Midterm 1

Take-home Midterm

Despite what I said in class, I've decided that you should work ALONE. However, you may consult any written or printed material, as well as material on the internet. You may discuss your exam with Amy and I, and should do so freely. This exam is meant to be a chance to learn, as well as a test of what you've already learned.

Do NOT turn in a printout, especially piles of commands. I believe that you know HOW to do xlispstat commands, I merely want to know what you did. You can paste small segments of output and graphs into your text as appropriate.All work should be supported by coherent comment. Write as if you were reporting to a statistically sophisticated, paying client. Discuss the choice and the validity of your methods, and the evolution of your data analyses from exploration to formal significance testing.

The exam is due at 5pm, Monday, May 20, in my office or slipped under the door.

Alzheimer Disease

The Mini-Mental Status Exam (MMSE) is a short (hence the "Mini") exam that attempts to quantify the mental status of a patient. Alzheimer's Disease (AD) is a degenerative illness that causes memory loss, disorientation, and other mental problems. An ongoing study at the UC San Diego Alzheimer Disease Research Center (ADRC) administered the MMSE to roughly 400 AD sufferers approximately once a year for up to ten years. From these data, the rate of change of the MMSE score was measured.

It is of some interest to be able to predict how severe the course of the illness will be. It is believed that a low MMSE slope (i.e. a steep, negative slope) corresponds to a rapid decline in health. Possible predictors of the MMSE slope are the MMSE score at the first visit, the age of the subject at the first visit, the estimated age of onset of AD, and the years of education of the subject. (Note, some patients were admitted before the onset, and so the age at first visit is sometimes less than the age at onset.)

These data are available through this webpage. You should "save" the data to your own directory, "clean" it up (by removing the HTML junk), and load it into xlispstat, or the statistics package of your choice. Note that there is HTML junk at both the beginning and the end of the file.

The data are real, and proprietary, so you may not disseminate, publish, etc.

Explore the relationship between the above covariates and the MMSE slope. Use plots, correlations, comment on what you see. Develop a reasonable linear model.

Home