Mistake in my experiment



I’ll be defending my thesis within the next three weeks.

I discovered a mistake in the way I collected part of the data during my experiment. It is a within subject experiment, where I asked users to use one interface under the first condition and to use interface B under the second condition. Instead of giving users a usability survey at the end of using each interface I gave them all the two surveys at end ( after they used interface a and interface B). I am worried that may have affect the results that I got and I am also worried that I might get a major correction or an MPhl.
I have not mentioned in my thesis when I have the survey to the user l, but the examiner may ask . And even if they did not ask I feel that I have to mention that and added it to my thesis.
Any advice? Has anyone came across similar mistak?


Hi there

This is a design flaw / something to consider that could affect your results. But it is not something that by itself is likely to affect the outcome of a viva.

By the way, did you counterbalance the order in which participants were exposed to the two conditions? That is, did half of them see Interface B first, and the other half see Interface A first? If so, that would remove any effect of having the survey directly after the second condition. But if you didn't counterbalance, again, it is just a limitation that probably should be commented on in the Discussion.

Personally, I would be ready with a clear answer that shows you have thought about it (eg. I didn't counterbalance the conditions but should have, as blah... ). But I wouldn't raise it unless they ask you directly about it.

I hope this helps. Have a good viva :-)

Avatar for rewt

I wouldn't worry as a small flaw will not invalidate the whole thesis. The data may be slightly skewed but still usable as it will not significantly alter the conclusions. If the examiners mention it you can just acknowledge it, say you would have done it differently with hindsight and then explain how small of an affect it will have. No thesis project is perfect and understanding the limitations of your work is a key part of a viva.



The most important thing for you to do at viva is acknowledge the error and demonstrate your understanding of it's implications.

I say this as an examiner (with the obvious proviso all examiners are not the same, but having worked with internals/externals I think I have a reasonable feel for the general consensus):

A 'perfect' thesis is not required to get a PhD. There needs to be a contribution to science, but it need not be groundbreaking.

More importantly, as an examiner, you're often asking the question 'is this person sufficiently skilled to be an independent researcher?'. Having made mistakes, realising them, and their implications, is not a bad thing.

The first big mistake you could make would be to attempt to defend flawed work at viva (or, to a lesser extent, be unaware of it). When this happens it's a big cause of difficult discussions when the candidate leaves the room, because someone who's either willing to 'fudge' data/results, or is abundantly naive, is not sufficiently skilled to be an independent researcher. The second big mistake is to defend conclusions data doesn't support - the classic being attempting to generalise to a population from a tiny sample. Often candidates worry too much on the 'need for a contribution' side and end up trying to defend a contribution their data doesn't support. Corrections issued are often about toning down the conclusions in this regard - because it's dangerous/damaging in academia to claim things that aren't proven.

Don't be 'might have' affected the results, be 'x study showed.... influence of this form of surveying'. The most important thing at viva in terms of outcome (to the limited degree you can influence it at that stage) is to be knowledgeable, confident, honest, and speaking like an academic, rather than someone desperately trying to fake it through a test.