Week 3

Sociology 106: Quantitative Sociological Methods

February 3, 2026

Housekeeping

Weekly assignments

  • Due on Thursdays nights (except for hw1)
  • Generally, if a couple of things not quite right → still full credit. If more than a couple → half credit
  • Each assignment is only 3% of your grade, and lowest grade over the course of the semester is dropped
  • But, multiple assignments not turned in will negatively affect your grade!

HW #1: due Tuesday, February 10, 11:59 PM – Extended deadline - any questions?

HW #2: due Thursday, February 12, 11:59 PM

Learning goals

  1. Identify and distinguish between different variable types (ratio, interval, ordinal, nominal)
  2. Create and customize appropriate figures for different data types using ggplot2
  3. Evaluate which visualization methods best communicates specific data patterns
  4. Start thinking about where to get data for final paper

Agenda

  • Finish lab from last week
  • Getting data to use in future weekly assignments and your final paper
  • Visualizing data using ggplot
  • Weekly assignments example
  • Begin paper proposal discussion

Finish Lab #1

Let’s jump back in to Lab #1:

  • Open soc106/labs/lab1.qmd
  • Decide if you want to use Source or Visual to interact with the script

What type of data?

We want data that will help us say something about sociological theories / social processes

This encompasses a wide variety of data!

  • Our observations can be individuals, groups/organizations, countries
  • We will want to focus on topics that sociologists care about
  • But, fortunately for you, we care about almost everything!

Not sure if your data source/topic is appropriate?

Feel free to ask!

Getting data: online repositories

Online repositories contain multiple datasets, and you can generally search for datasets related to your topic of interest

Features:

  • Some repositories need a UC Berkeley login
  • Some repositories allow you to do some data analysis on the webpage without accessing the raw data
  • For this class: we will want to download the raw data and work with it in R (importing it from .csv or from another data type if necessary)

Data repositories

Repository Focus URL
SDA Berkeley Opinion data (GSS, ANES, racial attitudes) sda.berkeley.edu
ICPSR Comprehensive social science icpsr.umich.edu
ARDA Religion, health, public opinion thearda.com
IPUMS Census, CPS, health, time use ipums.org
Roper Center Public opinion polls ropercenter.cornell.edu
CEIC Economic data ceicdata.com
Google Dataset Search General search tool datasetsearch.research.google.com

Government data

Level Focus URL
Federal Health, education, climate, agriculture data.gov
California State demographics, economy, environment data.ca.gov
San Francisco City services, transportation, housing datasf.org
Berkeley Local permits, public works, demographics data.cityofberkeley.info

Questions?

Visualizing data

Most important first step in analyzing data is to take a close look at the data itself

An easy way to take a close look at the data is to visualize it with a figure:

  • How variables are distributed—that is, what values our variables have across all observations
  • How the values of variables change over time
  • How the values of one variable are related to the values of another variable

Review: levels of measurement

Four common levels of measurement we use in data analysis.

Level Key Property Examples Type
Ratio True zero point Age, income, years of education Continuous
Interval Equal distances, no true zero IQ, SAT scores, SES Continuous
Ordinal Ordered categories Letter grades, liberal-conservative Categorical
Nominal Unordered categories Gender, religion, political party Categorical

Why this matters: How we visualize a variable depends on its level of measurement

Basics of a ggplot plot

ggplot follows the grammar of graphics. Every plot builds up from three components: data, mapping, and layers. Let’s look at an example:

Start with your dataframe and ggplot(). This creates an empty canvas.

ggplot(data = mpg)

Add aes() to connect columns to visual properties (x, y, color). Now we have axes.

ggplot(mpg, aes(x = cty, y = hwy))

Add a geom_point() to display the data as points. Use + to add layers.

ggplot(mpg, aes(cty, hwy)) +
  geom_point()

Add a geom_smooth() to run a regression line through the points

ggplot(mpg, aes(cty, hwy)) +
  geom_point() +
  geom_smooth(formula = y ~ x, method = "lm")

Visualizing variables

Which data visualization should you use with which combinations of variables?

# of Variables Variable Type Visualization
1 Categorical Bar chart
1 Continuous Histogram, Density plot, Box plot
2 Continuous by Continuous Line chart (over time)
2 Continuous by Continuous Scatterplot
2 Continuous by Categorical Boxplot

Bar charts: geom_bar()

For categorical (nominal or ordinal) variables, we can visualize their distribution in a dataset using a bar chart.

  • A graph of a frequency table
  • Height of bars indicates the relative frequency of values in our dataset
  • X axis: a series of categories
  • Y axis: absolute frequency of values
  • We can do this using the geom_bar() function

Frequency of marital status in attain:

ggplot(data = attain, aes(x = marital)) +
  geom_bar(fill = "#0d467cff") +
  labs(
    title = "Marital Status Distribution",
    x = "Marital Status",
    y = "Count"
  ) + 
  theme_minimal()

Histogram: geom_histogram()

For continuous (interval or ratio) variables, we can visualize their distribution in a dataset using a histogram.

  • A graph in which a continuous variable is divided into categorical values
  • Frequencies shown in a bar graph
  • We can do this using the geom_histogram() function

R will automatically ‘bin’ a continuous variable into groups and then plot their frequency.

It defaults to 30.

ggplot(data = attain, aes(x = age)) +
  geom_histogram(fill = "#0d467cff") +
  labs(
    title = "Age Distribution",
    x = "Age",
    y = "Frequency"
  ) + 
  theme_minimal()

You can override geom_histogram()’s default by specifying the number of bins(). Here we set it to 100.

ggplot(data = attain, aes(x = age)) +
  # set bins to 100
  geom_histogram(bins = 100, fill = "#7B2C2F") +
  labs(
    title = "Age Distribution",
    x = "Age",
    y = "Frequency"
  ) + 
  theme_minimal()

Alternatively, you can make the bins wider to include more observations. Here, we set the binwidth = 10, which makes the bins span 10 units.

ggplot(data = attain, aes(x = age)) +
  # set binwidth to 10
  geom_histogram(binwidth = 10, fill = "#4CBB17") +
  labs(
    title = "Age Distribution",
    x = "Age",
    y = "Frequency"
  ) + 
  theme_minimal()

Line chart: geom_line()

A line chart is similar to a bar graph but the tops of the bars are represented by points joined by lines.

  • X and Y variables are ordered in some way (ordinal, interval, ratio)
  • Most useful to show change in a Y variable over your X variable
  • May not work well with missing values
  • We can do this using the geom_line() function

Here’s a look at unemployment rates over time. Note that if we have data where we have average values of Y for ordered values of X, we can also use line charts.

economics |> 
ggplot(aes(x = date, y = unemploy)) + 
geom_line() + 
  labs(
    title = "Employment over time",
    x = "Date",
    y = "Unemployment count"
  ) + 
  theme_minimal()

Scatterplot: geom_point()

A scatterplot is a two-dimensional rectangular plot for visualizing the relationship between two continuous variables.

  • X-axis: values of one continuous variable
  • Y-axis: values of another continuous variable
  • We can do this using the geom_point() function

Let’s look at the relationship between the age at which an individual was married and the age when their child was born.

attain |> 
ggplot(aes(x = agewed, y = agekdbrn)) + 
geom_point() + 
  labs(
    title = "Age of marriage vs childbirth",
    x = "Age married",
    y = "Age child was born"
  ) + 
  theme_minimal()

Boxplot: geom_boxplot()

The boxplot compactly displays the distribution of a continuous variable, usually by a categorical one. It visualises five summary statistics (the median, 25th and 75th percentile form the box), and all “outlying” points individually.

Most useful to see the entire distribution of a variable, by another categorical one

  • Most useful looking at 2 vars
  • We can do this using the geom_boxplot() function

In this case, we are looking at the distribution of income by sex.

attain |> 
ggplot(aes(x = sex, y = rincom91)) + 
geom_boxplot() + 
  labs(
    title = "Income by sex",
    x = "Sex",
    y = "Income"
  ) + 
  theme_minimal()

Questions?

Weekly Assignment Example

You’ll be provided a hw#.qmd each week that you’ll use as a template to complete your homework assignment. It will be available in bCourses under the “Assignments” folder for that week

Some weeks you’ll just answer specific questions, but in weeks that require you to use your own data, you’ll approach the assignment like a research paper, which you can use to build into the final paper.

Let’s look at an example to better understand what I’m asking.

Weekly Assignment #2

Assignment:

Using a dataset of your choice, choose a research question and construct a plot that helps answer it.

HW #2: due Thursday, February 12, 11:59 PM

  • any questions?

Paper Proposal (5%)

A two-page double-spaced proposal for your final paper is due on bCourses by Thursday, February 26 at 11:59 PM. Here’s an example.

Your proposal should include:

  1. Research question - What are you trying to answer?
  2. Why it matters - Why should we care about this question?
  3. Hypotheses - What do you think the answer is, and why?
  4. Data source - What dataset will you use?
  5. Key variables - Identify your independent and dependent variables

Note: You do not need to discuss statistical techniques at this point.