I realize there is a somewhat similar question posted, but my question is different in that I found a pretty big mistake in my thesis. It's such a big problem, it changes my results. I almost wish I hadn't noticed, and been so careful to go back and check every thing, because I really don't know what to do about it except fix the errors/interpretation errors and discussion (today). I defend very soon (in days). It's possible that my committee won't notice it, but I feel like the guilt might drive me insane. I'm worried that my chair noticed it in the past, but they didn't correct me because they feel I'm incompetent (which I feel in ways I am when it comes to stats). Should I re-write it and let my chair know and ask what to do? I need to defend soon in any case. I wonder if I could just present as though I'd already corrected it and e-mail the revised manuscript beforehand. I'm just not sure how they would perceive someone who overlooked such a obvious mistake. Any advice would be appreciated.
The thesis and thesis defense is less about having the results you wanted to have, and more about demonstrating that you know how to do good quality research and can work on that somewhat independently. It's about figuring out what questions to ask and what methods can be used to find the answers, and then applying those methods to come up with answers. What the answers actually are is not as important [for the purposes of passing a thesis] as the process you took to get those answers. Your discovering this issue and taking prompt action to fix it shows attention to important details and integrity in the knowledge-discovery process. Sure, it would've been better if you'd caught that earlier, but you've caught it now, before your defense, and you're rewriting the discussion and conclusions to reflect your best analysis of the data.
In my opinion, your having found this and your efforts now to promptly fix it say more [positively] about what a thesis is supposed to evaluate than most completed theses.
Don't panic. Talk to your supervisor and committee. Tell them what you found. Revise your document to reflect the new understanding. Maybe you'll have to delay the defense a bit, but more likely you'll present at the same time and talk about what you found; the committee might require you to deliver a revised document [some weeks after the presentation] reflecting that before they sign off. That might take you some time to do but it should be OK, and will leave you with work you can feel is more solid.
Kudos to you for finding the issue and having the integrity to stand up for it. This should help you in the long run and the core evaluation at issue here, at the cost of some extra work to revise and maybe some scrambling to re-practice your revised presentation.
Edit: Congratulations on passing!
Get off of Stack Exchange and contact your advisor, right now. He/she is the most qualified person to help you understand what is going on and what to do about it.
It's possible you might have to delay your defense in order to fix it. That would be unfortunate, but not the end of the world.
On the other hand, if you know about a serious error and defend anyway in the hopes that your committee doesn't notice, that is deeply unethical. We are talking "kick you out of grad school", "revoke your degree years later" unethical. That is not an option. Forget about it.
It seems to me unlikely that your committee knows about the error but is intentionally ignoring it in order to "trap" you. That would be very inappropriate behavior on their part, and I've never heard of it happening. There would be nothing for them to gain by doing so. And even if this were the case, pretending you don't know about the error would only make things worse.
This happened to me. I found an error in a complicated mathematical proof in the appendix to one of my papers. As I flew home to defend it I was fixing the mistake on the plane. In the defence, I told them about the error. They told me to fix it and awarded me the degree. This was a formal political theory paper; I doubt if they even read the appendix, but I'm glad I told them.
Unless this mistake blows away your results, it's best to be honest. (Actually, it's best to be honest in any case. There are lots of bullshitter academics out there; don't be one of them.)
I'll add a suggestion to the excellent answers provided already:
When we find an error in our proof or even our claims - especially in something as significant in our lives as a these (even if it's an M.Sc. thesis) - we tend to believe that everything is ruined and the research is useless. It's really not. Even if you can't fix it in a day. I would go into details regarding why that is, but that doesn't matter now.
The important thing to remember is: Don't make the reporting of the mistake the focus of your thesis presentation. You should definitely be fair and open: When you get to the part which directly relies on the error, tell the committee that as you were preparing for your defense, you found a mistake in that claim / in the proof of that claim. Don't start going on and on about the mistake and how you made it and how it invalidates everything; make your presentation like you would if you hadn't found your mistake, and when you get to where the mistake actually happens, that's when you say what the mistake was instead of presenting the erroneous argument. Let the committee decide if they want to focus on it or whether they would rather hear the rest.
Andrew Wiles' famous proof on Elliptical Curves also proves Fermat's last theorem 356 years after Fermat proposed it. The 1993 proof contained an error that took Wiles over a year to fix with the help of his assistant Richard Taylor. Even the greats make mistakes or overlook things. This is why papers are peer reviewed in the first place. Even with the mistake, the original proof was valuable as it showed innovative approaches to the problem.
If you could fix the thesis in under a day, it wasn't a major error. In the words of Benjamin Franklin "When in doubt, tell the truth. It will amaze your friends and confound your enemies." Just be glad for word processors. Back when I was in college, I had to type out my papers by hand on a typewriter, and created the graphs using rulers, triangles, and Rapidograph pens.
First, Speak to your advisor immediately. That person is in the best position to guide you. Second, realize that the most valuable stock-in-trade in academic life along with competency is integrity. Research mistakes WILL happen, that's a fact of life and the imperfect world we live in. [Albert Einstein did not get General Relativity right the first time he published it and save for WWI, experimental data would have disproved the incorrect version of GR.]
Needing or waiting for someone else to point your error out (especially since you discovered it already) speaks poorly of both one's competence AND integrity and surely you don't want that outcome.
If it's too late to amend the thesis prior to its defense, it's better that they learn from you of the error than having someone else point it out. That would demonstrate both your competence AND integrity.
A good academic committee will recognise the positive significance of a candidate who proactively checks their work and takes errors seriously. They would see it as a plus for you, not a minus, whatever the effect on your thesis, because the thesis itself is a tool to assess you as a researcher in the field, and this action will indeed reflect well on you. (And you won't have to worry about someone else noticing it in future!)
Write them a formal note explaining what you found, and your initial assessment of its impact - extra time, any changes to the paper, etc - and tell them what you'd like (an extra week or month to rewrite that section, or update it to fix the issue, or to consider if any other part is affected and needs changing as a result).
Be matter-of-fact and cordial - and have a verbal chat with your supervisor in which you show him/her the note you have drafted and check it looks OK to them. That also gives your supervisor a heads up so they don't look foolish or caught by surprise, and a chance to give any other advice.
In independent study as an undergrad I reviewed draft MS and PHD papers for Math errors. Every paper had some error or errors, some were minor, and some were major. It was never a big deal unless the grad student got defensive. Accepting the criticism and moving on was part of the process. The students who positively engaged with my adviser to understand the flaws and gain a deeper understanding were really appreciated by their adviser and the committee.
Are you (1) a life scientist and (2) worried about the stats?
I worked as a statistical advisor for several high-profile journals in the life sciences. Your hunch that your stats is wrong is probably right. Your hunch that the examiners may not have noticed is probably also... well, I should not say too much.
Be frank about your misgivings. If you are able to make coherent arguments during your defence, that will in itself be a positive thing.
And learn stats properly, for crying out loud.