ERIC Identifier: ED338702
Publication Date: 1991-12-00
Author: Crafts, Jennifer
Source: ERIC Clearinghouse on
Tests Measurement and Evaluation Washington DC.
Using Biodata as a Selection Instrument. ERIC/TM Digest.
Biographical inventory is a selection device used as an alternative or
supplement to cognitive testing because this measurement method predicts aspects
of job performance that are not predicted by cognitive measures. Examples of
these aspects of performance are interpersonal relationships (e.g., with
co-workers or clients) and motivation.
Biographical inventories have been empirically developed against such varied
criteria as amount of insurance sold by life insurance agents, turnover of bank
clerks, productivity of research scientists, and performance of naval personnel
in diver training. Such inventories have proved valid as predictors of job
performance in groups ranging from unskilled workers, office clerks, and service
station dealers to chemists, engineers, and high-level executives. Personal
history types of items that discriminate can provide a great deal of information
about what kinds of employees remain on a job and what kinds do not, and what
kinds are promotable and what kinds are not.
This Digest discusses some of the issues and concerns about using
biographical inventories as well as the rationale behind them; their
verifiability; their format; legal issues associated with them; evidence about
their reliability; and validity and fairness.
The assumption that underlies the use of biodata
is that past behavior is a valid predictor of future behavior. More
specifically, it is assumed that information obtained from job applicants about
previous work experiences, education, etc. can be used to predict job
performance. Items included in these inventories are selected on the basis of
previous research which demonstrates significant relationships between item
responses and job performance. Personal history items commonly used fall into
the following areas:
socioeconomic level-financial status
personal characteristics, attitudes expressed.
Biodata items can be differentiated on the
basis of how directly and easily they can be verified. "Hard" items can be
verified against records, whereas "soft" items cannot be checked for
truthfulness, beyond the word of the respondent. The advantage of using only
verifiable items is that individuals tend to provide honest responses. However,
the range of information generated by this approach is somewhat restricted. The
combination of both item types broadens the variety of information collected.
However, individual subjectivity often colors responses to the soft items (e.g.,
items asking for opinions and interpretations). The most predictive items are
those which require individuals to summarize their feelings about a range of
Biodata items are typically cast into a short answer
or multiple choice format, which can be objectively scored. Owens (1976) has
cataloged the following types of items: @ @ @ @ @ @
Owens (1976) advocates using items with response options that lie along a
continuum (either apparent or demonstrated), for ease of statistical analysis.
BIODATA AND LEGAL ISSUES
Current guidelines, regulations,
and statutes restrict certain types of information from being included on
biodata inventories. These limitations have been imposed to protect applicants
from being denied employment based on factors unrelated to jobs. Unless
demonstrated to be job-relevant, items addressing race, gender, marital status,
number of dependents, birth order, and spouse's occupation are not considered
appropriate as a basis for selection decisions. As long as they are correlated
with job success or related to "business necessity," other personal items such
as grade point average or level of education can be used for personnel
Several documents have been generated that set forth the legal and technical
standards for test development, use, and validation. The Uniform Guidelines on
Employee Selection Procedures, the Standards for Educational and Psychological
Testing, and the Principles for the Validation and Use of Personnel Selection
Procedures provide frames of reference to ensure that major relevant test
construction issues are addressed.
Estimates of biodata reliability vary greatly
with the specific content of items included in an inventory and with the type of
reliability estimate computed. To the extent that the items measure different
constructs, internal consistency estimates will be low. Thus, an inventory built
from empirically keyed items, as opposed to rationally derived constructs, will
usually have low item intercorrelations and therefore must be longer to increase
VALIDITY AND FAIRNESS EVIDENCE
Many researchers have
conducted reviews of biodata validities. Results demonstrate that biodata
predict a variety of job performance criteria, such as training success, job
proficiency, tenure, and salary levels. Reilly and Chao (1982) conducted an
extensive review of biodata predictor studies and found an average validity of
.35 over many occupations and over several different criteria. The accuracy of
personal history items as predictors of future work behavior was also supported
by Hunter and Hunter's (1984) meta-analysis result of an average validity of
.37. Ghiselli (1966) reported that biodata predictors had higher mean
validities, averaged over a variety of occupations, than other types of
predictors. The mean correlation with trainability was .44, and the mean
correlation with job proficiency was .41.
One issue associated with the use of biodata is that
almost all inventories are developed using the concurrent validation strategy.
Therefore, range restriction effects are a concern. Another problem with biodata
is stability of prediction. A number of studies which showed significant
one-time validity results have also demonstrated that these estimates decreased
over time and across situations.
Other issues of concern for biodata are accuracy, fakeability, invasion of
privacy, and adverse impact. Very little information exists to support or refute
allegations of inaccuracy, invasion of privacy, or fakeability. Owens (1976)
reviewed investigations of adverse impact and reported that "the major
dimensions of biodata responses are quite stable across cultures, age, race, and
sex groups, and companies." Adverse impact may depend on the degree to which
items elicit information that directly or indirectly reflects cultural
differences in social, educational, or economic advancement opportunities. Thus,
in constructing inventories, care must be taken to include items with potential
for reducing adverse impact.
Ghiselli, E.E. (1966) The validity of
occupational aptitude tests. New York: Wiley.
Hunter, J.E. and R.F. Hunter (1984) The validity and utility of alternative
selection predictors of job performance, Psychological Bulletin, 96, 72-98.
Owens, W.A. (1976) Background data. In M.D. Dunnette (ed.) Handbook of
industrial and organizational psychology. Chicago: Rand McNally.
Reilly, R.R. and G.T. Chao (1982) Validity and fairness of some alternative
employee selection procedures, Personnel Psychology, 35, 1-62.