Hey,

I'm panicking quite a bit right now.

I'm close to finishing my thesis. I am editing one of the chapters, which is based on work I did about 3-4 years ago in my first year when I understood things a little less than I do now.

I was doing some image classification and I generated some features. I have realised that two features (x1 and x2) I used are in fact equivalent. For example, let's say you found two ways to measure the density of a material. I used both values, which were meant to express different properties but in fact, were both density. I think I got confused over the mathematical formulation of what I was working in and probably wasn't as rigorous as I should have been.

My thesis chapters are self-contained and anything in this chapter does not affect the remaining 4 methods chapters.

This error doesn't change the results since theoretically, it doesn't add new (and wrong) information to the model. The overall conclusion/discussion of that chapter remains unchanged (in the most part).

However, I am really really scared that this will definitely turn up in my viva. I am going to speak to my supervisor next week when he's back from holiday but I was wondering if anybody experienced the same or something similar?

Do I add an extra paragraph to my discussion to state that in fact, x1 and x2 are equivalent? I can't just pretend I don't know then if brought up by an examiner, be like "yes, upon reflection, these are the same, but it doesn't change x and y, it just means two features are colinear". Or should I just ignore it, and just declare it when prompted?

In the end, I don't think it is a major error as it doesn't invalidate anything, it just weakens the experiment, which I find is weak enough as it is, compared to all my other work. But I am still worried.

Thanks!

Hey first of all don't panic. Or panic, but know that panicking is as effective as trying to solve and algebra by chewing a bubble gum.

Second of all, would it be that bad if you just explained this? Like, added a paragraph at the end of the chapter as well as the discussion stating they are equivalent? I mean, since it doesn't change or affect the results than why wouldn't you make it clear? Either way you run the risk of the committee figuring it out themselves and I think that'd be much worse because it would portray you as not careful and they would automatically assume that you didn't notice this. But if you address it yourself beforehand, it's a different story.

Good luck and let us know how it goes

Thank you for the reply.

I agree with you and my supervisor mentioned the same. I cracked and e-mailed him this morning, quite early. I told him what had happened, that I found the chapter weak and asked for his suggestion.

He told me that I should basically mention it in the discussion and make clear how I have built on this earlier work to perform much more thorough work, that self-criticism in the discussion of this chapter pointing to new work is fine. However, he mentioned that I should no overdo the criticism!

I think that is the best course of action. There isn't anything I can do now, and I have no time to re-run experiments. Theoretically, nothing changes except the strength of the experiment.

I still feel a bit stressed about it but feel a lot better now that I've told my supervisor. I presume examiners look for self-critical reflection of work in a thesis? The quality of my work in my 1st year is drastically different than the work I am now performing.

Well, that's quite alright. I don't think anyone has the same quality of work from the beginning. I am sure lots of postgrads look back at their first year and wonder "why on earth were they thinking". So take it smoother. It also seems that your supervisor is pretty cool with that and that should help you. Also,I am sure self-criticism can work in a good way too. So good luck. It will be great.

Sounds like a good plan to me. You're in a different field (I'm social science) to me so it might not be quite the same, but my supervisors emphasise how important it is to critically discuss your own research methods. It shows a much more sophisticated researcher than the one who says 'I planned it perfectly, I did it perfectly, nothing could have been done any better' because there are always mistakes and decisions to be made along the way.