Recruitment and
Selection 2
(Usefulness of a selection procedure)
Usefulness of a Selection Procedure/Predictor
•
Concerned with the accuracy of prediction.
•
It is about how a predictor is good in enhancing selection
decisions.
•
Thus how well a selection tool predicts a job behavior
•
Some factors used to assess the usefulness of a selection
technique
•
Reliability/Validity
•
Base Rate
• The main reason why a selection procedure is developed is to show
that that procedure will bring out successful performers on the job.
• Selection system/procedures/techniques must be reliable and valid.
Terms to note
• Performance predictors / Predictors(x)- selection techniques, e.g.
employment tests, interviews, work experience, recommendation letters.
• Criterion (y)- standard for assessing /determining successful
performance on the job
• The required standard for accepting that performance is satisfactory
or not
Reliability and Validity Issues in Selection
• Reliability
• The extent to which the measurement tool produces a consistent score
or a set of scores.
Illustrations on reliability – Class Activity
• Test retest- administration of same test to same sample twice at
different times
• Split half - administration of test to sample once but items are split
into two halves
• Parallel forms – administration of two different equivalent tests to
same sample
• There are several computer softwares that can be used for this
Reliability and Validity Issues in
Selection
Reliability and Validity Issues in Selection
Validation(Cont.)
Validity
Tells whether the measure is right for the purpose.
Measurement tools or selection procedures must be valid.
Validation is the process of establishing validity of the selection techniques/procedures.
Four Ways of assessing validity
•
Construct validity
•
Content validity
•
Criterion validity
Reliability and Validity Issues in
Selection Validation(Cont.)
Construct validity:
• Constructs are inferences about psychological phenomena that
cannot be directly measured so they have to be translated into attitude or behavior in operational terms that can be measured.
• The test of construct validity is to determine whether the indirect
Reliability and Validity Issues in
Selection Validation(Cont.)
Content validity:
•
The selection procedure should cover all the indicators of
Reliability and Validity Issues in
Selection Validation (Cont.)
Face validity
•
Whether a measure appears relevant to the test domain.
Some tests are more obviously job relevant (eg., work
samples involving concrete tasks) and others are not (eg.,
personality tests/inventories).
•
The importance of face validity is mainly for how applicants
Reliability and Validity Issues in
Selection Validation (Cont.)
Criterion validity:
• Pertains to whether the measurement tool or the predictor can
account for significant and concomitant variation in the criterion.
• Extent to which a selection device makes accurate forecast/prediction
about future job performance.
• A predictor has criterion validity if it yields scores that predict job
success.
• Criterion validation is the process of establishing relationship between
Estimating the Validity of Personnel Selection Procedures (Cont.)
•
In establishing the relationship between the predictor (x) and
the criterion (y), the ‘r’ calculated (Pearson’s moment
correlation coefficient) is the validity coefficient.
•
The stronger the magnitude of the ‘r’, the stronger the
predictor in predicting/forecasting future job performance.
•
Two ways of criterion validation
•
Predictive Validation
Predictive
validity/validation
•
Involves testing job applicants and then correlating their test
scores (x) with their criterion or on-the-job performance
scores (y) later when they are employed, to obtain the validity
coefficient (r).
•
I.e. potential selection technique (eg. an aptitude/ cognitive
ability test) is administered to all job applicants. When the
Predictive validity/validation
(Cont.)
•
If a high degree of correlation or relationship exists between
the scores, i.e. (predictor) and later job performance,
Concurrent
validity/validation
•
By contrast, involves testing current job holders and correlating
their test scores (x) with their performance scores (y) to obtain
the validity coefficient (r).
•
I.e. current job holders instead of applicants, are tested and
Concurrent validity/validation
(Cont.)
•
If the predictor scores are highly correlated to the job
performance, i.e. criteria scores, (obtained from performance
appraisal results) the technique may be useful for selecting
Class Discussion
•
Does HR practitioners have to go through all these
steps before selecting worker?
•
To what extent do you agree that if the performance
Prediction of Criterion Scores from Selection Test Scores
• Validity Co-efficient- Correlation between predictor scores and criterion
scores
The accuracy of the selection methods is obtained by assessing the criterion-related validity of the predictors
The extent to which variations in x scores lead to concomitant variations in y scores determine the criterion related validity
When the correlation between two variables ( x & y) is high, it is possible to predict the score of one (y) given the other (x).
When a selection test score (x) for an applicant is known , the criterion (y), i.e. predicted performance can be obtained.
Prediction of Criterion Scores from
Selection Test Scores(Cont.)
Two ways of predicting criterion scores The graphical approach
Generic regression equations approach
• 1) Graphical Approach
• The criterion-related validity of a selection procedure is
normally indicated by the magnitude of the validity coefficient (the correlation between the predictor (X) and criterion (Y)
Hypothetical Predictor and criterion Data on 16 Employees obtained using Concurrent/Predictive Validation in an Organisation
Predictor (X) Criterion (Y)
Graphical Approach
Y C ri te ri a50 80 100 150 200
Prediction of Criterion Scores from
Selection Test Scores(Cont.)
2) Regression Approach A. Single predictor
• A Predictor can be used for selection purposes by developing prediction
or regression equation
• The prediction equation is used for selection among future applicants on
whom we have no job performance data. It enables us to predict applicant’s job performance before selection
To make the predictions, the correlation or validity coefficient is
Simple Regression Approach
C ri te ri o n Predictor Score a b X ∆ X ∆ YRegression (Prediction Line) Y=a +bx
Simple Regression
Approach
Predicting Performance/ Criterion
Score (Example)
• Y= a +bx
• Given that a = 25 (intercept on Y, see graph)
• b = 0.7 (validity coefficient of Predictor X computed from x and y values)
• Predictor Equation
• Y = 25 + 0.7X (Substituting a and b values)
• Assuming a candidate scored 100 marks on a test (predictor X) • Then the predicted criterion score, Y?
• Y = 25 + 0.7x100
• = 25 + 70 = 95
Limitations in the use of single predictor
• In predicting job success the greater the absolute value of r, (validity
coefficient) the better the prediction of a criterion performance ,given a knowledge of predictor performance.
• The square of r (r²) gives variance in the criterion accounted for a given
predictor.
• Assuming a predictor (selection test) – criterion correlation i.e., the validity
coefficient r of a single predictor is 0.70, r² = .49 That is, 49% of the variance in the criterion may be determined /explained by the predictor (test).
• The statistic r² =coefficient of determination. The remaining 51% could be
due to other factors (predictors) ; the need for the use of multiple predictors, e.g., interviews, biographical data ,recommendation letters, etc. in the
B. Multiple predictors- Multiple Regression Approach
• If we generate two or more sets of scores based on two or more
predictors (for example, selection test, interview and recommendation/ reference letter), we can derive a ‘multiple’ regression equation
• An example of a multiple regression equation: • Y = a + b1X1 + b2X2 + b3X3 + … bnXn
Where
• Y = the predicted criterion or job performance.
• a = the constant, “Y–intercept”.
• b1, b2 and b3 = the constant regression coefficients (validity coefficients)
Multiple predictors- Multiple
Regression Approach
• This approach is a compensatory process/model.
• designed to recognize that applicants’ limitations on some
qualification can be counter – balanced by strengths in others.
• In addition to minimizing errors in prediction, the model combines
Multiple Cut – off Approach / Multiple
Hurdles Process
• assumes that an applicant’s strengths and weaknesses do not
balance each other
• Selection is then made from the group of applicants who meet
Multiple Cut – off Approach /
Multiple Hurdles Process (Cont.)
Appropriate when :
• Stringent procedures are required
• You have a large applicant with low selection ratio
• Compensation is claimed inappropriate, for example, visual acuity,
auditory acuity, manual dexterity, high degree of eye – hand co-ordination are needed for aircraft pilot success.
• It is essential that every single hurdle must be job – oriented and free from
Hybrid Approach / Process
• In this process, multiple hurdles and compensatory logic are both
used.
• For most jobs, certain minimum qualifications are required for
successful performance, e.g. college degree, typing speed, two year experience etc.
• Usually the hybrid process begins with hurdles to screen out those
Usefulness Utility of a Selection
Procedure
(Cont.)
Validity
• This is the relationship between the selection and procedure
(predictor) and the job behavior (criterion).
• The correlation between the predictor and the criterion
variable- the validity coefficient (r) in selection design computed using Pearson r.
• The stronger or higher the validity coefficient of the predictor
100
75
50
25
50 80 100 150 200
low high Selection Procedure
Usefulness Utility of a Selection Procedure
Base Rate
• Base Rate (BR)
• -a measure of the quality of the applicant pool. It is the proportion of applicants who will be satisfactory performers on the job. (See figure)
BR = No. of applicants who will be satisfactory Performers on the job/ Total no. of applicants
= A + D
A + B + C + D
• Significance of the Base Rate
If all the applicants are satisfactory, (base rate = 1) then there is no need for a selection technique no matter how valid. Anyone that the recruiters turn up will be satisfactory.
Usefulness Utility of a Selection Procedure
(Selection Ratios)
• Selection Ratio (SR)
• The proportion of applicants hired/selected for a job. The ratio of the number of available job openings /vacancies/ number of applicants offered jobs to the total of available applicants. (See figure )
• SR = Number of applicants offered jobs
Total number of applicants
• = A+B A+B+C+D
• A very high selection ratio means that more of the applicants are required to fill job
vacancies. As the selection ratio increases, the selection procedure has less
Correct Selection
Decisions and Errors
• Correct Selection Decisions and Errors
D = False negatives (error)
Rejection of persons who turn out to be successful
B = False positives (error)
Selection of persons who turn out to be unsuccessful
C = True negatives (correct decision) Rejecting poor performers
Usefulness /Utility of a Selection Procedure
(Cont.)
• There are two objectives of the selection procedure
• To increase the proportions of correct decisions (A and C)
• To minimize the chance of making the two types of errors (D
and B)
• Main concern is to avoid the last error B
• The cost of hiring a poor performer would have enormous
consequences to the organization
• Designers of the selection procedure attempt to minimize the