Department of Statistics, The Ohio State University



 2002 Program Highlights & Details

Summer 2001 Experience Report

 How To Apply?

If you wish to inquire further about the summer program or have questions, please send an e-mail to

The graduate programs in statistics and biostatistics in the Department of Statistics (, the Center for Biostatistics (, and the School of Public Health at The Ohio State University will once again host a Research Experience for Undergraduates (REU) program during summer 2002.  This follows the highly successful program that took place last year.



Each participant will be assigned a project requiring the application of statistical methods to real problems.  During the first five weeks of the program participants will also take two courses.  The first course deals with the proper design of experiments and appropriate statistical methodology for analyzing the resulting data.  The second course focuses on introductory probability and statistical theory as well as associated mathematical tools.  In these courses, participants will learn to reason statistically, use statistical software, manage data, and perform standard statistical analyses.  Throughout the entire eight weeks participants will meet with faculty mentors and other statistics faculty, staff, and graduate research associates concerning the subject matter issues related to their specific projects.  Each participant will prepare a written report documenting the relevant scientific questions and the associated statistical analyses for their project and make a verbal presentation of their report to the other participants.  A poster session will be organized at which the program participants can also share their summer experiences with the scientific community at The Ohio State University.


To apply, please send the following:

(a) short biographical statement, including your reasons for wishing to participate in the program,
(b) two letters of reference,
(c) an official copy of your undergraduate transcript to date, and
(d) copies of the scores for any standardized tests that you have  completed (SAT, ACT, or GRE, if available) to

Professor Douglas A. Wolfe
Chair, Department of Statistics
404 Cockins Hall
1958 Neil Avenue
The Ohio State University
Columbus,  OH  43210-1247

The deadline for receipt of completed applications is February 18, 2002.  Final selection of the participants for the summer research program in applied statistics and biostatistics will be completed by March 11, 2002.

Report on Summer 2001 REU in Biostatistics 
at Ohio State University

2001 REU Participant List

During the summer of 2001, we invited the following ten students to participate in our first REU program with an emphasis on biostatistical applications.
Student Name
Undergraduate Institution State of Residence
Jason Gershman Rice University Florida
Jodie Jansen Luther College Minnesota
Amy McGregor Virginia Tech University North Carolina
Anne-Michelle Noone Boston University Michigan
Colin O'Rourke University of New Mexico New Mexico
Lynne Peeples St. Olaf College Washington
Tania Robbins Purdue University Indiana
Jennifer Ryea University of New Hampshire New Hampshire
Stacey Stillion Ohio Northern University Ohio
Brandy Wiegers University of Idaho Idaho

Relying on contacts made by faculty and staff of the Center for Biostatistics and the Department of Statistics, the REU participants were assigned to a wide range of clinical and basic science research problems.

We secured a large house (with nine bedrooms) owned by the university to provide a 'home' for the exclusive use of the REU students during their time at Ohio State. The students participated in a variety of activities (both academic and social) during their eight-week program at Ohio State with constant support and advice from two of our very best (both entered our program under University Fellowship support) senior graduate students, Ms. Kristin Blenk and Ms. Marilisa Gibellato. These activities included both course work and research on a project involving the application of statistics/biostatistics. The coursework included selected short courses in biostatistics topics during the first two weeks of the program to five-week term courses in mathematical statistics or biostatistical methods. The students were involved in major research projects during the entire eight weeks of the program. The topics and scientific mentors for their research projects and brief summaries are given below. These are taken from the student presentations given at the end of the program.


Jason Gershman (Advisor: Dr. Gary D. Stoner,  OSU Cancer Center)
NMBA Induced Carcinogenesis in the Rat Esophagus

Esophageal cancer is an extremely deadly disease, with approximately 12,000 new cases diagnosed each year in the United States and thousands more worldwide. Chemoprevention is the prevention of cancer by chemicals found in drugs or in the diet. Our study dealt with the testing of a potential chemopreventive agent. We studied one type of esophageal cancer, Squamous Cell Carcinoma (SCC), which is found at high rates in parts of China. The animal model used was nitrosamine (NMBA) induced cancer in the rat esophagus, which mimics human esophageal SCC. The potential chemopreventive agent analyzed in this study was piroxicam, which is an anti-inflammatory drug shown to effectively inhibit various cancers in rodents.

The statistical analysis compared tumor size and multiplicity between three groups of fifty rats: those given just NMBA, those given NMBA and a low-dose of piroxicam, and those given NMBA and a high-dose of piroxicam. An identical number of rats in each group were sacrificed at each of three different times. In comparing tumor size data, most tumor sizes in all groups and all times were small, between 0.1 and 20 cubic millimeters, but a few tumors were large, up to 150 cubic millimeters. A logarithmic transformation of tumor sizes improved the dataís normality and pair-wise t-tests were performed between the groups at the three times. No pattern of statistical significance was found between the three groups over time. Under proper assumptions, tumor multiplicity data was modeled as a Poisson random variable. The appropriate likelihood ratio test was performed and it found that there was no statistically significant difference in the tumor count data in the three groups. The findings are consistent with intuition, based on graphs of the data, as well as the findings of the Stoner laboratories.

Jodie Jansen (Advisors: Dr. Sunny Kim, Lynne Giljahn; Ohio Board of Health, Columbus)
Analysis of Adolescent Obesity & Nutritional Behaviors in Ohio

Adolescent obesity is quickly becoming one of the most pressing health risks facing youth in America. Not only does adolescent obesity affect nearly every organ system in children; it can have detrimental affects on the psychological welfare of the child. Detailed analysis of the Center for Disease Controlís Ohio Youth Risk Behavior Survey was conducted to further investigate the nutritional and psychological behaviors of overweight and obese adolescents. Using STATA (a statistical software package), numerous chi-squared and correlation tests were used to find relationships between many of these behaviors.

Overall, significantly more females than males described themselves as overweight. However, there were no significant differences between or within different races. Significantly more females (64%) than males (21%) reported they were trying to lose weight. Also, significantly more overweight females (89.2%) reported trying to lose weight than their male counterparts (55.4%). Over 60% of students that were dieting were using both exercise and eating less as a means to do so. However, only 1.6% of adolescents in Ohio ate the recommended daily allowances of fruits, vegetables and milk. Of this minute percent, males were significantly more likely than females to be "healthy."

Amy McGregor (Advisors: Dr. Haikady Nagaraja, Dr. Philip Diaz; OSU Pulmonary and Critical Care Div. )
Analysis of Longitudinal Data on Lung Damage in HIV Positive Patients

The % of alveolar macrophages containing hemosiderin obtained from broncoalveolar lavages of HIV+ and HIV- patients from the central Ohio area were analyzed in six month intervals from 1993-1998; where the average number of visits per patient is 4.2. Using SAS JMP student version 4, a baseline comparison of HIV+ and HIV- patients (Control: n=31, HIV+: n=89) matched for age and smoking status showed a significant difference in the % of alveolar macrophages that did not contain hemosiderin (Control: 66.88%, HIV+: 76.65%, p = .0391). Over time, there was a slightly negative trend in the % of alveolar macrophages in HIV+ patients (p=.0567). We found no significant difference between HIV+ smokers and nonsmokers as we had predicted (p=.3275).

Anne-Michelle Noone (Advisor: Dr. John Orban, Battelle Memorial Institute, Columbus)
Analysis of Driving Behavior

The United Sates Department of Transportation (USDOT) began the Intelligent Vehicle Initiative (IVI) in order to improve the safety systems in commercial trucks. Specifically, in a partnership with Freightliner and Praxair, they are testing a system with a roll stability advisor and a roll stability control. The roll stability advisor will provide advisory notices after completion of a dangerous maneuver and in-vehicle messages. The roll stability control is able to take partial and temporary control on the vehicle while in a dangerous maneuver. Dangerous maneuvers will be determined by a preset criterion. A few of the overall evaluation goals are to determine if drivers drive more safely with the system, to estimate the number of crashes that will be avoided, and to determine if the system can be deployed nationwide.

This aim of this project was to analyze the baseline data in order to find any possible confounders. The associations between conflict rate and diver age (p=0.7556), conflict rate and driver experience (p=0.8463) were considered and these were not significant. The association between wiper status and conflict rate, where conflict rate was calculated for each driver, was considered and that showed a significant relationship where conflict rate was lower with wipers on. Drivers being more cautious in inclement weather can explain this; however, more analysis is needed to examine the relationship between drivers. Also, the fill status of the trailer showed a slightly lower conflict rate for a full loaded trailer but this also requires further analysis. And, finally, the hours into a driving session were considered. A scatter plot of hours into session and conflicts per hour showed a slight upward trend indicating an increase of conflict rate as time progressed, but this is also inconclusive because the road conditions, point into trip, and break times need to be considered.

Colin O'Rourke (Advisor: Dr. Elizabeth Stasny; OSU Center for Survey Research)
Modeling Response Mode for the OSU Poll

Once a year, the Ohio State University conducts the OSU Poll, a survey of students, staff, and faculty at the university. A unique feature of the OSU Poll is that it employs two survey modes, Internet and telephone. E-mail is sent out to all sampled members of the population asking them to respond to the survey on-line. This e-mail includes an identification number and password which are required to access the website. Sampled individuals who do not complete the web-based survey in the first few weeks are given a phone call, reminding them to complete the Internet survey. If the reminder is not effective, the survey is administered by telephone to sampled individuals who can be reached. Typically, Internet surveys are not practical due to large coverage biases. Some subjects in the population do not have a computer, have limited knowledge of computer usage or have software incompatible with the survey (i.e. old version of Netscape). It is thought that these concerns might be mitigated when surveying a university population since the university provides e-mail accounts and computer access to all faculty and students.

This project dealt exclusively with the undergraduate population. Respondents who completed the web-based survey were compared to those who were interviewed over the phone. The following question was asked: In light of the accessibility of computing resources at the university, is it still necessary to use a multi-modal approach to reduce bias in this survey? A regression model found characteristics within the undergraduate population that were predictive of the mode that the student used to respond. The resulting conclusion was that there were some differences between the groups who answered by Internet and telephone and that a solely Internet based poll is likely to produce biased estimates because it would systematically exclude part of the population.

Lynne Peeples (Advisors: Dr. Clara Bloomfield, Kellie Archer; OSU Cancer Center)
Cytogenic Abnormalities of the Elderly in Acute Myeloid Leukemia

Acute Myeloid Leukemia (AML), a cancer of the blood cells, is a rapidly progressing disease with a generally unfavorable prognosis. In our study, we characterized cytogenetic abnormalities commonly seen in de novo adult AML patients, examining the survival duration of patients with abnormalities categorized as the core-binding factor (CBF) type of AML. These have been shown to be significantly correlated with more favorable prognosis. Further, we compared younger and older patients (under 60 years and 60 years or over), adjusting for various confounding factors, including high-dose Ara-C treatment (HIDAC), white blood cell count, and previous leukemia infiltration to other body organs.

Using a Cox Proportional Hazards model, we were able to assert a prolonged survival for patients treated with HIDAC and/or diagnosed with CBF leukemia. After adjusting for these variables in the model, we still found a significant difference in the survival durations between younger and older patients with the same favorable prognostic cytogenetic abnormalities. However, the data also suggested the possibility that older patients with CBF AML had longer survival durations than younger patients, also with CBF AML, when treated with high-dose Ara-C. This brings into question the current protocols that limit HIDACís use in the treatment of patients older than 60 years. The findings of this research may further the understanding of the role of these cytogenetic abnormalities in relation to prognosis and treatment response among elderly AML patients.

Tania Robbins (Advisors: Dr. Fred Wright, Dr. William Lemon, Sandya Liyanarachchi; OSU Div. of Human Cancer Genetics)
Expression Profiling Using Model-Based Estimates

Gene expression from Affymetrix chips is currently estimated with the Average Difference and Log Average methods. These current methods take the difference in Perfect Match ó Mismatch intensity or the log (PM-MM) intensity and average it over the number of probes. A new method, which has been proposed by Li and Wong (Li C, Wong WH, 2001 PNAS 98:31-38), models gene expression by taking into account probe effects and systematic variations. The two models are compared both analytically and experimentally.
Data from gene expression of Acute Myeloid Leukemia (AML) were taken and compared using the methods. Hierarchical clustering diagrams show the difference between the two methods. The Li-Wong model is better able to differentiate among AML+8, AML-CN, and control groups. T-tests were performed using the Li-Wong estimates between the AML+8 and AML-CN groups to identify differentially expressed genes. A list shows the top 20 genes that are more expressed in AML+8 than AML-CN. Another list shows genes which are much more prevalent in AML-CN than AML+8. These lists correspond to known expression patterns.

Jennifer Ryea (Advisor: Dr. Steven Clinton; OSU Cancer Center)
A Comparison of Immunostains for CD34 and Factor VIII as Biomarkers of Prostate Cancer Angiogenesis

The goal of our study was to evaluate the objective quantification of vascular density by immunohistochemical stains and their relationship to a known biomarker of prostate tumor aggressive behavior (Gleason Score). The samples were obtained from 268 patients in the Health Professional follow-up study (from HSPH). Using computerized digital image analysis we were able to obtain an objective quantification of vascular density. Our findings include strong negative relationships between Gleason score and the area, diameter and perimeter per blood vessel for CD34 and a strong positive relationship between Gleason score and the number of blood vessels for CD34. However, with FVIII strong negative relationships were found between the Gleason score and the mean area, mean diameter, area sum and the number of blood vessels in the samples.

Stacey Stillion (Advisor: Dr. Fernando G. Cosio; OSU Div. of Neurology)
Covariates of Pancreas Transplant Survival in Recipients of Combined Kidney-Pancreas Grafts

Diabetes is the most common cause of kidney failure in the U.S. These patients may be treated with dialysis (mechanical replacement of renal function) or transplantation. The latter may include only a kidney or both a kidney and a pancreas. Simultaneous pancreas-kidney (SPK) transplantation is the best therapy available for patients with Type 1 diabetes mellitus (insulin-dependent) and kidney failure. This project examined the pre-transplant and post-transplant variables that correlate with pancreas graft survival and the correlation between kidney survival and pancreas survival. The rationale behind these studies is the hypothesis that the survival of the kidney graft and the survival of the pancreas graft are biologically and statistically interrelated. Results of logistic regression, Cox regression, and Kaplan-Meier curves have shown that kidney survival and pancreas survival are indeed highly correlated in SPK recipients and that the correlation is statistically independent of other covariates of graft survival.

Brandy Wiegers (Advisors: Dr. Steven Clinton, Dr. Zhiming Liao; OSU Cancer Center)
Prostate Tumor Morphometrics in Rats Fed Diets Restricted in Energy

A previously published study (JNCI 1999) by Mukherjee,P. et al. Entitled "Energy Intake and Prostate Tumor Growth, Angiogenesis, and Vascular Endothelial Growth Factor Expression" tested the effect of dietary restriction on tumor growth. This past study used three sets of rats bearing a transplantable prostatic tumor (Dunning tumor): one control group allowed to eat ad labitum, a dietary restricted group, and a castrate group also allowed to eat ad labitum. That experiment showed a significant difference in tumor growth, in the weight, size, and architecture exists between the different treatment groups. This current study expands upon the previous by objectively quantitating tumor architecture using computerized digital imaging and technology.

Tumor architecture (specifically gland matrix ratio and gland morphology) of AR prepared slides was analyzed using Adobe Photoshop and Image Pro Plus software. It can be demonstrated that there was a significant difference in percent gland area of a high powered field between all three treatment groups. There was no significant difference in the number of glands in a HPF between the groups. A bias against larger glands complicated gland morphology analysis that showed a significant difference between the control and castrate group but none between the control group and the restricted diet group for percent area of individual glands and roundness of glands. Conclusions supported the use of computerized technology to study tumor architecture and an effect on tumor architecture by diet restriction.

At the end of the program, each REU participant prepared a poster describing the activities and research results from their project and each gave a twenty minute oral presentation of their work at a final REU symposium day. These posters and their talks were excellent across the board!

Finally, we asked the participants to complete a questionnaire addressing various aspects of their REU experiences. The overwhelming sentiment expressed by these participants was that the REU program was very successful in achieving its multiple goals. Every single one of the participants indicated that they both learned a lot during the summer and that they enjoyed both their course work and their research projects. In response to a question about future plans, all of the respondents indicated that this REU experience convinced them to pursue graduate work in either statistics, biostatistics, or mathematical biology!! Some selected comments from the students' completed questionnaires are given below.

2001 REU Participant Sample Responses

1. How would you rate your REU experience as a whole?

"Definitely a great experience. For this being the first year of this REU, it was such a great, well-run program. I've had numerous friends participate in well established REUs in pure math and in physics at top universities, and each one has been disappointed. I think I'm the first person I know to be truly happy with an REU"
"It was definitely an educational summer that will benefit me for a lifetime."
"Honestly, this has been an overwhelming (in a good way) experience. I've learned SO much about both statistics and some growing areas of cancer/genetics research. While there were some periods of high stress & frustration--even that was a good experience in itself, as I was sure it is encountered in the "real", practicing world, too."
"It was a very, worthwhile experience. The last couple of weeks was stressful, but most of the time it was very relaxing. Most of the people were great and it was a nice change of pace."

2. What is the most valuable thing you learned while you were at OSU?

"I think everything ties for first. Well, I think the most important life lesson is that when you have great people at all levels (from program directors, to TAs, to students), great things happen. And they did this summer."
"ALWAYS have a plan before you begin analysis, and NEVER be afraid to ask questions."
"I learned that grad school could be fun because research can be worthwhile to all. Also, meeting most of the people in grad school makes me feel that it is not as inaccessible as it seemed before."
"The value of the complete research experience--and how much fun that can be : ) Thank you!"

3. Has your experience at the REU influenced your thoughts about graduate school? In what ways?

"Several. I have realized that there is much more out there than just Tech (Virginia Tech), and that I have so many options of very interesting careers.
"Yes, it made me realize that there are lots of people pursuing higher degrees. Often I feel pressure not to, or that it is a waste of time. So it was good to see how many people are still studying."
"Yes. I didn't know for sure that I wanted to go to grad school for statistics before attending the REU. Now I do!"
"Yes--it encouraged me to continue to pursue research opportunities available in the education environment."

4. Has your research experience at the REU changed your impressions of academic or clinical research? If so, how?

"A bit. I guess I hadn't realized before how tied together they can be (the academic & clinical sides). Going to a small liberal arts school, I don't see the same research going on around me like I have here at OSU. Actually, I could now see myself happy in a career like that..."
"It made research seem much less scary and much more fun. I loved my project, and I will stay in touch with my adviser as he continues the research, just because I'm very interested in the results."
"I was excited to be able to participate in the complete research experience -- from collecting and analyzing my data. And having this experience just recommits me to a life of research! : )"

5. Any suggestions for improving the REU experience?

"Not really. If there is every anything I can do to help out the REU in the future, don't hesitate to ask. It would be a real shame if the NSF did not fund this REU again, as there are so many mediocre REUs out there that to lose a real quality one here at OSU would really bother me. Everything was great, the research, OSU, Columbus, and Ohio in general."

6. Are you planning on attending graduate school in statistics or biostatistics? If so, do you plan on pursuing a Master's degree, or a Ph.D.? If not, what do you plan to do following graduation?
(All Responses.)

"I am planning to pursue a Ph.D. in Biostatistics."
"Sure, I'll go to grad school for statistics or biostatistics, most likely for a Ph.D."
"I want to pursue a Ph.D. in either Biostats, Stats, or Mathematical Biology."
"Yes, especially after my experience this summer, I've been convinced to pursue biostatistics--not sure which degree, but I'm leaning towards a Ph.D."
"I will be getting my Ph.D. degree in statistics or biostatistics."
"Yes, statistics. I would really like to go to grad school for both stat and education and do education research and teaching at a university."
"No. Ph.D. in math."
"Yes, in 2 years I will be going into a Ph.D. program after I get my Masters."

Update: Oct. 24, 2001