Dissertation results section
If you wait until you finish data collection to start thinking about your analysis, you've waited too long, says Randy Larsen, PhD, head of the psychology department at Washington University in St. Louis. Rather, before you begin, establish a "road map" that matches your hypotheses to specific analyses that will best test those hypotheses, he recommends. Moreover, try to meet with your dissertation committee to go over the road map, even if your department doesn't require it.
When presenting complex relationships or numerous variables, a good chart, table or graph can make all the difference."You have four additional sets of eyes looking at that plan to see if it's reasonable, " says Larsen.
Not having a road map-and failing to tap the resources you have-are among the common pitfalls students face in finishing their dissertations. Some students get hung up on data analysis, struggling with complex statistical procedures or wasting time on analyses that are tangential to their main research questions. Others have difficulty writing up their analyses in a clear and concise manner that meets professional standards.
The good news is that there are resources that can help you resolve such issues. From statistics workshops to style guides, such resources can help you get your dissertation done on schedule.
After all, says Larsen, "A dissertation is a project; it's got to have an end."
One of the trickiest dissertation challenges is sorting out your data. For particularly thorny statistical challenges, the expertise you need may not be available in your home department. Karen Kaczynski, a fifth-year doctoral student at the University of Miami, began planning a dissertation on gender differences in substance abuse among inner-city Hispanic-American adolescents. Although she was familiar with longitudinal data-analysis techniques, she worried that they might not be appropriate for her small sample size of 20 female participants.
So, she attended several workshops, including the APA-sponsored Advanced Training Institute on longitudinal methods, supervised by Jack McArdle, PhD, a professor of psychology at the University of Virginia.
"I was able to ask him specifically about my concerns with using complex analytic techniques with a limited sample size, " says Kaczynski. "It was really wonderful to be able to discuss my analyses with someone with his expertise." Back at her home department, she refined her analysis with the help of Maria Llabre, PhD, her department's resident statistics expert.
If you plan to continue in academia after your doctorate, delving into the details of complex statistical methods can be a wise investment, experts say. For her dissertation at Washington University, Nicole Speer, PhD, studied event perception using behavioral measures and functional brain imaging. She says that learning logistic regression, Monte Carlo simulations, brain-imaging analysis and other techniques-with the help of her adviser, other faculty members and fellow students-was well worth the effort.
"Even though it definitely ends up taking you more time to learn, in the long run it's really worth it because if you're continuing in research, you'll most likely be using similar techniques in the future, " says Speer.
For computationally intensive analyses, invest in-or have access to-the hardware and software you need. Speer used an APA dissertation grant to buy a powerful computer that shaved weeks off her data analysis, she says. Even so, the analysis took three or four months longer than she had anticipated, in part because she expanded her original plans to address new questions that arose after her data were collected-a move her adviser Jeffrey Zacks, PhD, supported.
"Many data are expensive, so spending time to really digest a dataset is often a wise investment, " says Zacks. As Sharon Foster, PhD, and John Cone, PhD, point out in "Dissertations and Theses From Start to Finish" (APA, 1993), adventitious findings are sometimes the most interesting ones.