Using your advisor’s data: My favorite part about being a professor is mentoring graduate students, and a large part of such mentoring is research. Advisors often love when students work with data they already have. I think all faculty feel as though we have mounds of data and we wish someone would just write it up for us. When students use their advisor’s data, life is easier for the advisor because then advising converges with the faculty member’s own research agenda. The advisor feels very knowledgeable about the topic because it is within their own research area. And, your advisor likely has high quality data, perhaps data that you could not collect on your own. You cannot, for instance, collect a 20-year longitudinal study for your dissertation, but perhaps your advisor already did. Whatever the data, using your advisor’s data is efficient and can lead to the quickest time from dissertation start to finish. However, if you use your advisor’s data, you have constraints on the research questions you can ask. You must work within these constraints, whether they are the sample not being exactly what you might like, the measures being a bit of a stretch for the research questions you want, or perhaps another student is already working on a project with those data that overlaps with some of your research questions. Nevertheless, you can have a very successful dissertation with your advisor’s data. For instance, one of my former students, Sara Vasilenko, did a multi-paper dissertation using all data my colleague and I had previously collected (though Sara was there during some of the data collection). She has multiple papers based on these data, including this paper on how daily affect varies with sexual behavior, as well as a conceptual chapter from her introduction in a volume we co-edited.
Adding questions to advisor’s ongoing project: Some advisors may provide an opportunity to add questions to an ongoing research project, particularly a longitudinal project where you add questions to one or more timepoints. This option is more likely if you have worked with this mentor for a while and contributed to the project already in other ways – you have shown an investment. It is beneficial for your advisor because they have an opportunity for more publications from their project. It benefits you because you can now ask more tailored questions or have measures more tailored to your interests if you simply used your advisor’s existing data. But, it may also give you access to larger or better populations than you would have the time or funding to collect on your own. However, you are unlikely to be able to add every question or every measure you might want if designing your own study – your advisor probably has limits on how many questions could be added. In addition, you still do not have control over the sample or data collection techniques. Despite possible limitations, this option can be very fruitful. My former student Meghan Gillen, was very interested in body image. We were collecting data for a longitudinal study, and she ended up selecting body image measures to add to the project. She ended up writing her master’s thesis and dissertation from these data, and published six first authored papers using aspects of these data (here’s one on the freshman 15).
Data from another faculty member: Frequently my students work with my colleagues to gain experience with different types of data and different mentoring styles. I have had students use data from a colleague in the same department, a colleague in a different department at the same university, and a colleague in another state. In such situations, you may find data better suited to your research interests than data your advisor already has. It provides you an opportunity to learn about new topics, new data collection techniques, or new mentoring styles. And, it provides you with another mentor who can support you and also can be part of your larger network. However, your advisor may be less invested in a project that does not use their data. There may be more negotiation of roles, both mentoring and authorship, when you use someone else’s data. So, the process may be more complicated than simply using only your mentor’s data. Although I’m not sure I’ve ever had a student use a colleague’s data in their dissertation per se, I have had several students work with and publish papers based on colleagues’ data. For instance, Rose Wesche worked both with my then colleague in sociology Derek Kreager to publish multiple papers, such as this one I recently blogged about, as well as another paper on casual sexual experiences with my colleague at Kent State, Manfred Van Dulmen.
Publically available secondary data: Some students find publically available secondary data to address research questions that they cannot address with their mentor’s data. This option has many of the same advantages and disadvantages as using data from another faculty member. A further potential challenge can be getting access to such data. Sometimes there are a number of hurdles required before you are allowed to use such data. In addition, sometimes you need particular conditions, such as a secure computer without internet access in a locked office. Rose also used Add Health partner data for one of three papers in her dissertation. There were some harrowing moments waiting for all of the right permissions to come through for her to be able to access the data, though in the end it arrived in time.
Collecting your own data: One thing I love about mentoring students in research is that they take me in directions that I may not have gone without them, but I learn lots of new things and sometimes develop my own new interest in these areas. Of course, it’s great when students primarily work with data I have. However, I also enjoy when a student takes initiative on a new project and I learn at least as much as they do along the way. When you collect your own data, you have a lot more control over the research questions, design, and sample. It is the most obvious way to do exactly what you want to do. However, it is also costlier – it can cost money to collect good quality data, and it certainly takes more time to collect your own data than to use data someone else already collected. Your funds may not allow you to collect a large enough sample to insure publication, and as discussed already, you are unlikely to be able to collect longitudinal data (other than very short term longitudinal data). And, depending on the culture in your department, your advisor may be hesitant to support you collecting your own data. When I was in grad school and decided to collect my own data (an intervention, with video observations, and four total visits), my advisor was rather resistant. But I persisted, collected my own data, and ended up with five publications from those data, including one in Child Development. My first student – Tanya Boone-Holladay, received an F31 to collect her own dissertation data. Her research had a design based off of my own dissertation data collection. Her dissertation pre-dated the department’s multi-paper dissertation option. We published one paper together from these data, and she published subsequent papers after she graduated. In contrast, Chelom Leavitt collected data on a topic much further removed from my own research. Chelom came to me with an incredibly ambitious dissertation idea – she wanted to collect data on midlife adults (not an easy to capture population), in married relationships (more constraints), in three different countries speaking three different languages (the translation!). Using mostly measures of constructs I had never measured. She did it. And her first paper from her dissertation is now in press.
A combination: In one way or another, many of my students, particularly students who did multi-paper dissertations, chose a combination of these options. With a combination, you can maneuver around many of the challenges described above – you have access to high quality data you may not be able to collect on your own, but can also design your own smaller study to address more specific questions exactly as you’d like to. The challenge might be that you make things more complicated for yourself, in that you may introduce some of the challenges of using secondary data, or multiple collaborators, or getting your mentor’s buy-in. Emily Waterman used data from one of my projects, and also collected her own small-scale data to address more specific questions, getting the best of both worlds.
Which option you choose depends on a number of factors, some in your control and some not. Your research questions, your timeline, the data your mentor has, the culture in your department, and your advisor’s preferences among others. When making a decision, consult with your advisor and other informal mentors to figure out the best option for you.
“Where should your dissertation data come from? first appeared on Eva Lefkowitz’s blog on June 11, 2019.”