Session 3: How can I perform univariate analysis?

A Slower Introduction to R

Yea-Hung Chen, PhD, MS

UCSF Library

Friday, October 16, 2026

Arguments

The `NA` mean

from last week:

mean(nhanes$bpxodi1)
#| [1] NA
class(nhanes$bpxodi1)
#| [1] "integer"
head(nhanes$bpxodi1)
#| [1] 98 84 79 NA NA 72

The help pages

to access the help page for a function, use the help() function:

> help(sqrt)

or, equivalently, use ?:

> ?sqrt

Live demonstration

🤸

Function arguments in `sqrt()`

you can optionally type x= when you use sqrt():

sqrt(x=81)
#| [1] 9

or, you can continue to omit it:

sqrt(81)
#| [1] 9

x is known as an argument

What are function arguments?

you may think of arguments as being like options or settings
using another cooking analogy, suppose that baking is a function
- we can think of the temperature setting on the oven as an argument to the function
- what’s another possible argument to baking?

What are function arguments?

sqrt(x=81)
#| [1] 9

81 is the argument value
x is the argument name

Function arguments in `mean()`

notice the commas** and the default values

Function arguments in `mean()`

the na.rm argument is:

a logical evaluating to TRUE or FALSE indicating whether NA values should be stripped before the computation proceeds.

the Usage section (previous slide) shows that the default value for na.rm is FALSE, suggesting that NA values will not be “stripped”

The missing mean

mean(nhanes$bpxodi1,na.rm=TRUE)
#| [1] 74.97469

🎉🎉🎉

The missing mean

mean(nhanes$bpxodi1,na.rm=TRUE)
#| [1] 74.97469

notice that I’ve used a comma to separate the two arguments
TRUE must be in all caps
na.rm=TRUE does not delete missing values from your data
all it does is make mean() ignore missing values

`na.rm`

Caution

The na.rm argument does not exist in every R function!

most of the functions in the next section have an na.rm argument
but, for example, the function for linear regression, lm(), does not have an na.rm argument

When should I include argument names?

sqrt(x=81)
#| [1] 9
sqrt(81)
#| [1] 9

if there is just one argument, the argument name is optional

When should I include argument names?

mean(nhanes$bpxodi1,na.rm=TRUE)
#| [1] 74.97469

if there is more than one argument, you should include argument names if:
- you skip an argument
- or, write the arguments out of order

When should I include argument names?

sd(nhanes$bpxodi1,TRUE)
#| [1] 11.41659
sd(nhanes$bpxodi1,na.rm=TRUE)
#| [1] 11.41659

including argument names can improve the readability of your code
so, it can be good to include them even when they are not necessary

Coding style

Common suggestions

put a space before and after mathematical operators, except exponentiation (^)
put a space before and after <-
put a space after commas
put a space before and after =

12312 / 27113508 * 10^5
#| [1] 45.4091
abs(2 - 12)
#| [1] 10
popsize <- 27113508
mean(nhanes$bpxodi1, na.rm = TRUE)
#| [1] 74.97469

Styling code in RStudio

Highlight the code.
Use the keyboard shortcut: ⌘⌘ shiftshift AA on Mac or ctrlctrl shiftshift AA on Windows.
- Equivalently, select Code > Reformat Selection.

Do I have to follow the common suggestions suggestions?

no, the suggestions are not rules
for example, R code works fine without extra spaces, as you have already seen
but, the suggestions can improve readability

Exercise 1

Open the help page for the quantile() function, which you’ll use in Exercise 2 to find percentiles of continuous variables.
What is the name of the second argument?
Does the function support the na.rm argument?

`c()`

Example: cancer prevalence

from Session 1:

popsize <- 27113508
12312 / popsize * 10^5
#| [1] 45.4091
9120 / popsize * 10^5
#| [1] 33.63637

the math occurs in two lines of code
we could, alternatively, use a vector of the numerators and do the math in one line of code…

`c()`

c(12312, 9120)
#| [1] 12312  9120

the c() function combines values in a vector
in other words:
- the arguments are values of the desired vector
- the output** is the desired vector
in the function, use commas to separate values

`c()`

another example:

c(33, 52, 48)
#| [1] 33 52 48

you can also use c() to create a character vector:

c('California', 'New York', 'Texas')
#| [1] "California" "New York"   "Texas"

Example: cancer prevalence

using the numerators one by one:

12312 / popsize * 10^5
#| [1] 45.4091
9120 / popsize * 10^5
#| [1] 33.63637

using a vector of the numerators:

c(12312, 9120) / popsize * 10^5
#| [1] 45.40910 33.63637

Example: cancer prevalence

alternatively, first using object assignment:

numerators <- c(12312, 9120)
numerators / popsize * 10^5
#| [1] 45.40910 33.63637

`c()`

c() is very useful
it is one of those hard-to-avoid fundamentals I referred to in the orientation!
you will see an example in the next section in which c() is an argument to a function

Summarizing continuous variables

What is a continuous variable?

continuous variable = a variable that has numeric values
examples:
- age
- height
- systolic blood pressure

Means, standard deviations, and variances

mean(nhanes$bpxodi1, na.rm = TRUE)
#| [1] 74.97469
sd(nhanes$bpxodi1, na.rm = TRUE)
#| [1] 11.41659
var(nhanes$bpxodi1, na.rm = TRUE)
#| [1] 130.3386

Percentiles

median(nhanes$bpxodi1, na.rm = TRUE)
#| [1] 74
quantile(nhanes$bpxodi1, probs = 0.5, na.rm = TRUE)
#| 50% 
#|  74
quantile(nhanes$bpxodi1, 0.5, na.rm = TRUE)
#| 50% 
#|  74

Percentiles

quantile(nhanes$bpxodi1, probs = 0.25, na.rm = TRUE)
#| 25% 
#|  67
quantile(nhanes$bpxodi1, probs = 0.75, na.rm = TRUE)
#| 75% 
#|  82

Percentiles

quantile(nhanes$bpxodi1,
         probs = c(0.25, 0.5, 0.75),
         na.rm = TRUE)
#| 25% 50% 75% 
#|  67  74  82

for the probs argument, I’ve used the c() function to create a vector containing 3 values

Percentiles

quantile(nhanes$bpxodi1,
         probs = c(0.25, 0.5, 0.75),
         na.rm = TRUE)
#| 25% 50% 75% 
#|  67  74  82

why do we need to use c() for probs, when we want to use more than one probs value?

Minimum and maximum values

min(nhanes$bpxodi1, na.rm = TRUE)
#| [1] 36
max(nhanes$bpxodi1, na.rm = TRUE)
#| [1] 142
range(nhanes$bpxodi1, na.rm = TRUE)
#| [1]  36 142

`summary()`

summary(nhanes$bpxodi1)
#|    Min. 1st Qu.  Median    Mean 3rd Qu.    Max.     NAs 
#|   36.00   67.00   74.00   74.97   82.00  142.00    2030

Histograms

hist(nhanes$bpxodi1)

Histogram titles and labels

main: the title at the top
xlab: the label on the x axis

Histograms titles and labels

hist(nhanes$bpxodi1, main = '', xlab = 'Diastolic blood pressure (mmHg)')

Histogram bars

to set the number of bars in the histogram, use the breaks argument

hist(nhanes$bpxodi1,
     main = '',
     xlab = 'Diastolic blood pressure (mmHg)',
     breaks = 50)

Histograms bars

hist(nhanes$bpxodi1,
     main = '',
     xlab = 'Diastolic blood pressure (mmHg)',
     breaks = 50)

Quotes in function arguments

Tip

If a function argument needs to be in quotes, you can use single quotes or double quotes.

hist(nhanes$bpxodi1,
     main = '',
     xlab = "Diastolic blood pressure (mmHg)",
     breaks = 50)

you must end the quote with the type you started with

Exercise 2

Across the exercises today, you’ll build an R script that will import the NHANES data and perform some univariate analysis. You’ll also practice going back and forth between the console and your script.

At the top of a new script, include code for importing the data. Hopefully, you already copied and pasted code for doing this from last week. If you didn’t, re-obtain the code by using RStudio’s import tool. Add a comment above that code. Run that code.

Exercise 2

In the console or the script, write code for displaying a histogram of fasting glucose (lbxglu). Now do this again, specifying that you want to use 60 bins. Feel free to try other numbers of bins.
A common threshold for a health level of fasting glucose is 100 mg/dL. In the console or the script, add a red vertical line to your histogram at that threshold: abline(v=100,col='red').

Exercise 2

In the script, type code for finding the mean level of fasting glucose. Add a comment above that code. Run that code.
In the script, type code for finding the 25th, 50th, and 75th percentiles of fasting glucose. Add a comment above that code. Run that code.

Confidence intervals for population means

Confidence interval for mean diastolic blood pressure

what is a 95% confidence interval for the mean diastolic blood pressure in the US adult population?

t.test(nhanes$bpxodi1)
#| 
#|  One Sample t-test
#| 
#| data:  nhanes$bpxodi1
#| t = 513.88, df = 6122, p-value < 2.2e-16
#| alternative hypothesis: true mean is not equal to 0
#| 95 percent confidence interval:
#|  74.68867 75.26070
#| sample estimates:
#| mean of x 
#|  74.97469

Function values

t.test() prints a summary report
it is possible to extract specific values

Function values

Function values

diastolic_ci <- t.test(nhanes$bpxodi1)
names(diastolic_ci)
#|  [1] "statistic"   "parameter"   "p.value"     "conf.int"    "estimate"   
#|  [6] "null.value"  "stderr"      "alternative" "method"      "data.name"

Extracting function values

diastolic_ci$estimate
#| mean of x 
#|  74.97469
diastolic_ci$conf.int
#| [1] 74.68867 75.26070
#| attr(,"conf.level")
#| [1] 0.95

Summarizing categorical variables

What is a categorical variable?

categorical variable = a variable that has groups
examples:
- age group
- level of educational attainment
- exposure (yes or no)
- disease (yes or no)
variables with two groups are known as dichotomous variables

Hypertension

class(nhanes$hypertension)
#| [1] "character"
head(nhanes$hypertension)
#| [1] "Stage 2" "Stage 1" "No"      NA        NA        "No"

Counts

table(nhanes$hypertension)
#| 
#| Elevated       No  Stage 1  Stage 2 
#|      841     2657     1409     1216

What about missing values?

what about the missing values?

The help page for `table()`

The help page for `table()`

the useNA argument can be used to control:

whether to include NA values in the table

Counts

table(nhanes$hypertension, useNA = 'always')
#| 
#| Elevated       No  Stage 1  Stage 2     <NA> 
#|      841     2657     1409     1216     2030
table(nhanes$hypertension, useNA = 'ifany')
#| 
#| Elevated       No  Stage 1  Stage 2     <NA> 
#|      841     2657     1409     1216     2030

Percents

prop.table(table(nhanes$hypertension, useNA = 'always'))
#| 
#|  Elevated        No   Stage 1   Stage 2      <NA> 
#| 0.1031522 0.3258923 0.1728198 0.1491476 0.2489881

notice that you must put table() inside of prop.table()

Percents

without the NA included:

prop.table(table(nhanes$hypertension))
#| 
#|  Elevated        No   Stage 1   Stage 2 
#| 0.1373510 0.4339376 0.2301160 0.1985955

Percents

on a percent scale:

prop.table(table(nhanes$hypertension, useNA = 'always')) * 100
#| 
#| Elevated       No  Stage 1  Stage 2     <NA> 
#| 10.31522 32.58923 17.28198 14.91476 24.89881

Percents

prop.table(table(nhanes$hypertension)) * 100
#| 
#| Elevated       No  Stage 1  Stage 2 
#| 13.73510 43.39376 23.01160 19.85955

Exercise 3

In the script from Exercise 2, write code for obtaining the percent of adults who have a history of smoking (smoking). Add a comment above that code. Run that code.
In the script, write code for obtaining the percent of adults who have a history of asthma (asthma). Add a comment above that code. Run that code.

Exercise 3

Select all of the text and code in your script. You can optionally use the keyboard shortcut: ⌘ A on Mac or ctrl A on Windows. Run the code, and verify that your computer reproduced all of the work.

Session 3: How can I perform univariate analysis?

Arguments

The NA mean

The help pages

Live demonstration

Function arguments in sqrt()

What are function arguments?

What are function arguments?

Function arguments in mean()

Function arguments in mean()

The missing mean

The missing mean

na.rm

When should I include argument names?

When should I include argument names?

When should I include argument names?

Coding style

Common suggestions

Styling code in RStudio

Do I have to follow the common suggestions suggestions?

Exercise 1

Exercise 1

c()

Example: cancer prevalence

c()

c()

Example: cancer prevalence

Example: cancer prevalence

c()

Summarizing continuous variables

What is a continuous variable?

Means, standard deviations, and variances

Percentiles

Percentiles

Percentiles

Percentiles

Minimum and maximum values

summary()

Histograms

Histogram titles and labels

Histograms titles and labels

Histogram bars

Histograms bars

Quotes in function arguments

Exercise 2

Exercise 2

Exercise 2

Exercise 2

Confidence intervals for population means

Confidence interval for mean diastolic blood pressure

Function values

Function values

Function values

Extracting function values

Summarizing categorical variables

What is a categorical variable?

Hypertension

Counts

What about missing values?

The help page for table()

The help page for table()

Counts

Percents

Percents

Percents

Percents

Exercise 3

Exercise 3

Exercise 3

The `NA` mean

Function arguments in `sqrt()`

Function arguments in `mean()`

Function arguments in `mean()`

`na.rm`

`c()`

`c()`

`c()`

`c()`

`summary()`

The help page for `table()`

The help page for `table()`