The primary scientific objective of the exam
is to answer the following question:

Primary Question: Does maternal smoking affect birth weight?

A secondary question asks more generally about the factors
affecting birth weight.

Secondary Question: What factors predict birth weight.

The data available to address these questions come from
the Child Health and Development Studies. More details about 
the dataset can be found at www.berkeley.edu/users/statlabs.
A description of this study can be found  by clicking
on "Maternal Smoking and Infant Health". The dataset you need for 
this is found by selecting "Statlabs Data". From there, go to
the dataset " data on parents ...." just under the "Birth Weight 2"
dataset.  This should be the large dataset containing
roughly 23 variables. A description of the variables is provided.
 You might want to rename some of the 
variables before starting. 

To answer the questions you will need to consider possible
confounding effects, such as the weight of the mother, the
mother's level of education, variables related to the father, 
and so forth. A natural way to adjust for these is to include
them in multiple linear regression models. However, variable
selection techniques will be important. Also, you'll want to
look for outliers, influential observations, and use diagnostics
to examine the appropriateness of the linear regression model.

I suggest breaking the report into the following
sections:

1. Introduction: State the aims of the analysis, and give some
background on the data. This should be informative to an audience
who have not before seen these data. However, you should be concise.

2. Methods of Analysis: Discuss how the analysis was done. This
would include how you selected variables, made any transformations,
excluded outliers, and any other analysis details that could apply.

3. Results: Discuss the results in plain words, and refer to
neatly displayed and labeled tables and figures that are most
relevant.

4. Conclusions: Based on the results discuss what you would
conclude from the analysis.

5. Appendix: Include any technical material, figures, or tables
of interest to a statistical audience apart from the main results
of the results section. For example, this might include a log-likelihood
plot for Box-Cox transformation, details about outlier tests, or
plots of Cook's distance.