M=N/logN? or M=ln(N)

[guest]

To Kent以东首帅哥 :
"假如说你是一个employer,现在有N个人来应征一个职位,N的数值比较大,使的面试所有人的资源消耗远远超过了你们的支配能力,那么是否有最佳解M,使得面试N中M个人就能在资源消耗和最终人选中得到一个很好的折衷?

咱好象从脑海深处拖出一个经验公式M=N/(logN), Log是以e为底的自然对数,有没有依据?"

My common sense tells me your formula is not correct. I know usually for an permanent position, an employer selects 4 or 5 candidates in a short list to give interviews from a large number of applications (50-250 in 1990's recession). If the employer is lucky, he/she will choose one, but in many cases, he needs to repost one, two, even three runs to get one he like. Say 100 个人来应征一个职位, N=100, your formula gives 100/ln(100)=100/4.60517=21.714. No employer will interview 21 persons for one position, since he has to pay the applicant for travel and accommodation expenses.

Actually, if you use M=ln(N) as a 经验公式, it would be close, e.g. when N=250, ln250=5.52, N=100, ln100=4.6. I said 4 - 5 is a reasonable number if applications are great than 30. Hope this helpful to your question.

 

[gdntfrank]

Till now I understand why most of the job applicants don't receive any
response!

 

[Kent以东首帅哥]

呵,承劳贴到这里.不过这样是先假定了employer的最大负载能力,所以也就没什么最优值可求了.

但看来也是employer的实际做法,只要对比一下发出简历数和得到的面试数,外推一下.

 

[guest]

"不过这样是先假定了employer的最大负载能力,所以也就没什么最优值可求了. "

I don't know what an objetive function is for your 最优值 model, but as long as you set up a restriction subject to: M<=5 for one position (this is true - 但看来也是employer的实际做法), your 最优值 will fall into this range.

贴到这里 since I didn't register.