View(), glimpse(), and summary().Check the online lecture before the workshop. Weekly problem sets begin in Week 3 and are completed after the relevant workshop.
The lecture distinguishes between describing the data we have and making claims about a wider population.
Today we focus on the descriptive R step: understand the sample before making population claims.
For each analysis, use the same basic workflow:
Running an analysis is useful; understanding what the output means is the point. If you don’t understand the output, please ask!
Work from inside the RStudio Project folder you created last week.
.Rproj file for this module.week_02_workshop.R.week_02_workshop.R into your RStudio Project folder.week_02_workshop.R from inside RStudio.Do this before Part A. For learning R, run one line at a time and inspect the console after each line.
Some R functions are available as soon as R starts. Other functions come from packages.
Think of packages like apps on a phone: they add extra tools to base R.
For this workshop, we use tidyverse because it gives us the functions we need to load, inspect, clean, and summarise data.
You should have installed tidyverse last week. Only run this line if you have not installed it yet:
install.packages("tidyverse")
Load tidyverse at the start of each script:
library(tidyverse)
Installation only needs to happen once. library(tidyverse) is needed each time you start a new R session.
Before we write formulas or code, we need the idea.
The maths and R code are just precise ways to express this idea.
Later workshops reuse this same logic: build a calculation from named quantities, then compare it with what R reports.
Functions such as mean() do the arithmetic for us, but it is useful to see the pieces once.
Formula
For scores 6, 8, and 10:
\[ \bar{x} = \frac{6 + 8 + 10}{3} = 8 \]
Total divided by number of scores.
R code
scores <- c(6, 8, 10) sum(scores) # 24 length(scores) # 3 manual_mean <- sum(scores) / length(scores) manual_mean # 8 mean(scores) # 8
The formula and the R code do the same calculation.
SD starts by measuring how far each score is from the mean.
Formula
For scores 6, 8, and 10:
\[ \bar{x} = 8 \] \[ x_i - \bar{x} = -2, 0, 2 \] \[ (x_i - \bar{x})^2 = 4, 0, 4 \]
We square the distances so negative and positive distances do not cancel out.
R code
scores <- c(6, 8, 10) mean_score <- mean(scores) mean_score # 8 deviations <- scores - mean_score deviations # -2 0 2 squared_deviations <- deviations^2 squared_deviations # 4 0 4
The sample SD divides by n - 1, then takes the square root.
Formula
\[ s = \sqrt{\frac{\sum(x_i - \bar{x})^2}{n - 1}} \] \[ s = \sqrt{\frac{4 + 0 + 4}{3 - 1}} \] \[ s = \sqrt{4} = 2 \]
R code
sum_of_squares <- sum(squared_deviations) sum_of_squares # 8 variance <- sum_of_squares / (length(scores) - 1) variance # 4 manual_sd <- sqrt(variance) manual_sd # 2 sd(scores) # 2
Now open week_02_workshop.R and complete Part A. Run one line at a time and check how the numbers change in the console.
Part B uses the Blomkvist data. Before starting Part B in week_02_workshop.R:
workshop_data.zip from the Data sets folder on NOW.workshop_data.zip into your RStudio Project folder.workshop_data.
The script uses paths such as workshop_data/analysis_blomkvist.csv, so the unzipped workshop_data folder needs to be directly inside the RStudio Project folder.
blomkvist <- read_csv("workshop_data/analysis_blomkvist.csv")
If read_csv() gives an error, first check that:
.Rproj file from last week.workshop_data.zip.workshop_data.workshop_data folder is directly inside the RStudio Project folder.analysis_blomkvist.csv.Other file formats exist, but today we practise CSV files. RStudio’s Import Dataset menu can help discover import code for unfamiliar files.
Follow along in Part B of week_02_workshop.R while I show the same steps on the slides.
A data frame is the usual shape for data analysis in R: rows are observations, columns are variables, and a cell is one value for one observation on one variable.
blomkvist <- read_csv("workshop_data/analysis_blomkvist.csv")
Each row is one participant. Key columns are:
id: participant IDage: age in yearsmedicine: medication scoresmoker: smoking categorymeds_group: medication groupage_group: age categoryrt: hand reaction timert_foot: foot reaction timeFor today, the main question is: does average rt differ across age_group?
glimpse() shows variable types next to each column name.
<dbl>: numeric values, such as age or rt.<chr>: text values, such as labels or IDs.<lgl>: logical values, TRUE or FALSE.<chr> and define groups.For summaries today, rt needs to be numeric and age_group needs to define groups.
Before summarising, check what the data frame contains.
blomkvist View(blomkvist) glimpse(blomkvist) summary(blomkvist$rt)
View() opens a spreadsheet-like view in RStudio.glimpse() shows variable names, types, and preview values.summary() gives a quick numeric summary of one variable.What does one row represent, which variables are numeric, and which variables could define groups?
count(blomkvist, age_group) count(blomkvist, age_group, meds_group)
count() is descriptive: it checks how many observations are in each category before we summarise or test them.
drop_na() keeps rows that have data for the variables named.
blomkvist_rt <- drop_na(blomkvist, rt, age_group, meds_group)
This gives one row for the whole data frame.
summarise(blomkvist_rt, mean_rt = mean(rt), sd_rt = sd(rt), n = n() )
Mean and SD are the main focus today. Minimum and maximum help us see the range of values.
summarise(blomkvist_rt, mean_rt = mean(rt), sd_rt = sd(rt), min_rt = min(rt), max_rt = max(rt), n = n() )
What do the minimum and maximum reaction times tell us about the range of values?
Add .by = age_group to calculate the same summary separately for each age group.
summarise(blomkvist_rt, mean_rt = mean(rt), sd_rt = sd(rt), min_rt = min(rt), max_rt = max(rt), n = n(), .by = age_group )
Which age group has the slower average reaction time?
Use .by = c(age_group, meds_group) when the summary should be split by two grouping variables.
summarise(blomkvist_rt, mean_rt = mean(rt), sd_rt = sd(rt), n = n(), .by = c(age_group, meds_group) )
Do the same age-group pattern and medication-group pattern appear together?
Descriptive tables are still R objects. To extract values, first save a small table.
summary_age <- summarise(blomkvist_rt, mean_rt = mean(rt), sd_rt = sd(rt), n = n(), .by = age_group ) summary_age[1, "mean_rt"] summary_age$mean_rt summary_age$mean_rt[1]
[] extracts by row and column: [row, column].$ extracts one named column as a vector.$ and [1] extracts the first value from that column.We use the same extraction idea later for model objects, where the output is harder to read.
Complete the remaining exercises in week_02_workshop.R.
Part C is independent practice with blanks. It uses analysis_chinese_ldt.csv to answer concrete questions about reaction time, list_group, and accuracy_group.
Optional: Part D is independent write-your-own-code practice. Use the same logic, but decide and write the full code block yourself.
Only open week_02_stretch.R if the core workshop script felt comfortable. There is no expectation that you complete the stretch script, and it is not required preparation for the exam.
Recommended reading: Andrews (2021), Chapter 5, Exploratory Data Analysis.
Optional background if you need it: Chapter 3, Data Wrangling, and Chapter 4, Data Visualization.
| Function | What it does |
|---|---|
read_csv() |
reads a CSV file into R |
View(), glimpse() |
shows variables, types, and preview values |
count() |
counts rows in categories |
drop_na() |
removes rows with missing values in named variables |
summarise() |
calculates descriptive statistics |
n() |
counts rows inside summarise() |
mean() |
calculates an average |
sd() |
calculates the sample standard deviation |
[] |
extracts values by position, row, or column |
$ |
extracts one named column from a data frame |
Andrews, M. (2021). Doing data science in R: An Introduction for Social Scientists. SAGE Publications Ltd.