ber87251_ch06_135-170 C 1/27/06 H A 15:48 Page 135 P E T R 6 PERSONNEL SELECTION* O V E R V I E W G It sounds simple: Match employees with A this task jobs. Researchers have made easier by developing selection T methods that successfully predict employee effectiveness. E what research Still, there is a void between indicates and how organizations S actually do personnel selection. Real-world personnel , selection is replete with examples of methods that have been proven to be ineffective or inferior to other methods. D retention) is Personnel selection (and a key to organizational effectiveness. The E most successful firms tend to use methods A that accurately predict future performance. We also are interested Nin selecting employees who will not only be effective, D but who will work for us as long as we R engage in need them, and who will not counterproductive behaviorAsuch as violence, substance abuse, avoidable accidents, and employee theft. A multiple-hurdle process 1 involving an application, reference and background 1 checks, various forms of standardized testing, and some form of 2 interview is the typical chronology of events for selection, 3 decisions. particularly for external hiring Internal decisions, such asT promotions, are typically done with less formality. S Selection is the process of gathering and assessing information about job candidates and ultimately making decisions about personnel. The process applies to both entry-level personnel decisions and decisions regarding promotions, transfers, and even job retention as part of corporate downsizing efforts. This chapter will introduce you to selection, describe some of the most popular types of screening procedures, review the research evidence on each, and discuss the social and legal implications of selection methods. We will first provide an overview of selection and the typical steps employed in the process. We will then introduce you to the various selection approaches in their usual order of use: First, we will review application and biographical blanks; next, we will review the use of background and reference checks; then we will review the various forms of standardized tests that purport to assess applicants’ KASOCs. Finally, the chapter will conclude with a discussion of the use of more sophisticated selection procedures, such as assessment centers, performance testing and work samples, and drug and medical tests in the preemployment selection process. Our context for this discussion will be the legal implications of the various personnel practices and areas where there are clear discrepancies between what typically happens and what academic research indicates. This is certainly one chapter where the discrepancies between findings in academic research and actual practice are quite great. O B J E C T I V E S After reading this chapter, you should be able to 1. Understand the concepts of reliability, validity, and utility. 2. Understand the validity evidence for various selection methods. 3. Discuss approaches to the more effective use for application blanks, *An early version of this chapter was written by Michael M. Harris and Barbara K. Brown. 135 ber87251_ch06_135-170 136 1/27/06 15:48 Page 136 PART II Acquiring Human Resource Capability reference checks, biographical data, and the interview in order to increase the validity and legal defensibility of each. 4. Discuss the approaches available for drug testing. 5. Review the validity of different approaches to interviewing. 6. Discuss how the various types of candidate information should be integrated and evaluated. Wackenhut Security had its share of selection challenges. Although recruitment efforts and a sluggish economy attracted a large number of applicants for its entry-level armed and unarmed security guard positions, new contract opportunities developed after the September 11, 2001, tragedy and new concern was raised about the quality of its personnel. The turnover rate for some positions exceeded 100 percent—meaning, the quit rate in one year exceeded the number of positions. Wackenhut Security also was dissatisfied with the quality of its supervisory personnel. The company contracted with BA&C (Behavioral Analysts and Consultants), a Florida psychological consulting firm that specializes in selection problems and personnel selection. Wackenhut asked BA&C to develop a new personnel selection system for entry-level guards and supervisors. Underlying this request was a need for Wackenhut to improve its competitive position in this highly competitive industry by increasing sales and contracts, decreasing costs, and, perhaps most important, making certain their security personnel could measure up. The company, which already compensated its guards and supervisors more than others in the industry, wanted to avoid an increase in compensation in these areas. The company estimated that the cost of training a new armed guard was about $1,800. With several hundred guards quitting in less than a year, the company often failed to even recover training costs in sales. Wackenhut needed new selection methods that could increase the effectiveness of the guards and supervisors and identify guard applicants most likely to stay with the company. You will recall from Chapter 4 that job analysis should identify the knowledge, abilities, skills, and other characteristics (KASOCs) that are necessary for successful performance and retention on the job. In this case, BA&C first conducted a job analysis of the various guard jobs to get better information on the KASOCs required for the work. After identifying the critical KASOCs, BA&C developed a reliable, valid, and job-related weighted application blank, screening test, and interview format. The process of selection varies substantially from company to company. While Wackenhut initially used only a high school diploma as a job specification, an application blank, a background check, and an interview by someone in personnel, other companies have used more in the Development and FIGURE 6-1 Steps Evaluation of a Selection Procedure JOB ANALYSIS/HUMAN RESOURCE PLANNING Identify knowledge, abilities, skills, and other characteristics (KASOCs) (aka: competencies) Use a competency model tied to strategy orientation RECRUITMENT STRATEGY: SELECT/DEVELOP SELECTION PROCEDURES Review options for assessing applicants on each of the KASOCs: Standardized tests (cognitive, personality, motivational, psychomotor) Application blanks, biographical data, background, reference checks, accomplishment record Performance tests, assessment centers, interviews G DETERMINE VALIDITY FOR SELECTION A METHODS Criterion-related validation T Expert judgment (content validity) E Validity generalization DETERMINE WEIGHTING SYSTEM FOR S SELECTION METHODS AND RESULTANT DATA , Dcomplex methods to select employees. American ProtecEtive Services, for example, the company that handled seAcurity for the Olympics, used a battery of psychological and aptitude tests along with a structured interview. N As with the job analysis and the recruitment Dprocess, personnel selection should be directly linked to the HR planning function and the strategic objectives of Rthe company. The mission goal of the Marriott CorporaAtion is to be the hotel chain of choice of frequent travel- ers. As part of this strategy, the company developed a selection system designed to identify people who could be 1particularly attentive to customer demands. Wackenhut also had a major marketing strategy aimed at new con1tracts for armed security guards who would be ex2tremely vigilant. They needed a legal selection system could identify people most likely to perform well in 3that this capacity. T Figure 6-1 presents a chronology of events in the seprocess and the major options available for perSlection sonnel selection. The previous chapters on job analysis, planning, and recruitment have gotten us to the point of selecting job candidates based on information from one or more selection methods. We will review each of these methods in this chapter. But keep in mind the focus should be on selecting or developing tools that will provide valid assessments on the critical KASOCs, competencies, or job specifications most important for strategy execution. So, the job analysis should identify the strategically important KASOCs or competencies. Then, particular selection methods (or tools) should be adopted to assess these job specifications. ber87251_ch06_135-170 1/27/06 15:48 Page 137 CHAPTER 6 Personnel Selection S ELECTION M ETHODS : A RE T HEY E FFECTIVE ? Our review includes a summary of the validity of each major approach to selection and an assessment of the relative cost to develop and administer each method. Three key terms related to effectiveness are reliability, validity, and utility. While these terms are strongly related to one another, the most important criterion for a selection method is validity. Remember our discussion of the research on High-Performance Work Systems. One of the HR practices shown to be related to corporate performance was the percentage of employees hired using “validated selection methods.”1 The essence of the term validity is the extent to which a selection method predicts one or more important criteria. While the most typical criterion of interest to selection specialists is job performance, companies also may be very interested in other criteria such as how long an employee may stay on the job or whether the employee will steal, be violent, or be involved in accidents. But before we address the validity of a method, let’s look at one of the necessary conditions for validity: the reliability of measurement. What Is Reliability? A necessary condition in order for a selection method to be valid is that it first be reliable. Reliability concerns the consistency of measurement. This consistency applies to the scores that derive from the selection method. These scores can come from a paper-and-pencil test, a job interview, a performance appraisal, or any other method that is used to make decisions about people. The CIA uses a very long multiple-choice test as an initial sceening device for job applicants to be agents. If applicants were to take the test twice three weeks apart, their scores on the test would stay pretty much the same (the same thing can be said for SAT scores). The level of reliability can be represented by a correlation coefficient. Correlations from 0 to 1.0 show the extent of the reliability. Generally, reliable methods have reliability coefficients that are .8 or higher, indicating a high degree of consistency in scores. No selection method achieves perfect reliability, but the goal should be to reduce error in measurement as much as possible. If raters are a part of the selection method, such as job interviewers or on-the-job performance evaluators, the extent to which different raters agree also can represent the reliability (or unreliability) of the method. Remember how we cast serious doubts upon graphology (or handwriting analysis) in Chapter 1? This method of selection is used by some U.S. companies and even more European firms. One problem with this method is that it is not even reliable, much less valid. If the same handwriting sample was given to two graphologists, they would generally not agree on scores on various employment-related attributes (e.g., drive, 137 creativity, intelligence). Even if they did agree this does not necessarily mean that their assessments are valid. More reliable tests tend to be longer. One of the reasons the SAT, the GRE, the GMAT, and the LSAT seemingly take forever to complete is so these tests will have very high reliability (and they do). But while high reliability is a necessary condition for high validity, high reliability does not ensure that a method is valid. The SAT may be highly reliable, but do scores on the SAT predict anything important such as how well you actually will perform in college? This question addresses the validity of the method. G A T E S , D E A N D R A 1 1 2 3 T S What Is Validity? The objective of the Wackenhut Security Consultants was to develop a reliable, valid, legally defensible, userfriendly, and inexpensive test that could predict both job performance and long job tenure for security guards. The extent to which the test was able to predict an important criterion was an indication of the test’s validity. The term validity is close in meaning but not synonymous with that critical legal term job relatedness, which we discussed in Chapters 3 and 4. Empirical or criterion-related validity involves the statistical relationship between performance or scores on some predictor or selection method (e.g., a test or an interview) and performance on some criterion measure such as on-the-job effectiveness (e.g., sales, supervisory ratings, job turnover, employee theft). At Wackenhut, a study was conducted in which scores on their proposed screening test were correlated with job performance and job tenure. Such a study would strongly support an argument of job relatedness. The statistical relationship is usually reported as a correlation coefficient. This describes the relationship between the predictor and measures of effectiveness (also called criteria). Correlations from 1 to 1 show the direction and strength of the relationship. Higher correlations indicate stronger validity. Assuming that the study was conducted properly, a significant correlation between a method’s scores and some important criterion could be offered as a strong argument for the job relatedness of the method if the method resulted in adverse impact against a protected class. Figure 6-2 presents a summary of the empirical validity evidence for the various selection tools, the cost of their development and administration, and group differences by ethnicity. The higher the correlation, the more predictive (and valid) the selection method. The correlation also can be used to calculate the financial value of a selection method, using a utility formula, which can convert correlations into dollar savings or profits that can be credited to a particular selection method. A method’s utility depends on its validity but other issues as well. For example, recall our discussion of selection ratio in Chapter 5. Selection ratio is the number of positions divided by the ber87251_ch06_135-170 138 1/27/06 15:48 Page 138 PART II Acquiring Human Resource Capability FIGURE 6-2 Selection Tools, Cost for Development and Administration, and Group Differences Validitya Tool Costs (Development/ Administration)b Group Differencesc Cognitive ability tests measure mental abilities such as logic, reading comprehension, verbal or mathematical reasoning, and perceptual abilities, typically with paper-and-pencil or computer-based instruments. .51 Low/low B/W: 1.0 H/W: .5 A/W: .2 W/M: 0 Structured interviews measure a variety of skills and abilities, particularly noncognitive skills (e.g., interpersonal skills, leadership style, etc.) using a standard set of questions and behavioral response anchors to evaluate the candidate. .51 High/high B/W: .23 H/W: .17 .31 Low/high B/W: .32 H/W: .71 .54 High/high B/W: .38 .48 High/low B/W: .38 .31 Low/low B/W: .06 H/W: .04 A/W: .08 W/M: .08 .35 High/low B/W: .78 for grades B/W: .27 biodata H/W: .08 biodata W/M: .15 biodata .34 High/low B/W: .61 on paper and pencil B/W: .43 on video H/W: .26 on paper and pencil H/W: .39 on video W/M: .26 on paper and pencil W/M  .19 on video Unstructured interviews measure a variety of skills and abilities, particularly noncognitive skills (e.g., interpersonal skills, leadership style, etc.) using questions that vary from candidate to candidate and interviewer to interviewer for the same job. Often, specific standards for evaluating responses are not used. Work samples measure job skills (e.g., electronic repair, planning and organizing), using the actual performance of tasks that are similar to those performed on the job. Typically, work samples use multiple, trained raters and detailed rating guides to classify and evaluate behaviors. Job knowledge tests measure bodies of knowledge (often technical) required by a job, often using formats such as multiple-choice questions or essay-type items. Conscientiousness measures the personality trait “conscientiousness,” typically with multiple-choice or true/false formats. Biographical information measures a variety of noncognitive skills and personal characteristics (e.g., conscientiousness, achievement orientation) through questions about education, training, work experience, and interests. Situational judgment tests measure a variety of noncognitive skills by presenting individuals with short scenarios (either in written or video format) and ask what would be their most likely response or what they see as the most effective response. G A T E S , D E A N D R A 1 1 2 3 T S ber87251_ch06_135-170 1/27/06 15:48 Page 139 CHAPTER 6 Personnel Selection 139 FIGURE 6-2 (Continued) Integrity tests measure attitudes and experiences related to a person’s honesty, dependability, trustworthiness, and reliability, typically with multiple-choice or true/false formats. .41 Low/low B/W: .04 H/W: .14 A/W: .04 W/M: .16 Assessment centers measure knowledge, skills, and abilities through a series of work samples/exercises that reflect job content and types of problems faced on the job, cognitive ability tests, personality inventories, and/or job knowledge tests. .37 High/high Varies by exercise; .02 to .58 Reference checks provide information about an applicant’s past performance or measure the accuracy of an applicant’s statements on the résumé or in interviews by asking individuals who have previous experience with a job candidate to provide an evaluation. .26 Low/low ?? G A Validity values range from 0 to 1.0, with higher numbers indicating better prediction of job performance. The labels “high” and “low” are designations relative to other tools rather than based T on some specific expense level. Values are effect sizes expressed in standard deviation units. Higher numbers indicate a greater difference; negative values mean the first group scores lower. B/W is black/white difference; H/W is Hispanic/white difference; A/W is Asian/white E difference; W/M is female/male difference. Source: Adapted from Ryan, A. M. & Tippins, N. T. (2004). “Attracting and selecting: What psychological research tells us.” HUMAN RESOURCE S MANAGEMENT, 43, pp. 307–308. Reprinted with permission of John Wiley & Sons. , a b c number of applicants for those positions. A test with perfect validity will have no utility if the selection ratio is 1.0 (one applicant per position). This is why recruitment and other HR issues such as compensation are so important for personnel selection. Valid selection methods only have great utility for an organization when that organization can be selective based on the scores on that method. There’s almost no point in developing and administrating a highly valid selection method if you have to hire anyone who took the test. This was a problem for the military in 2005 and perhaps later. Content validity assesses the degree to which the content of a selection method represents (or assesses) the requirements of the job. A knowledge-based test for “Certified Public Accountant” could be considered to have content validity for an accounting job. Subject matter experts are typically used to evaluate the compatibility of the content of a test with the actual requirements of a job (e.g., is the knowledge or skill assessed on the test compatible with the knowledge or skill required on the actual job?). Such a study also can be offered as evidence of job relatedness, but the study should follow the directions provided by the Supreme Court in Albemarle v. Moody (see Chapter 3) and, just to be safe, comply with the Uniform Guidelines on Employee Selection Procedures (UGESP). (See www.eeoc.gov for details on the UGESPs.) Validity generalization invokes evidence from past studies on a selection method that is then applied to a new and similar jobs and settings. What Is Utility? Utility concerns the economic gains derived from using a particular selection method. The basic formula involves D E A N D R A 1 1 2 3 T S estimating the increase in revenue as a function of the use of the selection method after subtracting the cost of the method. As we said above, good utility requires low selection ratios and thus is related to the ability of the organization to attract a large number of qualified applicants for each position they need to fill. Selection methods with high validity but that cost relatively little are the ideal for utility. Before contracting with BA&C, Wackenhut Security had studied the options and was not impressed with the validity or utility evidence reported by the test publishers, particularly in the context of the $10–$15 cost per applicant. This was the main reason Wackenhut decided to develop its own methods. BA&C investigated the validity of its proposed new selection systems using both criterion-related and content-validation procedures. This dual approach to validation provides stronger evidence for job relatedness. The BA&C study strongly suggested that new methods of personnel selection should be used if the company hoped to increase its sales and decrease the costly employee turnover. The resulting analysis showed substantial financial benefit to the company if it adopted the new methods for use in lieu of the old ineffective procedures. The first method BA&C considered was the application blank. A PPLICATION B LANKS AND B IOGRAPHICAL DATA Like most companies, Wackenhut first required a completed application blank requesting standard information about the applicant, such as previous employment history, experience, and education. Often used as an initial ber87251_ch06_135-170 140 1/27/06 15:48 Page 140 PART II Acquiring Human Resource Capability screening method, the application blank, when properly used, can provide much more than a first cut. However, application blanks, as with any other selection procedure used for screening people, falls under the scrutiny of the courts for possible EEO violations. HR managers should be cautious about using information on an application blank that disproportionately screens out protected class members, and they must be careful not to ask illegal questions. The passage of the Americans with Disabilities Act (ADA), for example, states that application blanks should not include questions about an applicant’s health, disabilities, and worker’s compensation history. Application blanks obviously can yield information relevant to an employment decision. Yet, it is often the weight—or lack of weight—assigned to specific information by particular decision makers that can seriously undermine their usefulness. Decision makers often disagree about the relative importance attached to information on application blanks. For instance they might disagree about the amount of education or experience required. Wackenhut required a bachelor’s degree in business or a related discipline for the supervisory job. This criterion alone, however, should not carry all the weight. Wackenhut’s personnel staff made no effort to develop a uniform practice of evaluating the information on the forms. They did not take into consideration indicators such as the fact that an applicant lives 20 miles from the workplace. This may indicate that, relative to other responses, the candidate is more likely to quit as soon as another job comes along that is closer to home. A Discrepancy between Research and Practice: The Use of Application and Biographical Data What companies do to evaluate application blank data and biographical information and what research suggests are worlds apart. Decision makers rarely use a uniform approach to evaluate data. Scholarly research clearly shows, with adequate data available, the best way to use and interpret application blank information is to derive an objective weighting system.2 The system is based on an empirical research study, resulting in a weighted application blank (WAB), with the weights derived from the results of the research. By empirical study, we mean the responses from the application blanks are statistically related to one or more important criteria such that the critical predictive relationships can be identified. For example, BA&C was able to show that where a security guard lived relative to his assigned duties was indeed a significant predictor of job turnover. Another useful predictor was the number of jobs held by the applicant during the past three years. Figure 6-3 shows some examples from a WAB. The process of statistically weighting the information on an application blank enhances use of the application blank’s information and improves the validity of FIGURE 6-3 Examples of WAB and BIB WAB EXAMPLES How many jobs have you held in the last five years? (a) none (0); (b) 1 (5); (c) 2–3 (1); (d) 4–5 (3); (e) over 5 (5) What distance must you travel from your home to work? (a) less than 1 mile (5); (b) 1–5 miles (3); (c) 6–10 miles (0); (d) 11–20 miles (–3); and (e) 21 or more miles (5) BIB EXAMPLES How often have you made speeches in front of a group of adults? How often have you set long-term goals or objectives for yourself? G How often have other students come to you for advice? A How often have you had to persuade someone to do what you wanted? T How often have you felt that you were an unimportant member of a group? E How often have you felt awkward about asking for help S on something? often do you work in “study groups” with other , How students? How often have you had difficulties in maintaining your priorities? D How often have you felt “burnt out” after working hard on a task? E How often have you felt pressured to do something when A you thought it was wrong? NSource: C. J. Russell, J. Matson, S. E. Devlin, and D. Atwater, “Predictive of Biodata Items Generated from Retrospective Life Experience DValidity Essays,” Journal of Applied Psychology 75 (1990), pp. 569–580. Copyright 1990 by the American Psychological Association. Reproduced with R©permission. A the whole process. The WAB simply is an application blank that is scored—similar to a paper-and-pencil test. 1It provides a score for each job candidate and makes it possible to compare the score with that of other 1candidates. For example, the numbers in parentheses for 2the WAB examples in Figure 6-3 were derived from an study showing particular responses were related 3actual to job tenure. T Biographical information blanks (BIB) are similar WABs except the items of a BIB tend to be more perSto sonal with questions about personal background and life experiences. Figure 6-3 shows examples of items from a BIB for the U.S. Navy. BIB research has shown the method can be an effective tool in the prediction of job turnover, job choice, and job performance. In one excellent study conducted at the Naval Academy, biographical information was derived from life-history essays, reflecting accomplishments that were then written in multiple-choice format (see Figure 6-3).3 BIB scoring is also derived from a study of how responses relate to important criteria. WABs and BIBs have been used in a variety of settings for many types of jobs. WABs are used primarily for ber87251_ch06_135-170 1/25/06 09:09 Page 141 CHAPTER 6 Personnel Selection clerical and sales jobs. BIBs have been used successfully in the military and the insurance industry. Many insurance companies, for example, use a very lengthy BIB to screen their applicants. Check out www.e-Selex.com for an online biodata testing service. The accomplishment record is an approach similar to a BIB. Job candidates are asked to write examples of their actual accomplishments, illustrating how they had mastered job-related problems or challenges. Obviously, the problems or challenges should be compatible with the problems or challenges facing the organization. The applicant writes these accomplishments for each of the major components of the job. For example, in a search for a new business school dean, applicants were asked to cite a fund-raising project they had successfully organized. HRM specialists evaluate these accomplishments for their predictive value or importance for the job to be filled. Accomplishment records are particularly effective for managerial, professional, and executive jobs.4 In general, research indicates that methods such as BIBs and accomplishment records are more valid than credentials. For example, having an MBA versus only a Bachelor’s degree is not a particularly valid predictor of successful management performance. What an applicant has accomplished in past jobs or assignments is a more valid approach to “leadership” assessment. How Do You Derive WAB or BIB or Accomplished Record Weights? To derive the weights for WABs or BIBs, you ideally need a large (at least 150) representative sample of application or biographical data and criterion data (e.g., job tenure and/or performance) of the employees in the position under study. You then can correlate responses to individual parts of the instrument with the performance data. If effective and ineffective employees responded to an item differently, responses to this item would then be given different weights, depending on the magnitude of the relationship. Weights for the accomplishment record are usually derived by expert judgment for various problems or challenges. Research supports the use of WABs, BIBs, and the accomplishment record in selection. The development of the scoring system requires considerable work, but it is worthwhile because the resulting decisions are often superior to those typically made based on a subjective interpretation of application blank information. However, since you need a large sample size to validate results, the WAB technique will probably be useful only for jobs with many incumbents. What if you can’t do the empirical validation study? Might you still get better results using a weighted system, in which the weights are based on expert judgment? Yes. This approach is superior to one in which there is no uniform weighting system and each application blank or résumé is evaluated in a more holistic manner by whoever is evaluating it. 141 R EFERENCE C HECKS AND BACKGROUND C HECKS G A T E S , D E A N D R A More than 80 percent of companies do some form of reference or background check.5 The goal is to gain insight about the potential employee from people who have had previous experience with him or her. An important role of the background check is to simply verify the information provided by the applicant regarding previous employment and experience. This is a good practice, considering research indicates that between 20 and 25 percent of job applications include at least one fabrication.6 Fear of negligent hiring lawsuits is a related reason employers do reference and background checks. A negligent hiring lawsuit is directed at an organization accused of hiring incompetent (or dangerous) employees. One HMO was sued for $10 million when a patient under the care of a psychologist was committed to a psychiatric institution and it was later revealed that the psychologist was unlicensed and lied about his previous experience. A second purpose for reference checks is to assess the potential success of the person for the new job. Reference checks provide information about a candidate’s past performance and are also used to assess the accuracy of information provided by candidates. However, HR professionals should be warned: a proliferation of lawsuits has engendered a great reluctance on the part of evaluators to provide anything other than a statement as to when a person was employed and in what capacity. These lawsuits have been directed at previous employers for defamation of character, fraud, and intentional infliction of emotional distress. This legal hurdle has prompted many organizations to stop employees from providing any information about former employees other than dates of employment and jobs. Turnaround is fair play—at least litigiously. Organizations are being sued and held liable if they do not give accurate information about a former employee when another company makes such a request. The bottom line appears simple: Tell the truth. There are laws in several states that provide protection for employers who provide candid and valid evaluations of former employees. 1 1 2 3 T What Are the Legal Implications of Doing S Background Checks on Job Candidates? Employers often request consumer reports or more detailed “investigative consumer reports” (ICVs) from a consumer credit services as a part of the background check. If this is so, employers need to be aware of state laws related to background checks and The Fair Credit Reporting Act (FCRA), amended in 2005, a federal law that regulates how such agencies provide information about consumers. State laws vary considerably on background checks. Experts maintain that it is legally safest to comply with the laws of the states where the job candidate resides, where the reporting agency is incorporated, ber87251_ch06_135-170 142 1/25/06 09:09 Page 142 PART II Acquiring Human Resource Capability and the employer has its principal place of business. In general, in order to abide by the FCRA or state law, four steps must be followed by the employer: (1) Give the job candidate investigated a notice in writing that you may request an investigative report, and obtain a signed consent form; (2) Provide a summary of rights under federal law (individuals must request a copy); (3) Certify to the investigation company that you will comply with federal and state laws by signing a form they should provide; and (4) Provide a copy of the report in a letter to the person investigated if a copy has been requested or if an adverse action is taken based on information in the report. One of the problems with letters of reference is that they are almost always very positive while there is some validity, it is low in general (.26). One approach to getting more useful (and valid) distinctions among applicants is to construct a “letter of reference” or recommendation that is essentially a performance appraisal form.7 One can construct a rating form and request that the evaluator indicate the extent to which the candidate was effective in performing a list of job tasks. This approach offers the added advantage of deriving comparable data for both internal and external job candidates, since the performance appraisal, or reference data, can be completed for both internal and external candidates. With this approach, both internal and external evaluators must evaluate performances on the tasks that are most important for the position to be filled. An alternative approach asks the evaluator to rate the extent of job-related knowledge, skill, ability, or competencies of a candidate. These ratings can then be weighted by experts based on the relative importance of the KASOCs or competencies for the position to be filled. This approach makes good sense whenever past performance is a strong predictor of future performance. For example, when selecting a manager from a pool of current or former managers, a candidate’s past performance as a manager is important. Performance appraisals or promotability ratings, particularly those provided by peers, are a valid source of information about job candidates. However, promotability ratings made by managers are not as valid as other potential sources of information about candidates, such as performance tests and assessment centers. Employers should do their utmost to obtain accurate reference information despite the difficulties. If for no other reason, a good-faith effort to obtain verification of employment history can make it possible for a company to avoid (or win) negligent hiring lawsuits. P ERSONNEL T ESTING Surveys indicate that between 15 and 20 percent of organizations use some form of ability or knowledge testing to make selection decisions.8 Many companies now use aptitude or cognitive ability tests to screen applicants, bolstered by considerable research indicating the tests are valid for virtually all jobs in the U.S. economy. The dilemma facing organizations is this: While mental or cognitive ability tests have been shown to be valid predictors of job performance, they can create legal problems because minorities tend to score lower. Corporate America also is increasing its use of various forms of personality or motivational testing—in part due to the body of evidence supporting the use of certain methods, concern over employee theft, the outlawing of the polygraph test, and potential corporate liability for the behavior of its employees. Lawsuits for negligent hiring and negligent retention, for example, attempt to hold an organization responsible for the behavior of employees when there is little or no attempt to assess critical characGteristics of those who are hired. Domino’s Pizza settled a lawsuit in which one of its delivery personnel was inAvolved in a fatal accident. The driver had a long and Tdisturbing psychiatric history and terrible driving record before he was hired. E Cognitive ability tests are the most frequently used Spaper-and-pencil tests in use today. These tests attempt to mental, clerical, mechanical, or sensory capabil, measure ities in job applicants. You are probably familiar with these cognitive ability tests: the Scholastic Aptitude Test the American College Test (ACT), and the GenD(SAT), eral Mental Ability Test (GMAT). Cognitive ability tests, Emost of which are administered in a paper-and-pencil or Acomputerized format under standardized conditions of test administration, are controversial. On the average, NAfrican Americans and Hispanics score lower than Dwhites on virtually all of these tests; thus, use of these tests can affect employment and other opportunities for Rminorities (see Figure 6-2). A We will address the critical issue of test score differences as a function of ethnicity later in the chapter. Let us begin our discussion with a definition of cognitive ability 1testing and provide brief descriptions of some of the most popular tests. Then we will review the validity evidence 1for these tests. We will conclude with a focus on the legal 2aspects of cognitive ability testing in the context of the latest research, ethnic score differences, and case law. 3 T What Is a Cognitive Ability Test? SCognitive ability tests measure one’s aptitude or mental capacity to acquire knowledge based on the accumulation of learning from all possible sources. Such tests are often distinguished from achievement tests, which attempt to measure the effects of knowledge obtained in a standardized environment (e.g., your final exam in this course could be considered a form of achievement test). Cognitive ability or aptitude tests are typically used to predict future performance. Examples are the SAT and ACT, which were developed to measure ability to master college-level material. Having made this distinction between achievement tests and cognitive ability tests, however, we hasten to ber87251_ch06_135-170 1/27/06 15:48 Page 143 CHAPTER 6 Personnel Selection point out that in practice there isn’t a clear distinction between these two classes of tests. Achievement tests can be used to predict future behavior and all tests measure some degree of accumulated knowledge. Knowledgebased tests assess a sample of what is required on the job. If you are hiring a computer programmer, a cognitive ability test score might predict who will learn to be a computer programmer; yet, you would benefit more with an assessment of actual programming knowledge. Knowledge-based tests are easier to defend in terms of job relatedness and are quite valid (.48). They can be expensive to develop. There are hundreds of mental or cognitive ability tests available. Some of the most frequently used and highly regarded tests are the Wechsler Adult Intelligence Scale, the Wonderlic Personnel Test, and the Armed Services Vocational Aptitude Battery. In addition, many of the largest U.S. companies have developed their own battery of cognitive ability tests. AT&T evaluates applicants for any of its nonsupervisory positions on the basis of scores on one or more of its 16 mental ability subtests— the weights given to a particular test depend on the particular job and the validation results. Knight-Ridder, the communications giant, has a battery of 10 aptitude tests, some of which are even used to select newspaper carriers. There are hundreds of cognitive ability tests available for commercial use. The Wechsler Adult Intelligence Scale is one of the most valid and heavily re searched. A valid and more practical test is the Wonderlic Personnel Test. The publisher of this test, first copyrighted in 1938, has data from more than 3 million applicants. The Wonderlic consists of 50 questions, covering a variety of areas including mathematics, vocabulary, spatial relations, perceptual speed, analogies, and miscellaneous topics. Here is an example of a typical mathematics question: “A watch lost 1 minute 18 seconds in 39 days. How many seconds did it lose per day?” A typical vocabulary question might be phrased as follows: “Usual is the opposite of: a. rare b. habitual c. regular d. stanch e. always.” An item that assesses ability in spatial relations would require the test taker to choose among five figures to form depicted shapes. Applicants have 12 minutes to complete the 50 items. The Wonderlic will cost an employer from $1.50 to $3.50 per applicant depending on whether the employer scores the test. The Wonderlic is used by the National Football League to provide data for potential draft picks (the average score of draftees is one point below the national population).9 You may remember the Wonderlic from our discussion of the Supreme Court ruling in Griggs v. Duke Power (discussed in Chapter 3) and Albemarle v. Moody. In Griggs, scores on the Wonderlic had an adverse impact against African Americans (a greater proportion of African Americans failed the test than did whites); and Duke Power did not show that the test was job related. Despite early courtroom setbacks and a decrease in use 143 following the Griggs decision, according to the test’s publisher, the use of the Wonderlic has increased in recent years. What Are Tests of Specific Ability? G A T E S , D E A N D R A 1 1 2 3 T S A variety of tests also have been developed to measure specific abilities, including specific cognitive abilities such as verbal comprehension, numerical reasoning, and verbal fluency, as well as tests assessing mechanical or clerical ability, physical or psychomotor ability, including coordination and sensory skills. The most widely used mechanical ability test is the Bennett Mechanical Comprehension Test (BMCT). First developed in the 1940s, the BMCT consists mainly of pictures depicting mechanical situations with questions pertaining to the situations. The respondent describes relationships between physical forces and mechanical issues. The BMCT is particularly effective in the prediction of success in mechanically oriented jobs. While there are several tests available for the assessment of clerical ability, the most popular is the Minnesota Clerical Test (MCT). The MCT requires test takers to quickly compare either names or numbers and to indicate pairs that are the same. The name comparison part of the test has been shown to be related to reading speed and spelling accuracy, while the number comparison is related to arithmetic ability. Physical, psychomotor, and sensory/perceptual are classifications of ability tests used when the job requires particular abilities. Physical ability tests are designed to assess a candidate’s muscular strength, movement quality, and cardiovascular endurance. Scores on physical ability tests have been linked to accidents and injuries. One study found that railroad workers who failed a physical ability test were much more likely to suffer an injury at work. Psychomotor tests assess processes such as eyehand coordination, arm-hand steadiness, and manual dexterity. Sensory/perceptual tests are designed to assess the extent to which an applicant can detect and recognize differences in environmental stimuli. These tests are ideal for jobs that require workers to edit or enter data at a high rate of speed. For example, Bank of America uses a battery of these tests to screen applicants for checking account data entry. As we discussed in Chapter 3, the validity of physical ability tests has been under close scrutiny lately, particularly with regard to their use for public safety jobs. Many lawsuits have been filed on behalf of female applicants applying for police and firefighter jobs who had failed some type of physical ability test, such as push-ups, sit-ups, or chin-ups. In fact, the probability is great for adverse impact against women when a physical ability test is used to make selection decisions.10 Sensory ability testing concentrates on the measurement of hearing and sight acuity, reaction time, and psychomotor skills, such ber87251_ch06_135-170 144 1/27/06 15:48 Page 144 PART II Acquiring Human Resource Capability as eye and hand coordination. Such tests have been shown to be related to quantity and quality of work output and accident rates. Are There Racial Differences in Test Performance? Many organizations discontinued the use of cognitive ability tests because of the Supreme Court ruling in Griggs. Despite fairly strong evidence that the tests are valid and increased use by U.S. businesses, the details of the Griggs case illustrate the continuing problem with the use of such tests. The Duke Power Company required new employees either to have a high school diploma or to pass the Wonderlic Personnel Test and the Bennett Mechanical Comprehension Test. Fifty-eight percent of whites who took the tests passed, while only 6 percent of African Americans passed. According to the Supreme Court, the Duke Power Company was unable to provide sufficient evidence to support the job relatedness of the tests or the business necessity for their use. Accordingly, the High Court ruled that the company had discriminated against African Americans under Title VII of the 1964 Civil Rights Act. As we discussed in Chapter 3, the rationale for the Supreme Court’s decision gave rise to the theory of disparate impact. The statistical data presented in the Griggs case are not unusual. African Americans, on average, score significantly lower than whites on cognitive ability tests; Hispanics, on average, fall about midway between average African American and white scores.11 See Figure 6-2 for a summary of group differences as a function of selection tool options. Thus, under the disparate impact theory of discrimination, plaintiffs are likely to establish adverse impact based on the proportion of African Americans versus whites who pass such tests. If the Griggs case wasn’t enough, the 1975 Supreme Court ruling in Albemarle Paper Company v. Moody probably convinced many organizations that the use of cognitive ability tests was too risky. In Albemarle, the Court applied specific and difficult guidelines to which the defendant had to conform in order to establish the job relatedness of the particular test. The Uniform Guidelines in Employee Selection Procedures, as issued by the Equal Employment Opportunity Commission, also established rigorous and potentially costly methods to be followed by an organization to support the job relatedness of the test if adverse impact should result. Current interest in cognitive ability tests was spurred by the research on validity generalization, which strongly supported the validity of these tests for virtually all jobs and projected substantial increases in utility for organizations that use the tests. The average validity of such tests was reported to be .51.12 (See Figure 6-2.) Some major questions remain regarding the validity generalization results for cognitive ability tests: Are these tests the most valid method of personnel selection across all job situations or are other methods, such as biographical data and personality tests, more valid for some jobs that were not the focus of previous research? Are there procedures that can make more accurate predictions than cognitive ability tests for some job situations? Are cognitive ability tests the best predictors of sales success, for example? (Remember the Unabomber? He had a Ph.D. in math from the University of Michigan. How would he do in sales?) Another issue is the extent to which validity can be inferred for jobs involving bilingual skills. Would the Wonderlic administered in English have strong validity for a job, such as a customs agent, requiring the worker to speak in two or more languages? Bilingual job specifications are increasing in the United States. Invoking Gthe “validity generalization” argument for this type of job based on research involving only the use of English is Asomewhat dubious. The validity of such tests to predict Tperformance for these jobs is probably not as strong as .5. Another issue concerns the extent to which other Emeasures can enhance predictions beyond what cognitive Sability tests can predict. Generally, human performance is to be a function of a person’s ability, motivation, , thought and personality. The highest estimate of the validity of cognitive ability tests is about .50. This means that percent of the variability in the criterion measure (e.g., D25 performance) can be accounted for by the predictor, or the Etest. That leaves 75 percent unaccounted for. Industrial Apsychologists think the answer lies in measures of one’s motivation to perform, personality, or the compatibility of Na person’s job preferences with actual job characteristics. D Would a combination of methods—perhaps, a cognitive ability test and a personality or motivational test— Rresult in significantly better prediction than the cognitive Aability test alone? Research indicates that a combination of cognitive and motivational tests may lead to a more comprehensive assessment of an individual.13 These tools 1add what is known as “incremental validity” in the prediction of job performance. In general, cognitive ability and 1job knowledge tests are valid but additional (and valid) 2tools can add validity to the prediction. Accordingly, the of other tests that address the motivational components 3use of human performance, in addition to a cognitive ability or Tknowledge-based test, can help an organization make better decisions. We will discuss these measures shortly. S Why Do Minorities Score Lower than Whites on Cognitive Ability Tests? This question has interested researchers for years; yet there appears to be no clear answer. Most HRM experts now generally take the view that these differences are not created by the tests, but are most related to inferior educational experiences. But the problem is not a defect or deficiency in the tests per se. The critical issue for HRM experts is not how to modify the test itself, but how to use the test in the most effective way. A panel of the National ber87251_ch06_135-170 1/27/06 15:48 Page 145 CHAPTER 6 Personnel Selection Academy of Sciences concluded that cognitive ability tests have limited but real ability to predict how well job applicants will perform, and these tests predict minority group performance as well as they predict the future performance of nonminorities. In other words, the tests themselves are not to blame for differences in scores. Obviously, the dilemma for organizations is the potential conflict in promoting diversity while at the same time using valid selection methods that have the potential for causing adverse impact.14 How Do Organizations Deal with Race Differences on Cognitive Ability Tests? The use of top-down selection decisions based strictly on scores on cognitive ability tests is likely to result in adverse impact against minorities. One solution to this problem is to set a cutoff score on the test so as not to violate the 80 percent rule, which defines adverse impact. Scores above the cutoff score are then ignored and selection decisions are made on some other basis. The major disadvantage of this approach is that there will be a significant decline in the utility of a valid test because people could be hired who are at the lower end of the scoring continuum, making them less qualified than people at the upper end of the continuum who may not be selected. Virtually all of the research on cognitive ability test validity indicates that the relationship between test scores and job performance is linear; that is, higher test scores go with higher performance and lower scores go with lower performance. Thus, setting a low cutoff score and ignoring score differences above this point can result in the hiring of people who are less qualified. So, while use of a low cutoff score may enable an organization to comply with the 80 percent adverse impact rule, the test will lose considerable utility. Another approach to dealing with potential adverse impact is to use a banding procedure that groups test scores based on data indicating that the bands of scores are not significantly different from one another. The decision maker then may select anyone from within this band of scores. Research shows that banding procedures have less effect on adverse impact than the characteristics of the applicant pool. Banding only has a big effect on adverse impact when minority preference within a band is used for selection. This approach is controversial and legally questionable.15 The use of cognitive ability tests obviously presents a dilemma for organizations. Evidence indicates that such tests are valid predictors of job performance across a wide array of jobs (see Figure 6-2). Employers who use such tests enjoy economic utility with greater productivity and considerable cost savings. However, selection decisions that are based solely on the scores of such tests will result in adverse impact against African Americans and Hispanics. Such adverse impact could entangle the organization in G A T E S , 145 costly litigation and result in considerable public relations problems. If the organization chooses to avoid adverse impact, the question becomes one of either throwing out a test that has been shown to be useful in predicting job performance or keeping the test and reducing or eliminating the level of adverse impact. Does such a policy leave a company open to reverse discrimination lawsuits by whites who were not selected for employment—their raw scores on the test were higher than scores obtained by some minorities who were hired? Many organizations, particularly in the public sector, have abandoned the use of cognitive ability tests in favor of other methods, such as interviews or performance tests, which result in less adverse impact and are more defensible in court. However, many other cities and municipalities have opted to keep such tests and then employed some form of banding in the selection of their police and firefighters primarily in order to make personnel decisions that do not result in statistical adverse impact. Researchers and practitioners are very interested in how to select the most effective candidates while meeting diversity goals and minimizing (or eliminating) adverse impact. Figure 6-4 presents a summary of common practices used to reduce adverse impact, the degree of support in research, and the research findings. D E What Is Personality/Motivational/ A Dispositional Testing? N While research supports the use of cognitive ability tests for D personnel selection, virtually all HRM professionals regard performance as a function of both ability and motivation. R Scores on ability tests say little or nothing about a person’s A motivation to do the job. We can all think of examples of 1 1 2 3 T S very intelligent individuals who were unsuccessful in many situations (we’re back to the Unabomber!). Most of us can remember a classmate who was very bright but received poor grades due to low motivation. The general validity of cognitive ability tests for predicting sales success is rather low and much could be done to improve prediction. Most personnel selection programs attempt an informal or formal assessment of an applicant’s motivation, attitudes, or disposition through psychological testing or a job interview. Some of these assessments are based on scores from standardized tests, performance testing such as job simulations, or assessment centers. Others are more informal, derived from an interviewer’s gut reaction or intuition. This section will review the abundant literature on the measurement and prediction of motivation, disposition, and personality using various forms of testing. There is an increased use of various types and formats for personality or motivational testings, including paper-and-pencil types, video and telephone testing, and, most recently, online testing. Some organizations place great weight on personality testing for employment decisions. BA&C, the company working with Wackenhut ber87251_ch06_135-170 146 1/27/06 15:48 Page 146 PART II Acquiring Human Resource Capability FIGURE 6-4 Practices Used to Reduce Adverse Impact Common Practices to Reduce Adverse Impact Degree of Support for Practice in Literature Research Findings Target recruitment strategies toward qualified minorities.  Characteristics of the applicant pool (e.g., proportion of minorities, average score levels of minorities) have the greatest effect on rates of adverse impact; changing these characteristics through targeted recruitment should help reduce adverse impact. However, simply increasing numbers of minorities in the pool will not help unless one is increasing numbers of qualified recruits. Use a selection system that focuses on predicting performance in areas such as helping coworkers, dedication, and reliability, in addition to task performance.  If the overall performance measure weights contextual performance (e.g., helping, reliability) more than task performance and the tests in a battery are uncorrelated, a test battery designed to predict this definition of overall performance will have smaller levels of adverse impact. Weighting task performance less than contextual performance in the overall performance measure will make cognitive ability less important in hiring and will lead to less adverse impact. Use a tool with high adverse impact and good validity in combination with a tool with low adverse impact to reduce the overall adverse impact of the system. 0 Provide orientation and preparation programs to candidates. 0 Remove cognitive ability testing from the selection process. Use banding of test scores. Use tools with less adverse impact as screening devices early in the selection process and those with greater adverse impact as later hurdles in the process. Change the more negative test taking perceptions of minority test takers about test validity, thereby increasing motivation and performance. G A T E S , 0 0 0 0 D E A N D R A 1 1 2 3 T S The degree to which adverse impact is reduced by combining tools with lower adverse impact is greatly overestimated; reductions may be small or the combination may actually increase adverse impact. Coaching and orientation programs have little effect on size of group differences but are well received by examinees. Using only noncognitive predictors (e.g., interview, conscientiousness, biodata) will lead to significantly reduced adverse impact, but significant black/white differences will remain. Also, cognitive ability tests are among the most valid predictors of job performance, and their removal may result in a selection system that is less effective. The use of banding has less effect on adverse impact than the characteristics of the applicant pool. Substantial reduction of adverse impact through banding only occurs when minority preference within a band is used for selection (i.e., preferential selection is employed). Using tools with less adverse impact as screening devices early in the process and those with greater adverse impact later in the process will aid minority hiring if the selection ratio is low, but will not have much effect if the selection ratio is high (i.e., few applicants per position). May provide a very small reduction in adverse impact. ber87251_ch06_135-170 1/25/06 09:09 Page 147 CHAPTER 6 Personnel Selection 147 FIGURE 6-4 (Continued) Identify and remove culturally biased test items. 0 Research suggests that clear patterns regarding what items favor one group or another do not exist and that removal of such items has little effect on test scores; however, item content should not be unfamiliar to those of a particular culture and should not be more verbally complex than warranted by job requirements. Use other modes of presenting test stimuli than multiple-choice, paper-and-pencil testing (e.g., video). 0 Changes in format often result in changes in what is actually measured and can be problematic; in cases where a format change was simply that (e.g., changed format without affecting what was measured), there was no strong reduction in group differences. Use portfolios, accomplishment records, and performance assessments (work samples) instead of paper-and-pencil measures. Relax time limits on timed tools. 0 0 G A T E S , Evidence suggests group differences may not be reduced by realistic assessments, and reliable scoring of these methods may be problematic. Well-developed work samples may have good validity and less adverse impact than cognitive ability tests. Research indicates that longer time limits do not reduce subgroup differences, and may actually increase them. Source: Adapted from Ryan, A. M. & Tippins, N. T. (2004). “Attracting and selecting: What psychological research tells us,” HUMAN RESOURCE MANAGEMENT, 43, pp. 312–313. Reprinted with permission of John Wiley & Sons. Security, does psychological screening for hundreds of companies using specialized reports based on the fivefactor model (FFM) of personality. One of the most popular personality assessment tools is the “Caliper Profile,” developed by the Caliper Corporation (www.calipercorp.com). Their Web site claims 25,000 clients. Avis uses the Caliper Profile to hire salespeople. Sears, Roebuck and Company, Standard Oil of New Jersey, and AT&T have used personality tests for years to select, place, and even promote employees. More companies today use some form of personality test to screen applicants for risk factors related to possible counterproductive behavior. We will begin this section with a definition of personality and provide brief descriptions of some of the more popular personality tests. We will review the validity of these tests and provide an overview of relevant legal and ethical issues. We will conclude with a description of four relatively new personality tests that have shown potential as selection and placement devices. What Is Personality? While personality has been defined in many ways, the most widely accepted definition is that personality refers to an individual’s consistent pattern of behavior. This consistent pattern is composed of psychological traits. Many researchers subscribe to a five-factor model (FFM) for describing personality.16 These so-called “Big Five” personality factors are as follows: (1) introversion/extra- D E A N D R A 1 1 2 3 T S version (outgoing, sociable); (2) emotional stability; (3) agreeableness/likability (friendliness, cooperative); (4) conscientiousness (dependability, carefulness); and (5) openness to experience (imaginative, curious, experimenting). There are several tests that measure the FFM. (Try http://users.wmin.ac./UK/~buchant/ for a free online “Big Five” test.) Two relatively new characterizations of personality are Emotional Intelligence (EI) and Core Self-Evaluations (CSE). EI is considered to be a multidimensional form or subset of social intelligence or a form of social literacy. There are many definitions and many different instruments that purport to measure EI. One definition is that EI is a set of abilities that enable individuals to recognize and understand their own emotions and those of others in order to guide their thinking and behavior to help them cope with the environment. A recent review of EI revealed an average validity of .23 in the prediction of performance. EI was also found to have low correlations with cognitive ability (.22), agreeableness (.23), and extraversion (.34) of the Big Five. Obviously, EI is another construct measure to consider in the development of (or selection of) a testing battery for job candidates.17 CSE is a broad and general personality trait composed of four heavily researched traits: (1) self-esteem (the overall value that one places on oneself as an individual); (2) self-efficacy (an evaluation of how well one can perform across situations); (3) neuroticism (the tendency to focus on the negative); and (4) locus of control (the ber87251_ch06_135-170 148 1/27/06 15:48 Page 148 PART II Acquiring Human Resource Capability Some Examples of Personality/ FIGURE 6-5 Dispositional/Motivational Tests PROJECTIVE TECHNIQUES AND INSTRUMENTS Thematic Apperception Test (TAT) Miner Sentence Completion Scale (MSCS) Graphology (Handwriting analysis) Rorschach Inkblot Test SELF-REPORT INVENTORIES—EXAMPLES The NEO Personality Inventory (FFM) Personal Characteristics Inventory Gordon Personal Preference Inventory Myers-Briggs Type Indicator Minnesota Multiphasic Personality Inventory (MHPI) California Personality Inventory (CPI) Sixteen Personality Factors Questionnaire (16 PF) Hogan Personality Inventory Job Compatibility Questionnaire (JCQ) Emotional Intelligence (e.g., EI Scale) Core Self-Evaluation Scale (CSES) Caliper Profile extent to which one believes s/he has control over life’s events). The core self-evaluation is a basic assessment of one’s capability and potential. There is some research that investigated the extent to which these new measures add predictive value (or incremental validity) beyond the Big Five or other selection tools. In general, this research indicates useful incremental validity for these measures beyond the big five and other selection models or tools. For example, research with a new instrument that purports to measure (CSE) shows scores on the scale are correlated with job performance and that CSE has incremental validity over the five-factor model.18 There are literally thousands of personality tests available that purport to measure hundreds of different traits or characteristics. (Go to www.unl.edu/buros/ for a sample). We will review the basic categories of personality testing next. Figure 6-5 presents a list of some of the most popular tests and methods. How Do We Measure Personality? Personality tests can be sorted into two broad categories: projective tests and self-report inventories. Of course, we also can use the interview and data from other sources such as performance appraisals as a means for assessing personality characteristics or competencies as well. Projective tests have many common characteristics, the most significant of which is that the purpose and scoring procedure of the test are disguised from the test taker. One of the most famous projective tests is the Rorschach Inkblot Test, which presents a series of inkblots to respondents who must then record what they see in each one. While numerous projective tests exist, the Miner Sentence Completion Scale (MSCS) is one of the few such tests specifically designed for use in the employment setting. Its aim is to measure managers’ motivation to manage others.19 And the test appears to work. The test consists of 40 incomplete sentences, such as “My family doctor . . . ,” “Playing golf . . . ,” and “Dictating letters . . . .” The test taker is instructed to complete each sentence. According to the developer of these tests, the way in which an applicant completes the sentences reflects his or her motivation along seven areas. These areas are capacity to deal with authority figures, dealing with competitive games, handling competitive situations, assertiveness, motivation to direct others, motivation to stand out in a group, and desire to perform day-to-day administrative tasks. On the downside, the MSCS is expensive and there isn’t a great deal of validity evidence to support its use. G Another projective test that has been used occasionally for employment purposes is the Thematic ApperAception Test, or TAT, a test that typically consists of a Tseries of pictures that depict one or more persons in different situations. Test takers are asked to describe who Ethe people are and what is happening in the situation, Swhich is somewhat ambiguous and open to interpretation. The test taker then determines the outcome of the situa, tion. Although a variety of scoring systems have been developed for interpreting a test taker’s responses, one of the popular approaches involves rating the responses Dmost with regard to the test taker’s need for power (i.e., the Eneed to control and influence others), achievement (i.e., to be successful), and affiliation (i.e., the need for Aneed emotional relationships). Like the MSCS, the TAT has Nbeen used primarily for managerial selection and the Dlimited research indicates some validity as a predictor of managerial and entrepreneurial success. R One form of projective test (which we alluded to Apreviously) that has received considerable attention recently is graphology, or handwriting analysis. With this approach, a sample of your handwriting is mailed to a graphologist 1who (for anywhere from $10 to $50) provides an assessment of your intelligence, creativity, emotional stability, ne1gotiation skills, problem-solving skills, and numerous other 2personal attributes. According to some writers, graphology used extensively in Europe as a hiring tool. The Wall 3isStreet Journal and Inc. magazine have reported an increase Tin the use of the method in the United States since 1989. As in The Wall Street Journal, “With the government Sdescribed pulling the plug on the polygraph, and employers clamming up on job references and liabilities from negligent hiring, it is one alternative managers are exploring in an effort to know whom they are hiring.” While the use of the method may be increasing, there is no compelling evidence that the method does anything but provide an assessment of penmanship. The only published studies on the validity of graphology have found no validity for the approach.20 Self-Report Personality Inventories Self-report inventories, which purport to measure personality or motivation with the respondent knowing the ber87251_ch06_135-170 1/27/06 15:48 Page 149 CHAPTER 6 Personnel Selection purpose and/or the scoring procedure of the test, are more popular today than projective techniques. Some instruments screen applicants for aberrant or deviant behavior (e.g., the MMPI), others attempt to identify potentially high performers, and others, particularly more recently developed tests, are directed at specific criteria such as employee theft, job tenure/turnover, accident proneness, or customer orientation.21 Self-report inventories typically consist of a series of short statements concerning one’s behavior, thoughts, emotions, attitudes, past experiences, preferences, or characteristics. The test taker responds to each statement using a standardized rating scale. During the testing, respondents may be asked to indicate the extent to which they are “happy” or “sad,” “like to work in groups,” “prefer working alone,” and so forth. One of the most popular and respected personality tests is the Minnesota Multiphasic Personality Inventory (MMPI). The MMPI is used extensively for jobs that concern the public safety or welfare, including positions in law enforcement, security, and nuclear power plants. The MMPI is designed to identify pathological problems in respondents, not to predict job effectiveness. The revised version of the MMPI consists of more than 566 statements: “I am fearful of going crazy.” “I am shy.” “Sometimes evil spirits control my actions.” “In walking, I am very careful to step over sidewalk cracks.” “Much of the time, my head seems to hurt all over.” Respondents indicate whether the statement is true, false, or cannot say. The MMPI reveals scores on 10 clinical scales, including depression, hysteria, paranoia, and schizophrenia, as well as four “validity” scales, which enable the interpreter to assess the credibility or truthfulness of the answers. Millions of people, from at least 46 different countries, from psychotics to Russian cosmonauts, have struggled through the strange questions. Litigation related to negligent hiring often focuses on whether an organization properly screened job applicants. Failure to use the MMPI in filling sensitive jobs has been cited in legal arguments as an indication of negligent hiring—although not always persuasively. Unfortunately, some companies are damned if they do and damned if they don’t. Target stores negotiated an out-of-court settlement based on a claim of invasion of privacy made by a California job candidate who objected to a few questions on the MMPI being used to hire armed guards. Had one of the armed guards who was hired used his or her weapon inappropriately (and Target had not used the MMPI), Target could have been slapped with a negligent hiring lawsuit. Another popular instrument is the 16 Personality Factors (16PF), which provides scores on the factors of the FFM, plus others. In addition to predicting performance, the test is used to screen applicants for counterproductive behavior, such as potential substance abuse or employee theft. AMC Theaters, C&S Corporation of Georgia, and the U. S. State Department are among the many organizations that use the 16PF to screen most employees. 149 Although there are many instruments available, the NEO Personality Inventory is one of the most reliable and valid measures of the FFM.22 Another very popular instrument for employee development but that is not considered a good selection instrument is the Myers-Briggs Type Indicator (MBTI).23 What Is the Validity of Personality Tests? G A T E S , D E A N D R A 1 1 2 3 T S Potentially useful personality tests exist among a great number of bad ones, making it difficult to derive general comments regarding their validity. Some instruments have shown adequate validity while others show no validity at all. One instrument with a good track record for selecting managers is the MSCS. A review of 26 studies involving the MSCS found an average validity coefficient of .35.24 In general, the validity is lower for personality tests than for cognitive ability tests. The latest review of the FFM found that Conscientiousness and Emotional Stability had useful predictive validity for all jobs but that Conscientiousness had the highest validity (see Figure 6-2). Extraversion, agreeableness, and openness to experience had useful predictive validity but for only certain situations.25 For example, extraverted workers are more effective in jobs with a strong social component, such as sales and management. More Agreeable workers are more effective team members. People with high scores on Openness to Experience are more receptive to new training. A particular combination of FFM factors also can predict important criteria. Research involving the FFM and managerial performance shows that Conscientiousness (.25), Extraversion (.21), and Emotional Stability (.24) are useful predictors of managerial success.26 Why do personality tests have such low validity relative to cognitive ability? Experts have given a number of explanations for the low (but useful) validity of such tests in the employment context. First, applicants can “fake” personality tests so their personality as reflected on the tests was compatible with the requirement of the job.27 Second, some proponents of personality testing have asserted that most of the validity studies involving personality tests are poorly designed with very small sample sizes. These experts contend that more carefully designed research would demonstrate higher validity for personality tests. Research shows the weight given to personality factors should derive from a job analysis of criterionrelated validation research. Another possible explanation is that behavior is to a great extent determined situationally, making stable personality traits unpredictable for criteria, such as job performance or employee turnover. Recall some of the examples of items from personality tests listed earlier in this chapter—note that most of the examples are not specific to the workplace; in fact, most of them are quite general. Research in other areas has found that behavior is dependent on the situation. A person who is friendly in outside work ber87251_ch06_135-170 150 1/27/06 15:48 Page 150 PART II Acquiring Human Resource Capability might be less sociable in the work setting. In order to enhance predictability, personality assessment should involve more than one method (e.g., tests, interviews). Personality assessment could be more specific to the workplace and target particular criterion measures of interest, such as employee theft, honesty or integrity, or job retention/turnover. Let us examine these newer approaches next. Approaches to the Prediction of Particular Criteria Some forms of personality, dispositional, or motivation assessment attempt to focus on either particular problems or criteria characteristic of the workplace. Examples are the prediction of voluntary turnover and the prediction of employee theft. Another instrument attempts to measure job compatibility in order to predict turnover. Other new instruments are designed for particular employment issues, such as customer service, violence, or accident proneness. Predicting (and Reducing) Voluntary Turnover Employee turnover can be a serious and costly problem for organizations. You may recall the discussion of Domino’s Pizza. They found that the cost of turnover was $2,500 each time an hourly employee quit and $20,000 each time a store manager quit. Among other things, Domino’s implemented a new and more valid test for selecting managers and hourly personnel that was aimed at predicting both job performance and voluntary turnover. As of 2005, the program was a great success on all counts. Turnover was down, store profits were up, and the stock was doing well in an otherwise difficult market. Attracting and keeping good employees was a key factor in their turnaround. There are numerous other examples of companies that have expensive and preventable high levels of turnover that can be reduced with better HR policy and practice. Recall the discussion of SAS, the North Carolina software company. Even at the height of the socalled “high tech” bubble in the late 90s, SAS had turnover rates that were well below the industry average. Attracting and keeping good employees is considered a key to the SAS success story. One recent study revealed guidlines regarding methods that have been shown to be effective at reducing voluntary turnover.28 A summary of the findings merged with previous research on turnover is presented in Figure 6-6. This most recent research drew several conclusions. First, voluntary turnover is less likely if a job candidate is referred by a current employee or has friends or family working at the organization. Candidates with more contacts within the organization are apt to better understand the nature of the job and the organization. Such candidates probably have a more realistic view of the job that may provide a “vaccination effect” that lowers expectations, thereby preventing job dissatisfaction and turnover (realistic job previews can also do this). Also, current job holders are less likely to refer job candidates who they feel are less capable or those who (they feel) would not fit in well with the organization’s culture. Another argument for an employee referral system is that having acquaintances within the organization is also likely to strengthen an employee’s commitment to the firm and thus reduce the probability that he or she will leave. Of course, this argument also applies to the employee who made the referral. Another reliable predictor of voluntary turnover is tenure in previous jobs. In general, if a person has a history of short-term employment, that person is likely to quit again. This tendency may also reflect a lower work ethic (lower Conscientiousness), which is correlated with organizational commitment and turnover. As discussed earlier, tenure in previous jobs, measured in a systematic manner as a part of a weighted application blank (WAB), Gis predictive of turnover. Intention to quit is also a solid predictor of, and perhaps the best predictor of, quitting. ABelieve it or not, questions on an application form such as T“How long do you think you’ll be working for this company?” are quite predictive of voluntary turnover. Prehire Edispositions or behavioral intentions, derived from quesStions such as this one or from interview questions, work well. Measures of the extent of an applicant’s desire , quite to work for the organization also predict subsequent turnover. However, almost all of the research on WABS, involved entry-level and nonmanagerial positions so Dhas its applicability to managerial positions is questionable. EThis is not true for biodata (or BIBs). Disguised-purpose Aattitudinal scales measuring self-confidence and decisiveness have been shown to predict turnover for higher level Npositions as well, including managerial positions. DAnswers to questions such as “How confident are you that you can do this job well?” and “When I make a deciRsion, I tend to stick to it” did predict turnover quite well. AIn addition, there was no evidence of adverse impact against protected classes using these measures. This research also revealed that disguised-purpose measures 1added incremental validity to the prediction of turnover beyond what could be predicted by biodata alone. 1 Another example of a disguised purpose disposi2tional measure is the Job Compatibility Questionnaire We discussed the JCQ in Chapters 4 and 5. The 3(JCQ). JCQ was developed to determine whether an applicant’s Tpreferences-for-work characteristics match the characterof the job.29 One theory is that the compatibility of Sistics preference with the job will predict job tenure and performance. Test takers are presented groups of items and are instructed to indicate which item is most desirable and which is least desirable. As we discussed in Chapter 4, the items are grouped based on a job analysis that identifies those characteristics that are common to the job(s) to be filled. Here is an example of a sample group: (a) being able to choose the order of my work tasks, (b) having different and challenging projects, (c) staying physically active on the job, (d) clearly seeing the effects of my hard work. The items are grouped together in such a way that the scoring key is hidden from the respondent, reducing the chance for faking. ber87251_ch06_135-170 1/27/06 15:48 Page 151 CHAPTER 6 Personnel Selection 151 FIGURE 6-6 Predictors of Voluntary Turnover 1. Rely on employee referrals Voluntary turnover is less likely if a job candidate is referred by a current employee or has friends or family working at the organization. Candidates with more contacts within the organization are apt to better understand the nature of the job and the organization. Having friends or family within the organization prior to hire is likely to strengthen the employee’s commitment to the firm and reduce the likelihood that he or she will leave. 2. Put weight on tenure in previous jobs A past habitual practice of seeking out short-term employment predicts future short-term employment. Short-term employment may reflect a poor work ethic, which is correlated with lack of organizational commitment and turnover. G A of turnover. Intention to quit is one of the best (if not the best) predictors Despite their transparency, expressions of intentions to stayTor quit before a person starts a new position are an effective predictor of subsequent turnover. (e.g., how long do you plan to work for the company?) E Measure the applicant’s desires/motivations for the position S New employees with a strong desire for employment will require less time to be assimilated into the organization’s , culture. 3. Measure intent to quit 4. 5. Use disguised-purpose dispositional measures High self-confidence should respond more favorably to theD challenges of a new environment. Employees with higher confidence in their abilities are lessElikely to quit than those who attribute their past performance to luck. Decisive individuals are likely to be more thoughtful aboutAtheir decisions, more committed to the decisions they make, and less likely to leave the organization. N Decisiveness is a component of the personality trait of Conscientiousness from the five-factor model. D turnover. Decisiveness affects organizational commitment and, indirectly, R A Source: Adapted from M. R. Barrick and R. D. Zimmerman, “Reducing Voluntary, Avoidable Turnover through Selection,” Journal of Applied Psychology 90 (2005), pp. 159–166. Studies involving customer service representatives, security guards, and theater personnel indicate that the JCQ can successfully predict employee turnover for lowskilled jobs. In addition, no evidence of adverse impact has been found. BA&C incorporated the JCQ in their test for security guards.30 The JCQ has never been used or validated for managerial positions. Predicting Employee Theft More than five million job applicants took some form of honesty or integrity test in 2005. These tests are commonly used for jobs in which workers have access to money, such as retail stores, fast-food chains, and banks. Honesty tests have become more popular since the polygraph, or lie detector, test was banned in 1988 by the Employee Polygraph Protection Act. This federal law outlawed the use of the polygraph for selection and greatly restricts the use of the test for other employment situations. There are some employment exemptions to the law, such as those involving security services, businesses involving controlled substances, and government employers. 1 1 2 3 T S Most honesty tests contain items concerning an applicant’s attitude towards theft. Sample items typically cover beliefs about the amount of theft that takes place, asking test takers questions such as the following: “What percentage of people take more than $1.00 per week from their employer?” The test also questions punitiveness towards theft: “Should a person be fired if caught stealing $5.00?” The test takers answer questions reflecting their thoughts about stealing: “Have you ever thought about taking company merchandise without actually taking any?” Other honesty tests include items that have been found to correlate with theft: “You freely admit your mistakes.” “You like to do things that shock people.” “You have had a lot of disagreements with your parents.” Many banks and retail establishments use honesty tests for employee screening. The validity evidence for honesty tests is fairly strong, with no adverse impact. Still, critics point to a number of problems with the validity studies. First, most of the validity studies have been conducted by the test publishers themselves; there have been very few independent validation studies. Second, very few of the criteria-related ber87251_ch06_135-170 152 1/25/06 09:09 Page 152 PART II Acquiring Human Resource Capability validity studies use employee theft as the criterion. A report by the American Psychological Association concluded that the evidence, albeit limited, supports the validity of some of the most carefully developed and validated honesty tests. The most recent studies on honesty tests support their use.31 Predicting Accident Proneness Accidents are a major problem in the workplace, causing deaths, injuries, and expense. Preemployment testing is one strategy some companies have turned to in an effort to lower accident rates. One test developed to predict (and prevent) accidents is the Safety Locus of Control (SLC), which is a paper-and-pencil test containing 17 items assessing attitudes towards safety. A sample item is as follows: “Avoiding accidents is a matter of luck.” The limited validity studies have been encouraging. Such studies have been conducted in several different industries, including transportation, hotel, and aviation. In addition, these investigations indicate no adverse impact against minorities and women.32 Predicting Customer Service The Service Orientation Index (SOI) was initially developed as a means of predicting the helpfulness of nurses’ aides in large, inner-city hospitals.33 The test items were selected from three main dimensions: patient service, assisting other personnel, and communication. Here are some examples of SOI items: “I always notice when people are upset” and “I never resent it when I don’t get my way.” Several other studies of the SOI involving clerical employees and truck drivers have reported positive results as well. How Do You Establish a Testing Program? Establishing a psychological testing program is a difficult undertaking—one that should involve the advice of an industrial psychologist. HR professionals should follow these guidelines before using psychological tests: 1. Most reputable testing publishers provide a test manual. Study the manual carefully, particularly the adverse impact and validity evidence. Has the test been shown to predict success in jobs similar to the jobs you’re trying to fill? Have adverse impact studies been performed? What are the findings? Are there positive, independent research studies in scholarly journals? Have qualified experts with advanced degrees in psychology or related fields been involved in the research? 2. Check to see if the test has been reviewed in Mental Measurements Yearbook (MMY). Published by the Buros Institute of the University of Nebraska, the MMY publishes scholarly reviews of the test by qualified academics who have no vested interest in the tests they are reviewing. You can also download Buros test reviews on line at www.unl.edu/buros. You can retrieve reviews by test name or by category (e.g., achievement, intelligence, personality). 3. Ask the test publishers for the names of several companies that have used the test. Call a sample of them and determine if they have conducted any adverse impact and validity studies. Determine if legal actions have been taken related to the test; if so, what are the implications for your situation? 4. Obtain a copy of the test from the publisher and carefully examine all of the test items. Consider G each item in the context of ethical, legal, and privacy ramifications. Organizations have lost court A cases because of specific items on a test. T Proceed cautiously in the selection and adoption of Epsychological tests. Don’t be wowed by a slick test brochure; take a step back and evaluate the product in the Ssame manner you would evaluate any product before buy, ing it. Be particularly critical of vendors’ claims and remember that you might be able to assess personality and motivation by other means. If you decide to adopt a test, Dmaintain the data so that you can evaluate whether the test working. In general, it is always advisable to contact Eissomeone who can give you an objective, expert appraisal. A N D RUG T ESTING D Drug abuse is one of the most serious problems in the RUnited States today, with productivity costs in the billions Aand on the rise. Drug abuse in the workplace also has been linked to employee theft, accidents, absences, use of sick time, and other counterproductive behavior. Detected am1phetamine use doubled between 2000 and 2004. Methamphetamine is the most commonly used form of ampheta1mine today. To combat this growing problem, many 2organizations are turning to drug testing for job applicants incumbents. One survey found 87 percent of major 3and U.S. corporations now use some form of drug testing.34 T While some of the tests are in the form of paper and examinations, the vast majority of tests conducted Spencil are clinical tests of urine or hair samples. Ninety-six percent of firms refuse to hire applicants who test positive for illegal drug use, methamphetamines, and some prescription drugs (e.g., oxycontin). While the most common practice is to test job applicants, drug testing of job incumbents, either through a randomized procedure or based on probable cause, is also on the increase. The most common form of urinalysis testing is the immunoassay test, which applies an enzyme solution to a urine sample and measures change in the density of the sample. The drawback of the $20 (per applicant) immunoassay test is that it is sensitive to some legal drugs as ber87251_ch06_135-170 1/27/06 15:48 Page 153 CHAPTER 6 Personnel Selection well as illegal drugs. Due to this problem, it is recommended that a positive immunoassay test be followed by a more reliable confirmatory test, such as gas chromatography. The only errors in testing that can occur with the confirmatory tests are due to two causes: positive results from passive inhalation, a rare event (caused by involuntarily inhaling marijuana), and laboratory blunders (e.g., mixing urine samples). Hair analysis is a more expensive but also more reliable and less invasive form of drug testing. Testing for methamphetamine use is more difficult since the ingredients pass through the body quickly. Positive test results say little regarding one’s ability to perform the job, and most testing gives little or no information about the amount of the drug that was used, when it was used, how frequently it was used, and whether the applicant or candidate will be (or is) less effective on the job. The legal implications of drug testing may have changed significantly since this chapter was written. Currently, drug testing is legal in all 50 states for preemployment screening and on-the-job assessment; however, employees in some states have successfully challenged dismissals based solely on a random drug test. For those employment situations in which a collectivebargaining agreement has allowed drug testing, the punitive action based on the results is subject to arbitration. One study found that the majority of dismissals based on drug tests were overturned by arbitrators.35 Among the arguments against drug testing are that it is an invasion of privacy, it is an unreasonable search and seizure, and it violates the rights of due process. Most experts agree that all three of these arguments may apply to public employers, such as governments, but do not apply to private industry. State law is relevant here since some drug testing programs have been challenged under privacy provisions of state constitutions. With regard to public employment, the Supreme Court has ruled that drug testing is legal if the employer can show a “special need” (e.g., public safety). We will explore the matter of drug testing in more detail in Chapter 14. Is Testing an Invasion of Privacy? Some have critiqued the widespread use of employment tests on the grounds that these procedures may be an invasion of an individual’s privacy and produce information that will affect an individual’s employment opportunities. Some types of selection methods that seem particularly prone to these concerns are drug tests and honesty/ integrity tests.36 Questions on tests or interviews that are political in nature are also illegal in some states. Experts in the field of employment testing who support the use of these types of selection procedures have responded to the challenges in a number of ways. First, various professional standards and guidelines have been devised to protect the confidentiality of test results. Second, almost any interpersonal interaction, whether it be an interview or an 153 informal discussion with an employer over lunch, involves the exchange of information. Thus, advocates of employment testing contend that every selection procedure comprises some invasion of the applicant’s privacy. Finally, in the interests of high productivity and staying within the law, organizations may need to violate an individual’s privacy to a certain degree. Companies with government contracts are among those that are obliged to maintain a safe work environment and may need to require drug testing and extensive background checks of employees. There are those who will continue to voice concern over the confidentiality and ethics of employment testing, particularly as computer-based databases expand in scope and availability to organizations. It also is likely that there will be increasing calls for more legislation at federal, state, and local levels to restrict company access to and use of employment-related information. G A T E P ERFORMANCE T ESTING /W ORK S S AMPLES 37 , Despite making valuable contributions to employee seD E A N D R A 1 1 2 3 T S lection, paper-and-pencil tests have their problems and limitations. The validity of cognitive ability tests is clear. Unfortunately, the potential legal implications that stem from their use is considerable. As we discussed, the validity of paper-and-pencil measures of applicant motivation or personality is not nearly as impressive. Many experts suggest that the prediction of job performance can be enhanced through performance testing or work samples that involve samples of actual or simulated job tasks and/or behaviors. There is also evidence that the use of such tests can result in less adverse impact than cognitive ability tests (see Figure 6-2). Performance testing is usually more complex than paper-and-pencil testing in that behavioral responses are required by test takers that are similar to the responses required on the job. A work sample consists of tasks representing the type, complexity, and difficulty level of the activities that are required on the job. Applicants must demonstrate that they possess the necessary competencies or skills needed for successful job performance. The most obvious example of a work sample is a word processing test for clerical personnel. More complex examples attempt to simulate what managers must do on the job. Assessment centers, for example, often entail several work samples or simulations of on-the-job behaviors typically exhibited by managers. The objective of performance testing is to assess candidates’ ability to do the job. Thus, applicants for clerical positions may be required to take word processing (typing) tests or demonstrate proficiency in shorthand or filing. These exercises are work samples because word processing, shorthand, and filing are representative of t...
Purchase answer to see full attachment