/threat of promotion, firing, hiring, salary increment)?
- What is the job description of the position the person is being tested for?
- What is the employer’s stated rationale for administering the test? Is the test being used for decisions about promotion, hiring, or continued employment?
- Is the employee being ‘singled out’ for this test? Are there other employees in a similar situation who were not asked to take the test?
- What are the total numbers of people in this organization taking the test?
- What other information is this test being correlated with before a decision is made (past, present or future job performance, other tests, letter from a doctor)?
- Is this test also part of a research programme, run by the organization or outside consultant?
- What is the full name of the test? (Also, edition number, year of revision, etc.)
- What kind of test is it (aptitude, personality, etc.)?
- Who is administering the test? Is it an outside consultant, or someone within the organization?
- What are the qualifications of the person(s) administering, scoring and interpreting the test (degree in psychology or organizational behaviour? If so, a specialty in psychometry? Is this person a member of the B.C. Psychological Association?)?