Effectiveness of recruitment

Bayesian analysis method of recruitment effectiveness in state-owned enterprises

Recruitment is one of the important functions of human resource management, which affects the overall quality level of human resources in enterprises from the source. In order to successfully find suitable candidates for enterprises, HR people have made great efforts in recruitment methods. Various interview methods and evaluation techniques are constantly enriching the theory and practice of human resource management.

After years of recruitment practice and theoretical study, enterprises and HR have more or less formed a set of their own recruitment concepts and methods. So how to evaluate whether these methods are really effective? Although some indicators such as employment rate, completion rate, application rate and employment rate can be used to measure the recruitment of human resources departments, we should consider the long-term and short-term goals and cost control of enterprises. However, these indicators still cannot show the effectiveness of the recruitment methods adopted by enterprises. Here, the author briefly introduces the Bayesian analysis method to evaluate the effectiveness of enterprise recruitment, hoping to play a role in attracting jade.

Bayesian analysis method is a probability-based analysis method. In practice, enterprises can preliminarily estimate the probability of current recruitment methods in some aspects, such as interview pass rate, according to historical data or subjective judgment of recruitment. Because these probabilities are the summary of previous data or experience, there will be great deviation in practical application. In probability theory, we call these probabilities transcendental probabilities.

Bayesian analysis is based on these prior probabilities. It corrects the prior probability through investigation and statistical analysis, obtains more accurate posterior probability, and coordinates managers to make decisions accordingly.

Bayesian analysis of enterprise recruitment methods usually requires the following steps:

1. Collect internal and external historical data and position information of enterprise recruitment.

2. By collecting post information for calculation and logical judgment, check prior probability, including historical probability and logical probability, to determine whether it is suitable for calculating posterior probability.

3. Analyze the validity according to Bayesian theorem.

For example, Company A decides to conduct Bayesian analysis on the recruitment method adopted for a management position, assuming that the company conducts interviews regularly to select the management position of the company. Through the enterprise's statistics and logical judgment of the historical recruitment data of this position, it shows that only 70% of all the candidates for this position actually meet the requirements of the enterprise, and the rest do not meet the requirements of the enterprise. Only 80% who meet the requirements of the enterprise can pass the screening of the interview, and 30% who do not meet the requirements of the enterprise can pass the interview.

In the recruitment work, enterprises hope to recruit people who meet the requirements of the enterprise and pass the interview, but in reality, those who meet the requirements of the enterprise often fail in the interview, or those who do not meet the requirements of the enterprise pass the interview. Of course, no matter which recruitment method, there will be this problem more or less. As a recruiter, what I want to know is the probability that a person who "passes the interview" "meets the requirements of the enterprise". If this probability is low, it proves that this recruitment is invalid.

Take Company A above as an example, and make Bayesian analysis on these data. Suppose an applicant has passed the interview of the enterprise, what is the probability that he is a "qualified" person?

According to Bayes theorem of probability theory, we use A 1 to represent a "qualified" candidate, and b stands for passing the interview. Assuming that an applicant has passed the interview, the probability that he is actually a person who meets the requirements of the enterprise is:

P(a 1 | B)= P(a 1)* P(B | a 1)/[P(a 1)* P(B | a 1)+P(A2)P(B | A2)]

=0.70*0.8/(0.70*0.8+0.30*0.30)=0.862

Excel formula code for Bayesian analysis of this position (see attachment)

From this, it can be judged that the recruitment method of this position is valuable for screening candidates. Because for this position, if you don't interview, you will randomly select one person from the candidates, and the probability that he meets the requirements is 70%; However, if the company only accepts applicants who have passed the interview, the probability will increase to 86.2%.

Through Bayesian analysis, we can clearly understand the effectiveness of a certain recruitment method in the selection, and then decide whether this recruitment method should be improved. The probability in the above example is 86.2%. If the enterprise's goal is more than 90%, then this interview method needs to be improved in technology and process to meet the requirements of the enterprise.

Practice shows that Bayesian analysis method can achieve good results in evaluating the effectiveness of enterprise recruitment. Of course, it also has some shortcomings. For example, enterprises need to keep a large number of historical recruitment data, and need to make a rational analysis of the external appearance of this information. However, in fact, many enterprises have not established a perfect database, so some data must use subjective probability, which makes the accuracy of evaluation results doubted by some people and hinders the application and popularization of Bayesian analysis method. In order to solve these problems, many theories and research methods of Bayesian analysis have been updated, such as Bayesian regression analysis, interpolation method, piecewise pricing model, sequence analysis and so on. Those who are interested in this can also learn about it.