Q.13. What do you mean by Basic Selection Model? What do you mean by Cross Validation of Selection Test? (AKTU. 2010 - 11)
Ans. Basic Selection Model: -
The purpose of the selection process is to take full advantage of the individual differences in order to select primarily those persons who possess the greatest amount of attributes important for success on the job.
Each step of basic selection model is important in developing a proper selection procedure.
Step 1: Examination of the job(s) having vacancies: - A thorough knowledge and understanding of job is of paramount importance and must precede the use of any test in the selection and placement of workers.
Step 2: Selection of criterion and predictor: - The second step involves two parts— choosing an indicator which measures the extent of how “good” or successful worker is (typically referred to as the criterion) and choosing a particular measure that can be used to predict how successful a worker will be on the job (typically referred to as the selection device or predictor
Step 3: Measurement of performance: - Once the criterion and the predictor have been selected it is necessary to obtain measures on both from a sample of workers on the job. This can be done either by giving the predictor to present employees and simultaneously obtaining criterion measures.
Step 4: Relating predictor to criterion: - The fourth step in the selection process involves determining whether a true and meaningful relationship exists between the employee scores on the predictor and the criterion. Only if such a relationship exists can the selection process be considered successful. Establishing the existence of such a relationship is called assessing the validity of a predictor.
Step 5: Deciding upon the utility of the selection device: - The predictor to select new job hires depends not only upon the size of the relationship found (in step 4) and its significance, but also upon many other conditions: the number of applicants, number of job openings, proportion of present employees considered successful (the base rate), and respective variances of the successful and unsuccessful worker groups.
Basic Selection Model
Step 6: Reevaluation: - The fact that the predictive situation is a dynamic ever-changing, applicants change, job change, and employment conditions change. Thus any good selection program should be reevaluated periodically to make certain it is doing the job for which it has been designed.
Cross-Validation Of Selection Test: -
One feature of most multiple prediction selection systems is that in their development, the chance variation that exists in the sample of employees being used for purposes of validation. This is particularly true with the multiple regression model, but applies as well to the multiple cutoff procedure. Because the multiple regression model has least-square properties, i.e., we deliberately minimize the errors in predicting our particular sample, it is likely that if we now apply our equation to a new sample (from the same population) we will not find our prediction as efficient as before. Thus, our computed R2 is an overestimate of what the future validity of our prediction system is apt to be, since using our equation for purposes of prediction automatically implies applying it to new samples of workers. This expected drop in is known in statistics as the shrinkage problem and can best be illustrated by examining figure.
There are formulas available for estimating the amount of shrinkage one can expect when using this equation on a new sample. One such formula is
where
= shrunken multiple correlation squared
= shrunken multiple correlation squared
R2 = multiple correlation squared obtained from validation sample
n = number of people in validation sample
k = number of predictors in regression equation
