Science fuckups

Now that you figured out the signal is 100% dependent on where the wells are located on the plate, isn't it possible to subtract this out as noise, and retest the procedure for a more accurate representation? If the procedure works in theory then the only thing that happened is the noise is dominating over your signal. If you get rid of the noise then the actual signal should appear...If not then there must be something wrong with the theory (which is actually a good thing to discover). If the procedure you use has no foundation in theory, then why the fuck are you using it?

>Now that you figured out the signal is 100% dependent on where the wells are located on the plate, isn't it possible to subtract this out as noise, and retest the procedure for a more accurate representation?

This is pretty much what I have been doing for the past month

>If the procedure works in theory then the only thing that happened is the noise is dominating over your signal. If you get rid of the noise then the actual signal should appear...If not then there must be something wrong with the theory (which is actually a good thing to discover).

>If the procedure you use has no foundation in theory, then why the fuck are you using it?

It is based on the work of a single fucking college intern that got """good""" results by using an extremely small sample size and manipulating the data to look good.

My boss then latched on to his results like they were the greatest thing on earth, and is extremely persistent on getting me to repeat the results.

I designed my own experiments to reach the same ends, but they never worked. It wasn't until I repeated it the same exact shitty way the other guy did it that I got "acceptable" results.

I am so fucking mad jesus christ

What does it detect?

He probably signed a contract for nondisclosure.

So if he hells you he's REALLY fired.

Hell even what principle he's attempting to test could be valuable information to test, just the abstract is enough to fuck shit up.

Consider if I'm a company, and now I know that xyz isn't really feasable, then I know what not to do for my own teams.

This, I can't really describe more than what I've said already

>It is based on the work of a single fucking college intern that got """good""" results by using an extremely small sample size and manipulating the data to look good.
Damn that sucks, but is there no theory to check it against other than this one person's experiment? For example, try asking yourself WHY the signal depends on the location of the well on the plate. It could lead to some insight about the technology or the theory behind it. I realize that you are working on this project right now and it would be compromising to give out too much info, but giving a few more details about the experiment like could help us to help you.

I think I am going to try it with a whole seperate plate, and scan the wells that are in the same corresponding positions to normalize them.

I was doing normalizations before, but it didnt take the positions into account.

Thanks user, I will report back

>I was doing normalizations before, but it didnt take the positions into account.
Try this, I really think it could make a difference. Good luck, don't get fired.

True capitalism has never been tried

OP here, just met with my boss.

I prepped a pretty comprehensive powerpoint to explain everything and stayed calm. I explained what I planned to do to try and fix things, including what suggested.

He was extremely understanding and rational about it. "Better you found the problem now then a year further into the work." He suggested a few more solutions, and everything went very smoothly.

Huge weight off my shoulders. Whew.

Thanks Veeky Forums!