Dissertation students

Lab manual for BSc or MSc dissertation students 

Lab culture

As you are probably undertaking a piece of research that is reasonably unique, your dissertation could be a rather lonely experience. Some of us thrive in this environment, but most people will find this pretty jarring compared to the very group-focused teaching you've experienced up until this point. But remember, you are part of a group of dissertation student who have their supervisor in common. Even though you probably weren't friends before you found yourself part of this 'group', you'll really benefit from getting to know one another. Talk about each other's projects, discuss things, share interesting papers you've read with each one (and me!). There's also research assistants and PhD students in the lab, who you should try to get to know - they can be a great resource for bouncing ideas off of, and finding out how to best manage your supervisor. One concrete thing I will do to try to foster a lab culture is to have a whole-group lab meeting once per term, which all undergraduate dissertation students must attend. In semester 1, the meeting will take place around week 6, and you will all present your dissertation project ideas to the group. Then, toward the end of semester 2, you will all present the outcomes of your project to one another (and me). 

Finally, try to enjoy your dissertation - it's got the potential to be the best part of the course, so really try to engage with the experience. Be enthusiastic about your data collection. Come to me (or each other) with problems - don't sit, stew, and worry about the marks you'd be losing if you meet with me (it really doesn't work like that). Get the ethics application done early, start reading as much as you have time for, and start writing your methods as soon as you get going with data collection.

Pre-registration and data simulation

I like students to pre-register their experimental protocol, hypotheses, and analysis plan using a short template designed to help with the ethics procedure. A key part of this process is simulating some data (details are on the form). After we have our first meeting, I will probably ask you to complete this form to make sure that we have a clear collective understanding of what the project will entail.

Ethics

First order of business is to complete your ethics form. All the information you need to complete this job is found here. Have a crack at this form by yourself, making sure you complete all the boxes - the main one to fill out in enough detail are the boxes in Section 7 - this is where you outline in detail what exactly is going to happen in the study. If you don't know what stats you need to perform on your data then you need to really go back to the drawing board and figure out what question you are trying to answer. Remember, there's no need to justify your sample size with a power calculation, but some form of sample size justification from prior literature is appropriate. Don't forget to make consent forms and information sheets based on the template found at the link above. Once you have it all completed to the best of your ability, compile it all together as a single word document and send it to me.

Your abstract

It's key that you fully understand the goals of your research project, in terms of articulating your IV(s) and DV(s). Because of this, one of the earliest exercises (after you submit ethics, but before you start data collection) will be to write your abstract! This might seem like a weird task, but if you write you abstract and include the results you are expecting to find, this will (1) force you to understand the outcomes of the data you are collecting, (2) make sure we are both on the same page and (3) make the writing of your abstract at the thesis submission time a whole lot easier.

Data collection

All the information you need for lab booking and data collection can be found in the lab manual - please read it carefully. Feel free to coordinate with other lab users informally, especially if you are planning to run a contiguous chunk of testing (not ideal, but necessary occasionally).

Statistics

If you're doing an object lifting study, then there are some specific guidelines here. Regardless of your project, there are some general points worth bearing in mind. 

First, try to do the stats by yourself. The first challenge here, of course, if actually inputting your data. The stats themselves should be pretty simple- you wrote the test on the ethics form and you still have access to all your old Y2 stats lectures. Andy Field's stats videos are a good refresher. If you need the stats helpdesk, that's fine - but don't wait until the day before the end of term to figure this out - they book up fast (all the more reason to get your data collection done early!)

Although you've been taught SPSS, I'd recommend using JASP or JAMOVI for your statistical analysis - it is easy to use (more intuitive than SPSS), easier to interpret than SPSS output (less garbage), and provides nice graphs with decent error bars etc. There's a lot of in depth guidance about how to do particular statistical tests online (e.g., here and here).

Of course, if you really want to get ahead in life you should be learning to code. Running your dissertation analysis in R (I'd recommend R Studio as your one stop shop here) is the perfect way to dip your toes in the water of programming.

All omnibus statistical tests (e.g., ANOVA) and post hoc analyses should be reported in full including effect size (e.g., F(1,16) = 3.68, p=.04, ω2=0.02). You should run your statistics through Statcheck, to make sure they are correct. 

Data visualization

I'm happy for you to present your data however you like, but bear in mind that clear and transparent visualization of your data plays a big part in the readability of your thesis, and this will have a large effect on your final mark. Always include some indication of variance on your plots, such as error bars or, better yet, the raw data. I do a lot of my plotting in Excel because I think it makes a range of graphs which look pretty good, especially for simple designs (detail here). But JASP and JAMOVI have some pretty decent visualizations for factorial design, and of course the gold standard is to plot your graphs using the R function ggplot2 (a tutorial for which can be found here). What I don't want to see are (1) tables repeating data in figures, (2) SPSS output tables pasted into a thesis, or (3) relentless paragraphs of numbers without any breaks or explanations of what they mean - clear communication should be your goal here.