Statistician - What to look for?

[jasonb067]

Our organization has recently decided to add a statistician. As a part of the group that will be reviewing resumes, interviewing etc. I am concerned we will not know good from bad.

All of us in the group know what we want this position to do for our organization (as much as anyone does before adding the spot and letting in mature in the group). We will be able to go down the list of requirements and check yes or no to them. I am just concerned that we will not have a enough "nuts and bolts" knowledge of the subject matter to separate bad from good, good from best...


Does anyone have any suggestions on what to look for in this particular position?

Should we consider having someone (third party) with more subject matter knowledge sit in on the decision making process?

Are there resources which I and other team members can review and become familure with that could be helpful?

Any other thoughts?

Just some background. Again, we have determined what we want out of this position. This is a position reporting the lead person responsible for the Operations of the company. The background of the interviewing group are three Engineering Managers (including Quality, all of which are Mechanical Engineers), Operations General Manager (also M.E.), Human Resources and Production Manager. Of that group the one with the most stasticial experience is the Quality Manager (SSBB, I know no one likes titles but just trying to give a scope of subject matter knowledge).

Thanks for you help.

[Sturmkind]

I think it is based on your objectives and culture. One tip is to write out pertinent questions with the ideal desired response to screen out candidates that are obviously not a choice fit.

I was charged with putting together a Supplier Development group a few years ago and after interviewing ~120 folks who were 'experienced' at Quality to find only 2. An explicit questionairre helped provide the remaining 4 that I hired within ~15 interviews. After all, if they are whiz-bang at the numbers but have trouble communicating or grasping how best to help your overall goals then a series of good questions can really help.

[20110131]

since this is a nunbers game, and a social one also. An online testing process is a good tool to weed through people. It can take the personal issues/viewing-bias out of an evaluation.

[bobdoering]

One thing I would look for is someone willing to learn - and not one who believes they know all that exists. Also, someone that can express their statistical results in manner that people can understand - not one who chooses to keep things complicated as a means to sustain a superiority complex.

I would also add, somebody that shows a good understanding of the data collections as well as its analysis. Most of the defective statistical conclusions I have found were traced back to sampling or measurement error of some sort. The plug-and-chug afterward was pristine...but the results were of dubious value.

[Sturmkind]

I 100% agree with Bob! I was blessed with working with some very practical hands-on statisticians and there is no substitute for those who first question the basic assumptions. One noted that there are three types of error: I. False out-of-control signal, II. False in-control signal, III Measuring the wrong signal.

[adickerson]

I would look at your current data sources and see if the applicant has any experience working with this data. For example do they know how to interface the data stored in your accounting package? How about you ERP system (if you have one)? Does you manufacturing equipment have any data outputs? How about inspection equipment data packages?

A statistician will be trained in using data but not a lot are trained in how to get data. It is one thing to make since of a bunch of data in Excel, it is quite another to export from a CMM machine into Excel.

More recent graduates will also have some background in programing languages. Do the languages they are familiar with have any relevance to your business? If you have a web portal programed in Pear it would be good for your stats guy to be able to interact with this portal at the code level. If you use Access a lot it would be good for your statistician to know how to use it.

[Statistical Steven]

Do not simply hire a SSBB or someone who has done data analysis but has no formal training in statistics. What to look for

1. Interpersonal skills. Cannot underestimate how important communication is to the relationship between the statistician and the organization.

2. Experience as a problem solver. Helps if they have industry experience, but must have experience in using data to solve problems.

3. Take some basic statistical concepts and ask the candidate to define it. For example, what does a p-value mean. If you can understand the answer, then they have good oral skills.

4. Look how they dress. I know this sounds silly, but if they cannot dress professionally, they will not be taken serious in the organization. I called this the "geek" factor. Statistician who are too geeky get labeled as too theoretical.

Hope that helps.

[bobdoering]

In Reply to Parent Post by Statistical Steven

4. Look how they dress. I know this sounds silly, but if they cannot dress professionally, they will not be taken serious in the organization. I called this the "geek" factor. Statistician who are too geeky get labeled as too theoretical.

This is always a tough one. Sure, too geeky can be an issue, sometimes people will infer that if you are dressed well for an interview (even though this used to be the minimum standard) - they interpret that you won't "get your hands dirty". Be very, very careful about inferences from interview dress. They are more wrong than right these days.