You may know the feeling. You have collected a lot of new data on an important experiment. Now you are faced with multiple groups of data, a sea of numbers, and a deadline for submitting your paper to a peer-reviewed journal. And you are not sure which data are relevant, or even the best way to present them.
The statisticians at Data Management Services (DMS) know how to help. This small group of experts provides a wide array of statistical and mathematical consulting services to the scientific community at NCI at Frederick and NCI-Bethesda. They work in concert with the investigators to formulate the appropriate question(s), design the experiments, collect the data, and interpret the results. They also contribute to the publication of scientific, peer-reviewed manuscripts.
Asking the Right Questions and Getting to the Answers
Conjecture, or asking the right question, is how all scientific research begins, according to Greg Alvord, Ph.D., director of Statistical Services, DMS.
“Formulating what the investigator is trying to measure or investigate is the key to the entire process,” Alvord said. “When investigators first come to us, they may not know what analysis may be required or what, exactly, they are trying to measure. We help them phrase or formulate the questions that will lead to the answers they seek.”
The next challenge for the statistician is to develop the most effective way to answer the question, Alvord said. He noted that general research questions naturally lead to more specific research hypotheses and refinements and, thus, to further analyses, methods, and approaches.
For example, a researcher may need help in determining the appropriate sample sizes, the best method for analyzing data from different treatment groups or from subsequent generations of samples, or the best way to display the data or other elements critical to an experiment.
“We help investigators formulate what they are trying to measure, evaluate, or analyze,” Alvord said. “In a nutshell, we help them design the experiment, as well as gather, format, organize, and visualize the data. We then write one or more scientific reports, which act as the basis for scientific, peer-reviewed publications.”
The Best Advice: Come Early
The best time to seek help from a statistician is before the experiment begins. “We encourage researchers to meet with us before they feel they have a need for our services,” Alvord said.
Sometimes researchers ask for a statistician’s assistance after they have had a paper rejected because of inadequate or inappropriate statistical analyses, he said. “When the investigator comes early, many of these problems can be avoided.”
Who Are the DMS Statisticians?
Alvord and his staff, Rob Leighty, Ph.D, and Matt Fivash, provide critical support to the research of NCI at Frederick. Leighty, a senior statistician who has been with DMS for eight years, said his work often involves “reformulating questions in terms of the statistical models.” He uses appropriate methods to test for the significance of parameter estimates and performs other statistical analyses that “add legitimacy to and confirm the statistical findings.”
Leighty was not always interested in statistics, however. He said he took undergraduate courses in the subject “because it was an easy A for a mathematics major.”
That all changed when he took a mathematical statistics course from Alan Gelfand, Ph.D., currently the J. B. Duke Professor of Statistics and Decision Sciences at Duke University. “Then, statistics came to life for me,” he said. Leighty went on to earn his Ph.D. in statistics at the University of Connecticut, and, following postdoctoral training, began consulting with scientists. That was when he realized how important statistics is for research, he said.
When not counting numbers, Leighty may be found counting steps in a classic ballroom dance. “I enjoy ballroom dancing with my wife,” he said. “My favorite dance is International Foxtrot.” He and his wife also enjoy attending the Baltimore symphony, and he likes to stay fit by swimming, using a stationary bicycle, and walking his dogs.
Fivash, a statistician at DMS since 1989, is responsible for developing analyses of multiple screening assays and other mathematical analyses. An expert in the analysis of surface plasmon resonance signals and related differential equations analyses, Fivash earned a master’s degree in mathematics from American University. Developing the appropriate analysis methods is the most challenging part of his job, but once that is accomplished, he said, he enjoys turning “a cloud of numbers into usable biological information.”
Fivash likes to work with his hands when he’s not working with numbers, and in his spare time he enjoys making furniture. He said his preference is to use hand tools over power tools because he can work faster. Some of the pieces he has made over the years include a poplar desk for his father and a cherry table for his wife.
Alvord, who came to DMS in 1985 to establish a statistical services department, earned his Ph.D. in statistics from the University of Maryland and completed additional graduate work in basic and biological sciences. An expert in the use of the R statistical language and environment, Alvord has authored or co-authored more than 150 peer-reviewed publications.
Outside the office, Alvord leads a men’s book group in discussions on such diverse topics as science, consciousness, cognitive psychology, history, philosophy, and literature. He is also an exercise enthusiast, and he participates in West-Coast swing dancing events.
For more information, or if you need help with a specific project, you may contact any of the statisticians directly, Greg Alvord (new clients), on 301-846-5105; Rob Leighty, on 301-846-5378; or Matt Fivash, on 301-846-5118.