Description
High systolic blood pressure is a strong predictor of heart attacks and strokes. A health researcher
wants to know how average systolic blood pressure varies with:
• Age
• Height
• Weight
For ease of interpretation, she does not wish to transform systolic blood pressure, but she is
willing to consider interpretable transformations of the explanatory variables. She thinks the trends
should be relatively smooth, but not necessarily linear. She suspects that for at least some of these
variables, the trends may be different for men and women. In addition to estimating the trends,
she wants to know how close observations typically are to the trend and whether the models might
have any explanatory value. However, she is not interested in making predictions for individuals.
She is not interested in formal inference right now, though she may be in the future. She knows
some R, so you may include R code in your report, but she can’t read your mind, so label your
graphs.
The questions
Use the NHANES data in the NHANES package to explore the researcher’s questions. The relevant
variables are:
• BPSysAve (the average of three measurements of systolic blood pressure)
• Age (in years; 80 or older is recorded as 80)
• Weight (in kilograms)
• Height (in centimeters)
• Gender (male or female)
1
Write a document in three sections, giving the relationship of average systolic blood pressure
with age, height, and weight respectively. (You can also include an introduction and conclusion if
you really want to.) Each section should include approximately TWO graphs (a set of faceted plots
counts as one graph) examining the trend and the residuals. Including many more graphs than
this may be penalized. In each section, include a brief justification of your modeling choices (type
of model, transformations or lack of transformations) and a verbal description of the differences
you see between men and women. Some (sensibly rounded) quantitative measures will probably be
useful, but you do not (and should not) list every single statistic you can think of.
Tips
• A safe approach would be to fit separate models for men and women for each explanatory
variable (though other approaches are possible.)
• The group and color arguments within aes() can be used to distinguish between men and
women.
• Because there’s a lot of data, you might have to play around with the plot settings to get
legible graphs. Google the help pages for the individuals geoms (e.g. geom point()) to learn
about aesthetic arguments for those functions.
• If the default axis limits don’t look nice, you can choose them with + xlim() and + ylim().
• If you see any weird results, try to explain them.
2
