Job specifications based on statistical analysis

In doing the job analysis, don’t miss the forest for the trees. Consider a study of 50 testing engineers at a Volvo plant in Sweden. When asked what determined job competence for a testing engineer, most of the engineers focused on traditional criteria such as to make the engine perform according to specifications. But the most effective testing engineers defined the job’s main task differently: to make sure the engine provides a customer with a good driving experience. As a result these engineers went about their jobs testing and tuning the engines not as engineers trying to hit a number, but as ordinary drivers — imagining themselves as seniors, students, commuters or vacationers. This subgroup of the testing engineers worked hard to develop their knowledge of customers’ driving needs, even when it meant reaching out of people outside their own group, such as designers or marketers. So, the job specifications for effective engineers tuned out to be quite different than the initial survey revealed.

The point says the researcher is that in developing the job description and job specification make sure you really understand the reason for the job and therefore the skills a person actually needs to be competent at it.

Basing job specifications on statistical analysis is the more defensible approach, but it’s also more difficult. The aim here is to determine statistically the relationship between (1) some predictor for human trait, such as height, intelligent, or finger dexterity; and (2) some indicator or criterion of job effectiveness, such as performance as rated by the supervisor.

The procedure has five steps: (1) analyze the job and decide how to measure job performance; (2) select personal traits like finger dexterity that you believe should predict successful performance; (3) test candidates for these traits; (4) measure these candidates’ subsequent job performances; and (5) statistically analyze the relationship between the human trait (finger dexterity) and job performance. Your objective is to determine whether the former predicts the latter.

This method is more defensible than the judgmental approach because equal rights legislation forbids using traits that you can’t prove distinguish between high and low job performers. For example, hiring standards that discriminate based on sex, race, religion, national origin, or age may have to be shown to predict job performance. Ideally this is done with a statistical validation study, as in the five step approach above. In practice most employers probably rely more on judgmental approaches.

Many employers and managers turn to the Web for a practical approach for creating job descriptions and specifications as the “When you’re on Your Own feature illustrates”.

Job analysis in a jobless world:

Job is generally defined as a set of closely related activities carried out for pay, but over the years the concept of a job has actually changed quite dramatically. In nutshell, jobs tend to be much more varied and loosely defined than in the past. For example, when an employer like Daimler-Chrysler moves from traditional assembly line production to using self managing teams, the employees jobs moves from narrowly defined to broad and flexible (some all this de-jobbing). This obviously has ramification for what job description look like. The New Workforce feature blow illustrates this.

A (Very) Brief History: From Specialized to enlarged jobs

The term job as we know it today is largely an outgrowth of the industrial revolution’s emphasis on efficiency. During this time, people like economist Adam Smith and consultant Frederick Taylor wrote enthusiastically of the positive correlation between specialized jobs (doing the same small thing over and over) and efficiency. Jobs and job descriptions until quite recently, tended to follow their prescriptions and to be fairly detailed and specific.