Feedback from Karen Smith (PhD 2003)
She writes: "A lot is expected of PhD students today, and no-where is this statement truer than in statistical analysis and interpretation. There are many things that can make you consider giving up a PhD, and analysing data is certainly one of them. The scope and depth of the PhD means that you end up with a huge volume of data to be analysed and condensed into a thesis format.
Statistical analysis is just one more thing you have to take responsibility for in a PhD. No-one will force you to consult with the Statistical Centre, some supervisors may actually discourage you from doing this. It is up to you to take the initiative. I think it is worthwhile recognising that stats is going to be a big part of your project and going along to meet a consultant sooner rather than later is certainly a good idea.
I was lucky to have a good background in statistics through postgraduate training provided by my faculty (Institute of Land and Food Resources). My experimental design was a classic one, and my supervisors were able to adequately advise me at this stage. So the data analysis stage was the the first time I consulted the Statistical Consulting Centre. For some, the design stage may be a more appropriate time to talk to a statistician, to bounce off ideas and get a different perspective from your supervisors.
I knew how to use basic statistical software and how to check assumptions before analysing my data, but the sheer volume of data was overwhelming. I needed some re-assurance getting going with the analysis. Whilst my design was a classic multi-factorial one, it is one that is confusing to analyse, and can easily be made incomprehensible at the thesis stage.
My stats consultant made sure my analysis was appropriate, helped me with the software, and gave me some useful tips on the nuances of the more difficult analyses. He was also useful in helping me present my design, analysis, and interpretation in an accessible manner. My examiners commented on how well written my thesis was, and how well presented my data was. I worked very hard at data presentation and formatting of tables whilst writing up, so it was great to have this acknowledged.
My supervisors did not necessarily approve of my approach to stats and my presentation in the thesis. But I decided it was my thesis and I had a right to persue what I wanted. Being experienced and capable, my supervisors coped with this, and in the great tradition of the PhD, let me go. However they did make me engage in plenty of robust debate, with the effect that I had to justify and explain my reasons and methods of doing things statistically. Looking back, I see this as a really constructive part of my professional development in the PhD. I'm sorry folks, you can't avoid statistics in a PhD. However, you can make life easier for yourself by accessing the resources the University has made available to you through the Statistical Consulting Centre."
Note. Karen's thesis, 'The effects of properties of designed soils on growth of Corymbia maculata', was submitted and successfully passed in 2003.