Hi,

I'm in the process of designing a quantitative study and I'm wondering once the results come in, is it OK to get a statistician to perform the actual tests? I'm new to quantitative research and SPSS, and wondering whether this might be an option? Has anyone done that? Is it OK for the thesis to do that?

I'm not sure I would feel 100% confident in reporting my results if the calculations haven't been at the very least checked by a quantitative researcher/statistician. My supervisor is not a quantitative researcher.

Thanks!

There may be different regulations at different universities and it may also depend on the focus of your thesis/your discipline. However, if you have any intention of working with quantitative data in the future, I would have the mindset of seeking help with the analysis, rather than handing it over to someone else entirely, as it will be harder for you to learn otherwise.

Please don't wait until the end to consult a statistician. Do it now at the design stage in case there are things you might overlook (e.g. minimum sample size, best way to code your data). They will be able to help you better with the analysis (whether that's checking it or assisting in doing it).

To reiterate what HazyJane is saying, if you're going to do a quantitative study then you need to know how you're going to analyse the data prior to collecting it - you can quite easily end up in a situation where you've collected loads of data but there is no appropriate statistical test/method for analysis.

You need to refine your research question in terms of the types of statistics you are going to be doing - are you looking at inferential statistical tests looking at comparing the differences between groups in response to some independent variable? Or are you looking at the relationship between variables where correlation or regression analysis might be more suitable?

You definitely, 100%, absolutely, never want to be in a situation where you are collecting data without knowing what you're doing with the analysis. Otherwise you can easily end up either not being able to do the analysis at all, or in some sort of fishing exercise.

I would think this is not acceptable. At PhD level you can't farm bits out to consultants - it's a basis requirement that the thesis should be all your own work, so you need to acquire the necessary skills without cutting corners. By all means seek advice from more experienced colleagues but you personally need to be in a position to defend your calculations and why they were used during the viva.

Well I think your thesis should be your own work, but sometimes others contribute to it, and as long as it's not the main component of your thesis, I think it's ok. I've seen theses where people have said 'someone provided x and I did y', so I think it's ok as long as you state your contribution and it's substantial.

I would very much doubt you'd get a statistician to agree to this! As part of the PhD you need to learn how to conduct research, so acquiring these skills is vital. There should be courses for training in SPSS. My supervisor didn't know about the particular techniques I was employing - it will often be the case you need to ask for help from other people.

That's not entirely true. You can certainly get other people to do things on your behalf, but it needs to be directed by you. For example, I was employed as a Research Assistant to support one of our PhD students in collecting their data - the student designed the study, organised the interviews, created an interview plan, etc. but it was my job to perform the interviews. Also, I've been paid to transcribe interviews for other PhD students in the past. It's "farming out the work" but not the decisions or the analysis and interpretation.

Actually, this can sometimes be seen as a positive for the PhD student - they're being resourceful and maximising their personal output by getting someone else to perform the data collection. It also shows management skills etc.

I also know of someone else who's PhD was focused around designing a computer game for a particular purpose. She designed the game, but had no programming skills herself, so she hired a games programmer to implement her design. She could have spent 2 years programming the game, but it wouldn't have been as successful or quick as getting someone skilled to do it. Her contribution was in designing and evaluating the game, not implementing it.

That said, when it comes to stats there's no cutting corners - it's integral to the research output, so you have to understand it yourself. And once you understand it, performing the calculations in SPSS is pretty simple, so there's no excuse for the OP.

It depend how complex though, surely? Maths is not my forte, but I've used SPSS for basic ANOVA and posthoc tests and found it really easy to use, but if I had to do something more complicated than that, I would definitely want someone to be looking at it to check. I wouldn't feel comfortable doing an Ecological PhD, for example, it just seems way too complicated for my limited knowledge, but basic stats in SPSS is fine.

The group I am working in outsources things like sequencing/mass spec and bioinformatics. We, the phD students, do the data interpretation (and sample collection beforehand) and subsequent experiments.. and we have no problem arguing that our work is not our own. After all we planned and did the experiments that created the samples and we interpret the results.

Most of us have basic knowledge about bioinformatics, but we can't compete knowledge wise with somebody really skilled at it. So yeah, I agree, basic statistic knowledge would be preferable for the OP, but I would nonetheless consult a statistican. SPSS is more or less easy to handle, but study design is very tricky and as long as you have no ideas about group sizing, matching of subjects, data collection or how to handle confounders I would highly advise seeking help from somebody skilled at it.

If you are unlucky, you have some flaws in your design and will get into trouble in a later stage of the project.

Thank you everyone! Some very good advice. I think I need a statistician to review my work and the calculations I propose to use before I conduct the study.