Degree In Sight

More data, less hassle

Your dissertation demands data, but your lab, your students and your adviser—not to mention your life outside of school—demand your time. It's tricky to strike the right balance, but you can take steps to make data collection more efficient, researchers say. It may sound like a Zen koan, but Patrick McKnight, PhD, a clinical psychologist at George Mason University in Fairfax, Va., says one of the best ways to gather data for your dissertation is not to collect it at all.

"Data collection is very time intensive," he says. But more important, many graduate students undertake data collection without knowing enough about the area and without sufficient training in research design or data analysis, he points out. "They end up with a mess."

Instead of running your own participants, McKnight recommends finding existing datasets for your question. Universities typically have bountiful data archives from past studies. Many have searchable databases online.

In fact, there's so much unused data out there already, "For many areas, it's a crying shame that students are collecting data on new subjects," McKnight says.

Among the richest databases for students to tap into is the Inter-university Consortium for Political and Social Research, which houses the world's largest digital archive of social science data. If your university is a consortium member, you can browse its database or ask your university for help.

Another database option is the Add Health database, run by the National Institute of Child Health and Human Development and 17 other agencies. This database contains longitudinal health and behavioral data on 21,000 individuals from 1994 to the present. There are public-use data as well as restricted data you can access by paying a fee. (For more information on public databases, see "Information gold mines" in the September 2008 gradPSYCH.)

By taking advantage of such existing datasets, McKnight says, students can sometimes shave up to a year off their time to dissertation completion.

Webbed feat

What if you've scoured the databases and literature for relevant data and come up empty? Try SurveyMonkey.com, suggests David Wimer, PhD, a postdoctoral intern at Pennsylvania Counseling Services in Reading, Pa. The online survey site and others like it, such as PsychData.com, HostedSurvey.com and SurveyGizmo.com, can be ideal for students whose studies involve questionnaires.

Wimer used the Web site to gather research data from college students while working on his dissertation two years ago as a temporary lecturer at Kutztown University in Kutztown, Pa. Since college students are on the Internet all the time anyway, it was convenient for them to just click a link and go right to his questionnaire.

"You don't even have to physically go into classrooms to recruit," he says.

Over the course of a year, about 300 students took his online survey, providing him with more than enough data for his research on the intersection of health risks and masculinity.

Getting noncollege populations to respond to online surveys is more challenging, but being creative can help. For instance, Wimer and his former adviser Ronald Levant, PhD, a clinical psychologist at the University of Akron in Ohio, wanted to investigate health risks among middle-age and older men, so he posted links to their online survey in an Internet message board dedicated to classic rock. They got more than 200 responses in a couple of months.

Of course, Web-based surveys have clear limitations. In a report published in a 2004 American Psychologist (Vol. 59, No. 2), researchers outlined some of their concerns about conducting research over the Internet. For example, it's almost impossible to eliminate sample biasing on the Web, the authors say. Even though more and more people regularly use the Internet, users are still more likely to be white and young than the general population.

Also, you can't control the setting in which people take the survey. "In the laboratory, for example, an experimenter can verify subjects' identities, age or gender; can tailor instructions to each subject; can monitor their behavior to ensure that they are involved and serious," the authors write. "The anonymous nature of the Internet allows people to participate frivolously or with malicious intent."

In light of these limitations, you might decide that a Web survey isn't the right tool for you. But even so, you can harness the Web to recruit people to come into your lab for your study, which can be more efficient than recruiting in person, Wimer says.

Tag team

If you do have to conduct in-person data collection, try to do it as part of a team.

In addition to helping you stay on track and divvying up the work, teaming up with like-minded students and researchers can help you streamline your ideas and is good practice for a world where collaborative research is the norm, says Robert Levenson, PhD, a neuropsychologist at the University of California, Berkeley.

In Levenson's lab, grad students typically work in groups of three. They each have their own research questions, but they collect and analyze the data together. As a result, the work gets done more quickly and students explore research avenues they might not have considered on their own.

"When you get three smart minds together, unexpected successes are almost the rule rather than the exception," he says.

Be sure, though, that when you collaborate you don't piggyback too many questions, Levenson says. The more you load up a study with measures and dependent variables, the more noise you introduce into your results.

"It's a 'sweet spot' problem," Levenson says. Too many measures in one dataset can make it tough to tease out meaningful results. Too few and you're not making efficient use of your time. The trick, he says, is to "measure only a few key things and measure them well."

That advice goes for all data collection, says Lore Dickey, a fifth-year counseling psychology graduate student at the University of North Dakota in Grand Forks. Making your research questions needlessly complex won't help you collect valuable data and it will only confuse participants in your study.

"Simple is better," Dickey says.

Article Sidebar