You've already learned a lot!

• Objects
• Classes
• Indexing 1-d vectors and 2-d data.frames
• Logical operators & functions
• How to look at the help documentation
• Installing/loading packages
• Making new script files
• Importing your data etc.

This time

• Lots of practice sets; letting it all come together

• Some random tidbits to make your life easier

MIDUS

For this portion, we will use the midus.csv dataset.

• Please download this file and save it somewhere you can easily find it.

You should do the exercises in this tutorial locally (on your own computer).

MIDUS

About the MIDUS dataset

Project Description: "The purpose of the study was to investigate the role of behavioral, psychological, and social factors in understanding age-related differences in physical and mental health." Variables available in this data file:

• Demographic variables: age, sex
• Physical health variables: self-rated physical health, heart problems, father had heart attack, BMI
• Mental health variables: self-rated meantal health, self-esteem, life satisfaction (life overall, work, health, relationship with spouse/partner, relationship with children), hostility (stress reactivity & agression)

Exercise 1

1. Open a new .R script file

2. Load the following packages using code in your new .R file:

   • psych
   • dplyr
   • ggplot2

3. Get your working directory

4. If you haven't already, download midus.csv from above. Drag and drop the file into a directory that you want to continue using for this class.

5. Import midus.csv into R. You can keep the name midus.

6. Make sure that the import code ends up in your SCRIPT (.R) file.

7. Save your new .R script file in the same working directory as midus.csv. You can name the script whatever you want.

Exercise 1...continued!

1. Check the classes of your variables. Below is a list of each variable and the class they should be. Change the class of any that don't match what you see.

   • ID should be a factor
   • sex should be a factor
   • age should be an integer
   • BMI should be numeric
   • physical_health_self & mental_health_self should be integers
   • self_esteem, life_satisfaction, and hostility should all be numeric
   • heart_self and heart_father should be factors

2. Save your script again. What happens to the text on the tab in RStudio?

Cleaning your global environment

Sometimes, your Global Environment (top right) can fill up with stuff that you don't need anymore. You can clean this!

A) Delete EVERYTHING using the broom icon (next to the Import Dataset tab)

B) Switch to GRID view, check boxes of individual objects you DON'T want, then press the broom icon. Be sure to switch back to LIST view when finished

Exercise 1...continued again!

Now...

Clear your global environment, select your entire script file, and RUN it!

...what happens? what does this mean?

Indexing...continued!

Indexing a vector (1 dimension):

• midus$age[1]
• The number 1 gives us the 1st item in the age vector that is part of the midus data.frame

Indexing a data.frame (2 dimensions):

• midus[1, 2:3]
• This gives us the 1st row and the 2nd thru 3rd columns of the midus data.frame

Logical Operator Review

Returns a value of TRUE or FALSE

A = 3 - "A is an objec that stores the number 3."

A == 3 - "Is A equal to 3?"

Multiple Logical Statements

You can combine logical operators to simultaneously see if multiple conditions have been met.

Multiple Logical Statements

Is this true? AND Is this true?

(10 < 100 & 24 == 23 + 1)

## [1] TRUE

(5 > 4 & 5 > 10)

## [1] FALSE

Is this true? OR Is this true?

(5 > 4 | 5 > 10)

## [1] TRUE

# Could also be something like:
# (A == B | C != D)

Subset your data

Sometimes, you will want to perform functions on only some of your data points.

You can subset your data to identify subjects in a certain subgroup.

Use the subset() function:

subset(x = empire, subset = (mass > 100))

## # A tibble: 2 x 6
##   name        height  mass sex   homeworld species
##   <chr>        <int> <dbl> <chr> <chr>     <fct>  
## 1 Darth Vader    202   136 male  Tatooine  Human  
## 2 Chewbacca      228   112 male  Kashyyyk  Wookiee

What if I don't remember the arguments?

Check the R documentation! ?subset

When in doubt, you can always search the "Help" tab or the internet

Functions within functions

Often, you can wrap functions within functions. This is called nesting.

Example:

round(x = sqrt(86), digits = 1)

## [1] 9.3

How does it do this? Inner to outer!

sqrt(86) --> 9.2736185 --> round(x = 9.2736185, digits = 1) --> 9.3

A million ways to do 1 thing

Add a varible to a data.frame

midus[,9] <- rep(x = "Wave2", times = 40)
midus$wave <- rep(x = "Wave2", times = 40)
midus["wave"] <- rep(x = "Wave2", times = 40)

Remove a variable from a data.frame

midus <- midus[,-9]
midus <- subset(midus, select = -wave)

Removing rows from a data.frame

midus <- midus[-c(1:10),]
midus <- subset(x = midus, subset = ID != "10175")

Exercise 2

Use nested functions to create a subset of the midus data.frame which contains only participants who have self_esteem scores above the mean.

Hints:

• Name this new data.frame something different from midus
• Use only complete cases (no missing) to get the mean. Where can you look to find the arguments of the mean function?

A note on missing values

• There are MANY ways to deal with missing data

• Decisions regarding missing data are best made by you and your colleagues/advisors since it will depend on your research question

• For now, we're going to use listwise deletion aka using complete cases

• To overwrite the midus data.frame so that it only shows participants that have a data point for every variable, we can use the following line of code:

  • midus <- na.omit(midus)

Dimensions

Sometimes it's useful to better understand the dimensions of your data.frame. Comes up in regards to error messages! For more, details on this, see the "Examples of a few functions" section at the end of the 5: Functions practice set.

dim(empire) # for both dimensions

## [1] 10  6

nrow(empire) # for rows

## [1] 10

ncol(empire) # for columns

## [1] 6

Saving data

So you've manipulated and cleaned your dataset, and now you want to save it...

write.csv(x = data.frame,
          file = "newFile.csv",
          row.names = FALSE)

Exploring data that is stored within a list obect (i.e., regression, ANOVA, t-test results) is really hard.

In these cases, I suggest pulling out the pieces of information that you actually care about, making it your own data.frame, and then writing out that data.frame. There are other ways, though (you'll need to Google it)!

Recoding variables

• Check out the recode() function from the dplyr package
• If there is something systematic you want to add or remove, you can use gsub()
  • EX: What if you wanted to replace all the underscores (_) with periods (.) in my column names?

colnames(midus) <- gsub(pattern = "_",
                        replacement = ".",
                        x = colnames(midus))

• EX: What if I wanted to remove everything after the underscore(_)?

colnames(midus) <- gsub(pattern = "_",
                        replacement = "",
                        x = colnames(midus))

Naming Conventions

Habits

Get into good coding habits now, so you don't hate yourself later!

Pro tip:

• Write code that you can understand now
• Write code that you can understand 6 months from now
• Write code that someone else can understand 6 months from now

What does this mean?

• COMMENT YOUR CODE
• Use consistent naming conventions
• Use descriptive object names (who cares if they're long when you have tab-complete)
• Organize your files
• Think computer & human readable
• Cloud-based storage
• Version control

Answers for Exercise 1

Your R script should look something like this:

library(psych)
library(dplyr)
library(ggplot2)
getwd()
# set your working directory to the folder you need...if not there already
# setwd("folder/with/midus.csv")
midus <- read.csv("folder/midus.csv")
# change 4 character variables into factors
# see the practice for 5: Functions for more on how to do this!
midus$ID <- factor(midus$ID)
midus$sex <- factor(midus$sex)
midus$heart_self <- factor(midus$heart_self)
midus$heart_father <- factor(midus$heart_father)

Answers for Exercise 2

Your code should look something like this:

midusMini <- subset(x = midus,
                    subset = self_esteem > mean(self_esteem, na.rm = TRUE))

To look up the arguments of the mean() function, use ?mean. Many functions have an argument called na.rm or similar to it that asks if you want the function to remove missing variables. Here, we're saying, "yes, please remove missing variables before calculating the mean"