Navigating Office Politics: Worker X's Fishy Model Revision Procedures

[Ma Xie Er]

I'm fresh out of school, working for a department where there's some office politics. A co-worker, call this person worker X, is friends with the boss. Worker X is using logistic regression to build a model, and whenever I question anything, the boss tells me worker X is too busy to answer questions.

I find out later, after reading a paper from worker X, that some fishy things are going on. Specifically I'm reading worker X's model revision procedures and...

Step 10) Keep variables which were statistically significant in samples A and B.

...

Step 19) Conduct cross validation study by applying model coefficients derived from sample A to sample B.

I am scratching my head here and thinking "wtf is worker X doing?".

If I call worker X out, I risk a lot. My boss is not mathematically inclined, and makes decisions based on concensus.

What to do?

 

[russ_watters]

1. The best way to deal with office politics is quite simply not to participate. It is easier than you think.
2. You need to remember where you are in the chain of command and that a chain of command is also a chain of responsibility.
3. The way you describe your tone/approach is adversarial. My approach to my bosses, when I disagree, is adversarial; a right I have earned through years of track record for being right. You haven't earned that. So:
4. "I'm too busy" may mean "ugh, not you again: let it go/leave me alone."

So, the bottom line is that I think you need to take more care/caution with your approach, remember that you primarily need to learn and not challenge and remember that if a mistake is being made, you didn't make it and won't own the consequences. Trust me: there are few worse employees than a kid fresh out of college who by definition knows nothing about the job, but thinks he knows everything -- and it doesn't matter if you are right about this particular issue or not. If you need more specifics:

Wrong attitude: "This doesn't look right; why did you do it?"
Right attitude: "I don't understand what you are doing here; can you explain how it works?"

 

[Ma Xie Er]

That's great advice, but if you were in my position, what would you do?

 

[russ_watters]

That's great advice, but if you were in my position, what would you do?

Um...I would apply the advice given...? It is difficult to give an exact course of action without knowing all of the exact details of what scenario you are looking for an exact course of action for.

 

[StatGuy2000]

To the OP:

If you have questions about the way that your co-worker's model revision procedures, the best way is not to question him/her on the spot, but see if you can schedule an appointment (through Microsoft Outlook or some other calendar system you have in your workplace) at a time that is convenient for both you and your co-worker. Then sit down and ask questions, in a respectful tone, about some of your questions, in the way that @russ_watters has suggested ("I don't understand what you are doing here; can you explain how it works?")

In this way, you are doing the following:

(1) Your are showing initiative in your willingness to learn about the job.
(2) You are being respectful of your co-workers.
(3) You just might be able to make a constructive suggestion, in a way that doesn't appear too aggressive or threatening.

 

[Stephen Tashi]

I'm fresh out of school, working for a department where there's some office politics.

Is the department doing anything important? - designing a nuclear reactor? -building a suspension bridge?

A co-worker, call this person worker X, is friends with the boss. Worker X is using logistic regression to build a model

Will Worker X's model have a chance to succeed or fail? Or will it just be part of a report that nobody will read?

 

[russ_watters]

...
(2) You are being respectful of your co-workers.
...

I agree with all, but want to re-emphasize: yes, they are co-workers, but the OP must not forget they are his superiors.

 

[Fervent Freyja]

What to do?

This depends on whether or not you are directly involved with this project. If not, I think you should let this one go and just document all future inconsistencies or "fishy" things. Accusing someone of falsifying data is very serious. However, with what little information you've given about the circumstance, I call that it's most likely a misunderstanding on your part. It is likely that they are running a test/error analysis on the model that will be published, in order to predict the accuracy of the model: A and B being a test set and not what will ultimately be published... If you are working on it as well then I think you should attempt to understand why they are using this particular revision method. If you are involved then this person should take the time to explain it to you or simply ask for references to the method that you can study on your own- without approaching them in an accusing manner.

The thing is, you have a long road ahead of you, things like this will surface occasionally, and so, if you run to the boss on a whim each time (and turn out wrong or with no evidence to support your claims), then it hurts everyone long term. I think that it's best to at least make an attempt to handle things of this nature informally, unless you have evidence of misconduct, and then go the formal route if you cannot resolve it on your own. I'm sure that your coworker would very much appreciate you approaching them on the matter first. That's part of teamwork. If you accuse them of misconduct, then find out later that what your coworker did was simply checking their work, how could that pan out?

 

[Choppy]

My friend Princess Elsa has some good advice for the OP...

On a slightly more serious note, I understand that you've found something that you think is incorrect with your co-worker's methods. What you haven't told us is the degree to which you have responsibility for this work, or what the big picture consequences of this potential error.

To be clear, I'm not suggesting that you just keep your head down and only worry about your own stuff. It's good to ask questions in general. But there's a big difference between asking for details about a procedure that you're ultimately going to be held accountable for, and asking about the details of a project that is outside of your responsibilities. Sometimes it can be difficult to understand the context for why a procedure is the way it is. On the surface, it may seem there is something wrong with it, but sometimes you need to understand the context of the work. Perhaps your co-worker is using an approximation that's justified under the circumstances? I can understand that perhaps that's why you're asking - but if it's not something you're directly responsible for, they may feel irritated when asked to justify it.

 

[Drakkith]

I'm fresh out of school, working for a department where there's some office politics. A co-worker, call this person worker X, is friends with the boss. Worker X is using logistic regression to build a model, and whenever I question anything, the boss tells me worker X is too busy to answer questions.

I find out later, after reading a paper from worker X, that some fishy things are going on. Specifically I'm reading worker X's model revision procedures and...

Step 10) Keep variables which were statistically significant in samples A and B.

...

Step 19) Conduct cross validation study by applying model coefficients derived from sample A to sample B.

I am scratching my head here and thinking "wtf is worker X doing?".

If I call worker X out, I risk a lot. My boss is not mathematically inclined, and makes decisions based on concensus.

What to do?

Have you approached worker X and asked them about what they are doing? Perhaps they are unaware they are making a mistake?

 

[jedishrfu]

You might learn some leadership politics from Sun Tze:

https://www.forbes.com/sites/ericja...est-pieces-of-leadership-advice/#728ec3ec5e5e

Choose your battles wisely, you have the inferior position. You might win this battle but coworker X can torpedo future opportunities unless he loses the managers favor which isn't likely. Sometimes folks will try to get the bad coworker promoted to get him out of the team, sounds unbelieveable but it happens.

The best course of action for you is to learn from coworker X when you can by asking questions and lead him to the right solution or perhaps it will give you some insight as to why he chose that route.

In any event, his work isn't your work but if you interfere then the tables might get flipped and you'll be blamed for the error in his work. Office politics can be very tricky and nasty.

Keep your resume up to date and always look for better opportunities and remember if you leave your present job because of a political issue like this it's unlikely you'll get a good reference from them.

Russ's advice is very good and very tactful allowing to become a part of the team.

Dr Courtney's advice, while noble is far more riskier as it may backfire. It's rare in industry for someone to respect you for your boldness after you've called them out privately or personally especially in this case where the coworker is close with your manager.

Higher level managers don't like disturbances and will punish both parties so you lose. It will most likely ruin your chances at promotion for a few years unless you leave which probably something you should prepare for anyway.

 

[jedishrfu]

It seems that the OP has been given a lot of good advice here and that this thread has run its course.

So now without further ado, I will close it thanking everyone who participated.

Jedi