Driver Selection Tests and Measurement (2012)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

67 GLOSSARY Ability test—A test that measures the current performance or estimates future performance of a person in some defined area of cognitive, psychomotor, or physical func- tioning (DOL 2000). Adverse impact—A situation in which members of a par- ticular race, sex, or ethnic group have a substantially lower rate of selection in hiring, promotion, or other employment decisions (DOL 2000). Assessment—Any test or procedure used to measure an individual’s employment or career-related qualifications or characteristics (DOL 2000). Associated factors (e.g., in the LTCCS)—Human, vehicle, or environmental conditions present at the time of the crash. Associated factors are not direct crash causes but are often viewed as contributing factors. Attitude—An individual’s positive or negative evaluations of a particular thing (person, topic, country, activity, etc.). Most important here are attitudes toward driving behaviors. Attitudes have cognitive (knowledge, belief) and emotional components, and are reflected in behav- iors. Safety-related attitudes are persistent and thus are a potential basis for driver selection. On the other hand, attitudes may change based on new knowledge, experi- ence, and maturation. Attribution bias—The strong tendency of most people to attribute their own behavior to situational factors while attributing the behavior of others to internal factors (e.g., their character, personality, abilities). Basic skills tests—Assessments of competence in reading, simple mathematics, and other skills that are widely required in training and employment settings (DOL 2000). Biodata—Information on personal characteristics, including physical, medical, and behavioral history information. Chronotype—A person’s fatigue susceptibility and sleep- and alertness-related characteristics. Although the same general factors affect people’s alertness levels, there are also significant individual differences, especially in vul- nerability to drowsiness. Construct—A concept or explanatory label for a personal characteristic that is not directly observable or that can- not be captured by a single observation or measure. For example, people skillful in reasoning and complex thought are considered high on the construct mental abil- ity. Mental ability is not directly visible, but its manifesta- tions and its significance for occupational success are easy to recognize. Construct validity—The degree to which a measure of a specific personal characteristic (e.g., constructs such as “mental ability,” “impulsivity,” and “agreeableness.”) is known to be relevant to the performance of a job. Content validity—The degree to which the content of a test corresponds to the knowledge or behavior content of a job. For example, an on-road assessment has high content validity in relation to on-the-job driving. Correlation—The degree of association or predictability between two variables within the same group of subjects (e.g., drivers). Examples include the correlation between sets of test scores, or between test scores and job perfor- mance measures. Correlation coefficient—A statistic summarizing direction and degree of association. Correlation coefficients range from −1.0 (a perfect inverse relation) through zero (no statistical association) to +1.0 (a perfect linear relation). Criterion—Any measure of work behavior or any outcome that can be used as the standard for successful job perfor- mance. Relevant examples include driver crash rate, vio- lation rate, tenure with company, or supervisory ratings of performance as a driver. Criterion-based validity—The degree to which test scores correlate with actual job performance criteria. Includes predictive validity (predicting future performance) and concurrent validity (correlates with current performance). Critical Reason (CR)—In the LTCCS, the human, vehicle, or environmental failure leading to the critical event and thus to the crash. The immediate or proximal cause of a crash. Differential driver risk—Persistent individual differences among drivers in crash risk. Related to various personal traits such as age, personality, character, medical condi- tions, other physical variations, and performance capabilities. Inventory—A questionnaire or checklist that elicits infor- mation about an individual in such areas as work values, interests, attitudes, and motivation (DOL 2000). Job analysis—Defining and describing a job in terms of the behaviors necessary to perform it. Includes job tasks and knowledge, skills, and attitudes necessary for successful performance. Mean—The arithmetic average score in a group of scores, computed by adding all the scores and dividing the sum by the number of cases.

68 Median—The middle score in a group of scores. The point or score that divides the group into two equal parts. Also known as the 50th percentile. Multiple-hurdles approach—An approach to personnel assessment that requires a candidate to pass all tests in sequence in order to qualify (DOL 2000). Normal distribution—The “bell-shaped curve” character- izing the distribution of many human traits such as height (within either gender), IQ score, and manual dexterity. Driver risk is not normally distributed. Normative score—A test score stated in relation to a peer group; for example, a percentile score in relation to other commercial drivers (DOL 2000). Norms—Descriptive statistics that are used to summarize the test performance of a specified group, such as a sam- ple of workers in a specific occupation. Norms are often assumed to represent a larger population, such as all workers in an occupation (DOL 2000). Odds ratio—A statistic often used to quantify relative risk or occurrence of an outcome for two different situations or groups. An odds ratio greater than 1.0 implies overin- volvement (e.g., in driving incidents), whereas an odds ratio less than 1.0 implies underinvolvement. Percentile score—The score on a test below which a given percentage of scores fall. For example, a score at the 65th percentile is equal to or higher than the scores obtained by 65% of the people who took the test (DOL 2000). Personality—Individual behavioral or psychological con- sistency over time and across different types of situa- tions. Style of interaction with other people and life situations. Examples include aggressiveness, impulsivity, sensation-seeking, extraversion-introversion, conscien- tiousness, and agreeableness. Reference group—The group of individuals used to develop a test; for example, commercial drivers, commercial drivers meeting some performance criterion. Reliability—The degree to which test scores are consistent, dependable, or repeatable. Reliability coefficient—A correlation coefficient indicating the degree to which two sets of test scores are associated or repeatable. Risk factor—Any prior factor—driver, vehicle, environ- mental, carrier—that affects the probability of a crash. Risk perception—A complex cognitive process represent- ing the level of perceived risk that drivers use to calibrate their risk-taking behaviours (Thiffault 2007). Selection ratio—In hiring, the ratio of job hires to job appli- cants. Other factors being equal, a low selection ratio (i.e., more selective hiring) results in higher average on-job performance of new hires. Selection success ratio—Conceptually, the percentage of correct decisions made in hiring. Specifically, this is the sum of the correct acceptances (hired employees who perform well) and correct rejections (nonhired who would have performed poorly) divided by all applicants. In practice, the success ratio cannot be calculated, but it is a useful concept for understanding employee selection. Sensitivity (test)—The ability of a test to correctly identify and reject unsafe or otherwise unsatisfactory drivers. In other words, the probability of driver failure in a criterion measure (e.g., on the job) given a test prediction of failure. Skewed distribution—A lopsided distribution in which there are more individuals at one end than the other. This is contrast to the normal distribution or “bell-shaped curve,” which is symmetrical with most people in the middle. For drivers in general and within almost any sub- group (e.g., a fleet), there are typically many relatively low-risk drivers, some drivers of medium risk, and a few drivers of much higher risk. Specificity (test)—The ability of a test to correctly identify and accept safe or otherwise satisfactory drivers. In other words, the probability of driver success in a criterion measure (e.g., on the job) given a test prediction of success. Standard deviation—A statistic used to describe the vari- ability within a set of scores. It indicates the extent to which scores vary around the mean or average score. Standardized test—A test developed using professionally prescribed methods which provides specific administra- tion requirements, instructions for scoring, and instruc- tions for interpreting scores (DOL 2000). Test—Any instrument or procedure that samples behavior or performance. A personnel or employment test is the general term for any assessment tool used to measure an individual’s employment qualifications, capabilities, or characteristics (DOL 2000). Traits vs. states—Traits are enduring personal characteris- tics (e.g., medical conditions, personality), whereas states are temporary characteristics (e.g., short illness, moods) that may reflect recent events. Validity—The degree to which an assessment actually mea- sures what it purports to measure. A test’s validity is determined in contexts such as content validity, construct validity, and criterion-based validity (predictive or concurrent). Validity coefficient—A numeric index that shows the strength of the relationship between a test score and a criterion, such as job performance. Expressed as a cor- relation between predictor(s) and job performance, and sometimes called a V-score.