You are on page 1of 4

PROFESSIONAL FORUM: PROJECTIVE PREDICATION OF RETENTION 229

PROFESSIONAL FORUM:
Prediction of Job Retention Using a
Brief Projective Test Battery:
a Preliminary Report
Alan J. Raphael and Charles J. Golden*

The use of objectively validated projective tests in personnel decisions has been limited in
recent years because of the perception that such tests are highly subjective, difficult to
administer, and difficult to score in a reliable manner. The present paper demonstrates the use
of a brief (‰ hour) projective test battery consisting of the Bender±Gestalt, House±Tree±
Person, and a free drawing test which can be administered in a personnel office and scored
blindly using an objective scoring system. The study showed that such a battery could predict
six month retention rates in a sample of recently hired corrections officers at statistically
significant rates (2 = 6.25, p < 0.05) despite the fact that the individuals had already been
thoroughly prescreened using the company's comprehensive normal procedures. The possible
uses and advantages of a language-free projective battery are discussed along with future
research directions.

Prediction of job retention using a brief believe as well that self-report tests appear
projective test battery 'unfair' to minority candidates. The potential
issue of response bias, where a client gives

W hile those who use personality tests have


long felt that there is a major role for
such tests in personnel selection, many outside
socially acceptable rather than true answers, in
such tests is also a concern to users despite
construction methods which attempt to minimize
the field continue to question their validity, such issues (Hogan et al. 1996)
reliability, and fairness to minority applicants An alternative to self-report or other language
(Hogan et al. 1996). Schneider (1987) believes based measures have always been the non-verbal
that personality tests are underused in personnel projective personality tests (Golden, 1991). For
selection because of early work that suggested the past 15 years, however, such open ended or
personality variables were unimportant; a focus projective measures have generally been con-
on psychopathology rather than job specific sidered too complex for employment assess-
variables like reliability or dependability; a ment. Such tests generally were regarded as
dependence on psychological measures unsuited requiring highly qualified individual administra-
for personnel selection; a failure to recognize tion, adding substantially to evaluation costs.
that traits important for one job may be The absence of reliable scoring methods,
unimportant for another job; and the impact of difficulty in administration, a perception that
privacy issues with regard to non job specific such tests are not and cannot be objective, and
personality variables. lack of awareness in industrial and organizational
While the Hogan et al. (1996) review settings have all impaired acceptance of these
concludes that well constructed measures using techniques as a substitute for multiple choice or
self report designs are fair, such research has not true±false tests.
adequately dealt with the impact of clients While some projective techniques do indeed
whose main language or culture may be require extensive face to face administrative
inconsistent with the language in which the test methods, this is not true of projective tests like
was developed. In some cases, such individuals the Bender±Gestalt and House±Tree±Person.
may not have adequate reading levels or more Such tests can be administered quickly and
importantly, reading and language compre- easily with a minimum of training. They also *Address for correspondence:
Dr Charles Golden, 2862 NE
hension levels to understand self report tests. have the advantage of requiring few specific 35th Court, Lighthouse Point,
Although Hogan et al. disagree, many in the field language skills (instructions can be easily FL 33064, USA.

ß Blackwell Publishers Ltd 1997, 108 Cowley Road, Oxford OX4 1JF, UK and
350 Main Street, Malden, MA 02148, USA. Volume 5 Number 4 October 1997
230 INTERNATIONAL JOURNAL OF SELECTION AND ASSESSMENT

translated into any language) and appear to Method


demand less specific cultural experiences,
offering tasks which are familiar to most if not The subjects were 60 new hires with a large
all ethnic, cultural, and educational groups (Hutt multinational corrections company, The candi-
1985; Reichenberg and Raphael 1992). dates were rigorously evaluated first by the
The obvious and significant disadvantage company for appropriateness for the job of
with these tests is the issue of scoring which Corrections Officer. They were then hired for
has typically been subjective despite the these positions. After hiring, all subjects were
availability of commonly used objective scoring administered the projective test battery by on-
systems (Hutt 1985; Naglieri 1980; Reichenberg site personnel following simple standardized
and Raphael 1992). The wide acceptance of instructions which came with the tests.
these systems in clinical practice by psycho- The battery included the Bender±Gestalt Test,
logists is well recognized (Piotrowski and Keller the House±Tree±Person, and an experimental
1985). free drawing test. A sole administrator
However, the predictive validity of such conducted the entire screening (usually in a
systems in a job applicant pool remains unclear group format) in approximately one hour. The
and relatively unresearched. Beutler et al. (1975) test instructions were simple. The candidate was
found that the Bender±Gestalt was a significant asked to copy the Bender figures, and then asked
predictor of police performance using only the to draw a house, a tree, a person, and a drawing
time to complete the test in stressed and of their own choice.
unstressed conditions as their measures. Hutt At the completion of the test, all the protocols
and Briskin (1960) found that the Bender±Gestalt were sent, without any job performance
(using the Hutt system) was a significant information, to be scored by the senior author.
predictor of military personnel performance Each protocol was scored blindly. The presence
under battle conditions. of specific characteristics of the drawings on the
Thus the use of such projective tests for tests were noted, using a simple scoring system
employment screening appears reasonable and of present or absent. These specific criteria
practical. They offer the advantages of ease of (currently nearly 200 in all) included such
administration, shortness of administration time, characteristics as size of drawings, location of
less language dependence, and fewer culture the drawings on the page, the relationship of
specific demands than many other tests. They figures within the drawings, the darkness of
offer the chief disadvantage of the lack of a lines, the presence or absence of specific features
coherent research program to show their on the house, tree and person, and other similar
effectiveness. criteria (see Raphael and Reichenberg 1992 for a
The present study is an initial attempt to detailed discussion of the system related to the
analyse the effectiveness of a battery of non- Bender). There are criteria which apply across
verbal projective tests administered by a the figures and drawings, while others which are
personnel department and scored objectively specific to only a particular figure or drawing.
by a `blind' expert. For this initial study, it was Each of these criteria are included on one or
decided to attempt to predict a relatively more scales which cover a wide range of
objective employment standard: likelihood of psychological characteristics ranging from severe
employee retention in a stressful employment pathology and brain injury (rarely seen in a
situation where high turnover rates are common. normal population) to those characteristics
This variable has the advantage of being included in this study which were hypothesized
completely objective as well as important, since prior to the study to be related to vocational
employee turnover is a significant source of retention. These categories included: (1)
expense for many companies as well as leading attention to detail; (2) effective communication
to a workforce which is less experienced and skills; (3) ability to work well with others; (4)
likely less effective. logic, reasoning, problem solving, and judge-
This variable also has obvious defects which ment; (5) organizational productivity; (6) ability
may compromise the magnitude of any findings. to accept direction; (7) ability to control stress
While it is reasonable to believe that retention is and contain emotional reactions; (8) ability to
related to the interaction of job demands and postpone gratification; (9) level of patience and
personality traits, it is also affected by many impulse control; and (10) level of self-confidence.
extraneous variables such as salary, availability Based on the performance on the criteria
of other jobs, personal and family issues, and related to a specific area, each candidate was
general health. However, it is believed that this assigned a point value, based on whether the
variable is a reasonable test of the potential performance was rated (1) low, (2) moderate or
efficacy of this assessment procedure. (3) high on each variable on the basis of the
projective test performance overall. These
judgments were made on the basis of data

Volume 5 Number 4 October 1997 ß Blackwell Publishers Ltd 1997


PROFESSIONAL FORUM: PROJECTIVE PREDICATION OF RETENTION 231

gathered over 20 years of use with this system. The ease of administration, low cost, racial
The 10 rating scores were totalled to get an and cultural fairness, gender fairness, educational
overall summary evaluation with these summary fairness and difficulty in faking good all suggest
scores ranging from a minimum of 10 to a that these techniques can be a valuable aid in
possible maximum of 30. Total scores of 10±16 personnel selection processes. One possible
or less were classified as minimally qualified; drawback of these features is also related to
scores of 17±23 were classified as moderately one of these strengths. This tests are difficult to
qualified; and scores of 24±30 were classified as fake because of their lack of clear face validity: it
highly qualified. is not immediately obvious to the test taker how
No feedback was given to the subjects on their the performance on the test relates to job factors.
scores or performance, nor were supervisors or As a result, the client cannot intentionally distort
other individuals involved in their retention answers in order to `look good', but generally
aware of the scores. At the end of six months, must simply do their best without awareness of
the authors were sent a list of which employees what that means.
were still employed and which were not. This lack of face validity however may lead
some clients to be concerned about the
meaningfulness or accuracy of the procedures.
Results Clients may be concerned or even upset with
tests when they do not believe that they
A chi-square analysis was conducted using a 2 understand how they work. While this could
(employment status) by 3 (minimal, moderate, or theoretically lead to a refusal to do the tests,
high status) design (df = 2). A chi-square of 6.25 there were no difficulties with our current sample
was found, p < 0.05. Tau (0.31) was also in this regard. Any concern on a client's part is
calculated and found to be significant at the 0.05 generally mitigated by the simplicity of the tasks
level. which most clients feel they are competent to
perform. Thus, while this issue is generally not a
problem, it is one that must be kept in mind for
Discussion those cases where this issue becomes important.
Of greater importance related to the issue of
Overall, the results of this double-blind study face validity is convincing a business firm or
confirmed the usefulness of the projective test manager of the validity of the tests. Managers
battery in predicting retention in this sample, are often more comfortable with face valid tests
These results are impressive in that accuracy was when making personnel decisions, The use of
likely reduced by many unrelated and uncon- projective tests like these must be accepted not
trollable factors, as well as the use of a restricted on the basis of face validity but must be based
sample caused by the prescreening of the on research and clinical experience which shows
employer which likely caused very poor the value of the techniques in specific settings.
candidates to be underrepresented in the sample While this should be true of all types of
(e.g. no candidate scored in the 10±13 range, the instruments, it is unfortunate that sometimes
lowest scores possible.) It is likely that earlier use face validity (which is not really a form of
of these measures in the process would result in validity) takes precedence over actual research
even more useful information. Even when findings.
administered at the end of the process, the The use of projective tests offer many
battery can alert the employer to employees potential advantages, but also emphasizes the
who have a lower potential for retention which need for clear outcome research rather than
could be used for final selection, saving decisions based on face validity. The present
potentially thousands of dollars in retraining results are encouraging and illustrate both the
costs, or used to develop new programs aimed at potential value of this approach and the role that
at-risk individuals to increase the chances of empirical research can play in establishing the
retention. value of these tests and procedures. Much more
The present results do not address the research is necessary before final conclusions can
question of the projective variables to predict be reached on the full value of these tests in
aspects of the quality of job performance, personnel selection. The positive results of this
tapping into such basic variables as motivation, study will hopefully provide a stimulus for such
dependability, obsessional traits, anxiety, and future research.
maturity. While the role of these variables may
differ across jobs, being positive in one and
negative in another, the awareness of the References
applicant's strengths and weaknesses will help
the employer in distinguishing from, otherwise Beutler, I. Storm, A., Kirkin, P., Scosin, F. and Gaines,
similar applicants. P. (1975) Parameters in the prediction of police

ß Blackwell Publishers Ltd 1997 Volume 5 Number 4 October 1997


232 INTERNATIONAL JOURNAL OF SELECTION AND ASSESSMENT

officer performance. Professional Psychology, 16, Stratton.


324±335. Naglieri, J.A. (1980) Draw a person: A quantitative
Golden, C. (1991) Clinical interpretation of objective scoring system. New York: Harcourt Brace
psychological tests (Second edition). Boston: Taylor Jovanovich.
and Francis. Piotrowski, C. and Keller, R. (1985) Psychodynamic
Hogan, R., Hogan, J. and Roberts, B.W. (1996) test usage ± a survey of the Society for
Personality measurement and employment Personality Assessment. Journal of Personality, 16,
questions ± Questions and answers. American 324±335.
Psychologist, 51, 469±477. Reichenberg, N. and Raphael, A.J. (1992) Advanced
Hutt, M. (1985) The Hutt adaptation of the Bender± psychodiagnostic interpretation of the Bender±Gestalt
Gestalt Test (Fourth Edition). Orlando: Grune and Test. Hartford, CT: Praeger.
Stratton. Schneider, B. (1987) The people make the place.
Hutt, M. and Briskin, G.J. (1960) The clinical use of the Personnel Psychology, 40, 437±453.
Bender±Gestalt Test. New York: Grune and

Volume 5 Number 4 October 1997 ß Blackwell Publishers Ltd 1997

You might also like