Dr Roxanne Connelly, University of Edinburgh 2021
Studying for a Masters or PhD is a unique experience, both challenging and enjoyable. The web is full of top tips for new graduate students, but this blog specifically focuses on practical tips for students using quantitative research methods. Most of these are things I wish I would have known and stuck to as a young researcher. Getting to grips with these ways of working will stand you in good stead for a career as a researcher.
Tip 1: Use a Reference Management Software
This tip is really for everyone, but it is too important to leave out. You should use a reference management software to organise your references (e.g. Zotero or Endnote). You want to automate the referencing process and minimise the risk of errors. Reference management software will do this. You absolutely do not want to spend long periods of time formatting references, and this can be easily avoided by dedicating a little time to learning how to use a reference management software early in your studies. Reference management software also allows you to switch easily between referencing styles, which may be necessary when preparing your work for publication.
Tip 2: Get Organised
How easily can you find files on your computer? Do you ever struggle to find the most recent version of a file? Have you ever lost a file? These are things you want to avoid.
A useful first step is to organise the file structure of your computer and to start using a standardised file naming protocol. There is no one correct way to organise or name your files, but the key advice is that your file organisation should be planned and not ad hoc.
I highly recommend ‘The Workflow of Data Analysis Using Stata’ by J. Scott Long which contains a detailed chapter on ‘Planning, Organizing, and Documenting’. Note whilst this book is aimed at Stata users the advice throughout is also applicable to students using other software.
Tip 3: Plan your Workflow
Closely connected to organising your files is putting thought into the organisation of your whole data analysis workflow. It is difficult to define the ‘workflow’ as it encompasses every part of data analysis from planning, cleaning data, undertaking analyses, presenting results and archiving your work. Without a planned workflow it is very easy for undetected errors to creep into your work, or for information to be lost. Professor Vernon Gayle provides useful advice in this online resource from the National Centre for Research Methods.
Planning your workflow will help you be a better data analyst and a better researcher. It can also help you be more relaxed, and avoid stressful errors in the research process.
Again, ‘The Workflow of Data Analysis Using Stata’ by J. Scott Long is the key text for those who want to improve their workflow.
Tip 4: Write Code for the Future You
As a quantitative researcher a large amount of your time will be spent writing code to clean, organise and analyse your data. When writing commands in software such as Stata, R or Python your primary focus will probably be on getting things to work. Alongside this you should not forget the important role of your code in keeping a record of what has been done. Your code should document all your data preparation, and your data analysis. You should also ensure that you provide comments throughout your programme in order to document what has been done and why.
You are likely to return to your code over a period of years. When you do so, if you have not thoroughly documented and annotated your programme, you may be utterly confused by what you see in your file.
Your first concern should therefore be helping future you! As you move towards publication you will also want to write code that will be suitable to share with others in the wider research community so they can fully understand and build on the work you have undertaken.
Tip 5: Get used to making things open
There is increasing concern across a wide range of social science disciplines that empirical results cannot be reproduced because of a lack of transparency in the research process. You should think about how you will share the materials associated with your research to maximise transparency and reproducibility. Sharing data can present challenges for social scientists, particularly when using secure data resources (see here for a detailed discussion). But sharing code is almost always possible.
Christensen, Freese & Miguel’s Transparent and reproducible social science research, and Ritchie’s, Science fictions: Exposing fraud, bias, negligence and hype in science provide excellent introductions to these issues.
Tip 6: Date your data
As quantitative researchers our main focus is often on data analysis (e.g. what type of model should I run?). It is easy to overlook the importance of your data. Your data is at the foundation of everything you will do. You should spend some time getting to know your data. You should carefully read the data documentation to fully understand how the data were collected, you should consider the degree of missingness, how the data files are structured, and how the variables have been operationalised. This will help you to understand whether you have to take the complex design of a survey into account in your analyses, or the extent to which the data requires a principled approach to handling missing data. The nature of your data should not be overlooked, and investing that quality time with your data will improve the quality of your work.
Tip 7: It’s ok not to know
Finally, a reminder that it is valuable to reflect on the gaps in your understanding. This is the perfect time to review foundational concepts. You should also think about where you want to extend your expertise, and whether you want to develop a specialism in a particular method. When using quantitative methods it is certainly true that ‘the more you know, the more you know you don’t know’. It is always ok to ask questions and admit that there are things you don’t understand.
We are all always learning and even a professor in the field for many years will have to revisit concepts.