stevedata: Steve’s Toy Data for Teaching About a Variety of Methodological, Social, and Political Topics

My stevedata  hexlogo

stevedata is an R package full of toy data sets that you may find useful for various purposes. Namely, I’ve created probably over a hundred toy data sets along the way, either to riff on some topic on my blog, show my students something in one of my many classes, or just to entertain myself for spamming Twitter with my assorted thoughts. I had stuffed a lot of these into stevemisc, but I want to keep that package mostly about the functions (and whatever data are useful for showing off the functions). stevedata will have all my toy data going forward.

I anticipate two sets of R users may find these data useful. First, instructors may find these data useful for classes on a variety of topics, but prominently quantitative methods and international relations. Many of the toy data sets included in this R package are data I’ve acquired or assembled to teach about topics in quantitative methods or international relations in a reproducible way. Users should see my Github repositories for my classes on introduction to international relations, quantitative methods in political science, and foundations of social science research for public policy to see how I’ve used these data (or development versions of them). Topics here are diverse, including (but not limited to) carbon dioxide emissions over 800,000 years (as an illustration of climate change), coffee prices (as an illustration of the worsening terms of trade), the justifiability of bribe-taking (as an illustration of information-poor and discrete variables that a researcher may be tempted to treat as drawn from a normal distribution), the canonical case of illiteracy rates in the 1930 U.S. Census (as an illustration of an ecological fallacy), and many, many more topics.

Second, my students in these classes (but especially my methods classes) should find this R package useful. I will also be having my methods students (undergraduate and graduate) download this package to work through problem sets in the R programming language. It’d be a benefit to them (and less hassle/headache for myself) to have my students download this package from CRAN rather than work through potential curl issues by installing through Github.

In almost all instances, each data set has an underlying code/script that generates them. These are in a data-raw directory that is not included in the Github repository or the R package. However, I invite users to reach out with questions about the data if they have them.


The goal is to get this on CRAN, eventually. Until then, you can install the development version of stevedata from Github via the devtools package. I suppose using the remotes package would work as well.



The data set already has a lot to offer those who might be curious about its contents. You can do this to see what is in it.

data(package = "stevedata")

The ensuing output will look like this.

Object Name Title/Description
Arca NYSE Arca Steel Index data, 2017–present
CFT15 Randomization Inference in the Regression Discontinuity Design: An Application to Party Advantages in the U.S. Senate
CP77 Education Expenditure Data (Chatterjee and Price, 1977)
DJIA Dow Jones Industrial Average, 1885-Present
DST Casualties/Fatalities in the U.S. for Drunk-Driving, Suicide, and Terrorism
Datasaurus The Datasaurus Dozen
Dee04 Are There Civics Returns to Education?
ESS9GB British Attitudes Toward Immigration (2018-19)
ESSBE5 Trust in the Police in Belgium (European Social Survey, Round 5)
Guber99 School Expenditures and Test Scores for 50 States, 1994-95
LOTI Land-Ocean Temperature Index, 1880-2020
LTPT Long-Term Price Trends for Computers, TVs, and Related Items
LTWT “Let Them Watch TV”
Newhouse77 Medical-Care Expenditure: A Cross-National Survey (Newhouse, 1977)
ODGI Ozone Depleting Gas Index Data, 1992-2019
Presidents U.S. Presidents and Their Terms in Office
TV16 The Individual Correlates of the Trump Vote in 2016
aluminum_premiums LME Aluminum Premiums Data
anes_prochoice Abortion Attitudes (ANES, 2012)
arcticseaice Arctic Sea Ice Extent Data, 1901-2015
arg_tariff Simple Mean Tariff Rate for Argentina
asn_stats Aviation Safety Network Statistics, 1942-2019
clemson_temps Daily Clemson Temperature Data
co2emissions Carbon Dioxide Emissions Data
coffee_imports Coffee Imports for Select Importing Countries
coffee_price The Primary Commodity Price for Coffee (Arabica, Robustas)
election_turnout State-Level Education and Voter Turnout in 2016
eq_passengercars Export Quality Data for Passenger Cars, 1963-2014
eustates EU Member States (Current as of 2019)
fakeLogit Fake Data for a Logistic Regression
fakeTSCS Fake Data for a Time-Series Cross-Section
fakeTSD Fake Data for a Time-Series
ghp100k Gun Homicide Rate per 100,000 People, by Country
gss_abortion Abortion Opinions in the General Social Survey
gss_spending Attitudes Toward National Spending in the General Social Survey (2018)
gss_wages The Gender Pay Gap in the General Social Survey
illiteracy30 Illiteracy in the Population 10 Years Old and Over, 1930
mm_mlda Minimum Legal Drinking Age Fatalities Data
mm_nhis Data from the 2009 National Health Interview Survey (NHIS)
mm_randhie Data from the RAND Health Insurance Experiment (HIE)
mvprod Motor Vehicle Production by Country, 1950-2019
pwt_sample Penn World Table (9.1) Macroeconomic Data for Select Countries, 1950-2017
quartets Anscombe’s (1973) Quartets
recessions United States Recessions, 1855-present
sealevels Global Average Absolute Sea Level Change, 1880–2015
so2concentrations Sulfur Dioxide Emissions, 1980-2017
steves_clothes Steve’s (Professional) Clothes, as of March 3, 2019
sugar_price IMF Primary Commodity Price Data for Sugar
therms Thermometer Ratings for Donald Trump and Barack Obama
uniondensity Cross-National Rates of Trade Union Density
usa_chn_gdp_forecasts United States-China GDP and GDP Forecasts, 1960-2050
usa_computers Percentage of U.S. Households with Computer Access, by Year
usa_states State Abbreviations, Names, and Regions/Divisions
wvs_ccodes Syncing Word Values Survey Country Codes with CoW Codes
wvs_immig Attitudes about Immigration in the World Values Survey
wvs_justifbribe Attitudes about the Justifiability of Bribe-Taking in the World Values Survey
yugo_sales Yugo Sales in the United States, 1985-1992

Here is a simple scraping job to provide more information (by way of the description field in the associated R Documentation file). I include these descriptions as a vignette as well.

Object Name Description
Arca Daily data on the NYSE Arca Steel Index. These data are useful for me in teaching how Trump’s 2018 steel tariffs didn’t do much good for the steel industry.
CFT15 This is the replication data for “Randomization Inference in the Regression Discontinuity Design: An Application to Party Advantages in the U.S. Senate”, published in 2015 in Journal of Causal Inference. I use these data to teach about regression discontinuity designs.
CP77 This is a simple data set provided by Chatterjee and Price (1977, p. 108) that serves as a known example of heteroscedasticity.
DJIA This data set contains the value of the Dow Jones Industrial Average on daily close for all available dates (to the best of my knowledge) from 1885 to the most recently concluded calendar year. Extensions shouldn’t be too difficult with existing packages.
DST These are fatalities (and, in the case of terrorism, casualties as well) for drunk-driving, suicide, and acts of terrorism in the U.S. spanning 1970 to 2018. Only one of these is sufficiently important to command public attention despite being the least severe public bad. Do you want to guess which one?
Datasaurus An illustrative exercise in never trusting the summary statistics without also visualizing them.
Dee04 This should be a data set for a (partial?) replication of Dee’s (2004) article on the purported civics returns to education. I use these data for in-class illustration about instrumental variable analyses.
ESS9GB This is a replication data originally set to accompany a blog post and presentation to students at the University of Nottingham in March 2020. However, COVID-19 led to the cancellation of the talk.
ESSBE5 This is a sample data set cobbled from the fifth round of European Social Survey data for Belgium. It offers a means to do a basic replication of some of Chapter 5 of The SAGE Handbook of Regression Analysis and Causal Inference.
Guber99 A data set for a canonical case of a Simpson’s paradox, useful for in-class instruction on the topic.
LOTI These data contain monthly mean temperature anomalies expressed as deviations from the corresponding 1951-1980 means. They are useful for showing how we can measure climate change.
LTPT These data are a monthly time-series of changes in the consumer price index relative to a Dec. 1997 starting date for televisions, computers, and related items. I use this as in-class illustration that globalization has made consumer electronics cheaper across the board for Americans.
LTWT “Let Them Watch TV”: These data contain price indices for various items for the general urban consumer. Categories include medical services, college tuition, college textbooks, child care, housing, food and beverages, all items (i.e. general CPI), new vehicles, apparel, and televisions. The base period in value was originally the 1982-4 average, but I converted the base period to January 2000. I use these data for in-class discussion about how liberalized trade has made consumer electronics (like TVs) fractions of their past prices. Yet, young adults face mounting costs for college, child-raising, and health care that government policy has failed to address.
Newhouse77 These are the data in Newhouse’s (1977) simple OLS model from 1977. In his case, he’s trying to explain medical care expenditures as a function of GDP per capita for these countries. It’s probably the easiest OLS model I can find in print because Newhouse helpfully provides all the data in one simple table.
ODGI The NOAA Earth System Research Laboratory has an “ozone depleting gas index” (ODGI) data set from 1992 to 2018. This dataset summarizes Table 1 and Table 2 from its website. The primary interest here (for my purposes) is the ODGI indices (including the new 2012 measure). The data set includes constituent greenhouse gases/chlorines as well in parts per trillion. The primary use here is for in-class illustration.
Presidents This should be self-evident. Here are all U.S. presidents who have completed their terms in office (i.e. excluding the current one).
TV16 These data come from the 2016 CCES and allow interested students to model the individual correlates of the Trump vote in 2016. Code/analysis heavily indebted to a 2017 analysis I did on my blog (see references).
aluminum_premiums A near daily data set on the price of aluminum premiums (USD/MT) for LME in the U.S., Western Europe, East Asia, and Southeast Asia. I like these data as illustrative of some of the shortsightedness of the aluminum tariffs that Donald Trump announced in March 2018. The tariffs had no discernible effect on manufacturing employment or earnings, but they created a supply shock that made aluminum more expensive.
anes_prochoice A simple data set for in-class illustration about how to estimate and interpret interactive relationships. The data here are deliberately minimal for that end.
arcticseaice This data set from Connelly et al. (2017) measures the Arctic sea ice extent in 10^6 square kilometers. It includes lower bounds and upper bounds on annual averages.
arg_tariff Simple mean tariff rate for Argentina, starting in 1980. The goal is to keep these data current.
asn_stats These are yearly counts on air accidents and fatalities, including measures for corporate jet accidents and hijackings. The hijackings are of particular interest to me, at least from a historical terrorism perspective.
clemson_temps This data set contains daily temperatures (highs) for Clemson, South Carolina from Jan. 1, 1930 to the end of the most recent calendar year. The goal is to update this periodically with new data for as long as I live in this town.
co2emissions This is a sample data set, cobbled from various sources, about carbon dioxide emissions in the history of the planet from 800,000 BCE to the most recently concluded calendar year. I use this for a data visualization example for a lecture on climate change and international politics. Data communicate yearly averages/estimates.
coffee_imports A simple time series on coffee imports for select importing countries (i.e. European Union + Japan + Russia + Tunisia + United States).
coffee_price This is primary commodity price data for coffee (Arabica, Robustas) from 1980 to the present. I manually update these data since FRED’s coverage since 2017 has been spotty.
election_turnout A simple data set on education and state-level (+ DC) turnout in the 2016 presidential election. This is inspired by what Pollock (2012) does in his book.
eq_passengercars Data from the International Monetary Fund for the export quality and unit/trade value of passenger cars for all available countries and years from 1963 to 2014.
eustates European Union membership by accession date
fakeLogit This is a simple fake data set to illustrate a logistic regression.
fakeTSCS This is a toy (i.e. “fake”) data set created by the fabricatr package. There are 100 observations for 25 hypothetical countries. The outcome y is a linear function of a baseline for each hypothetical country, plus a yearly growth trend as well as varying growth errors for each country. x1 is supposed to have a linear effect of .5 on y, all things considered. x2 is supposed to have a linear effect of 1 on y for each unit change in x2, all things considered.
fakeTSD This is a toy (i.e. “fake”) data set created by the fabricatr package. There are 100 observations. The outcome y is a linear function of 20 + (.25 * year) + .(25 * x1) + (1 * x2) + e. This clearly implies some autocorrelation in the data. I.e. it’s a time-series.
ghp100k This is the yearly rate of gun homicides per 100,000 people in the population, selecting on “Western” countries of interest.
gss_abortion This is a toy data set derived from the General Social Survey that I intend to use for several purposes. First, the battery of abortion items can serve as toy data to illustrate mixed effects modeling as equivalent to a one-parameter (Rasch) model. Second, I include some covariates to also do some basic regressions. I think abortion opinions are useful learning tools for statistical inference for college students. Third, there’s a time-series component as well for understanding how abortion attitudes have changed over time.
gss_spending This is a toy data set that collects attitudes on toward national spending for various things in the General Social Survey for 2018. I use these data for in-class illustration about ordinal variables and ordinal models.
gss_wages Wage data from the General Social Survey (1974-2018) to illustrate wage discrepancies by gender (while also considering respondent occupation, age, and education).
illiteracy30 This is perhaps the canonical data set for illustrating the ecological fallacy.
mm_mlda These are data you can use to replicate the regression discontinuity design analyses throughout Chapter 4 of Mastering ’Metrics. Original analyses come from Carpenter and Dobkin (2009, 2011).
mm_nhis These are data from the 2009 NHIS survey. People who have read Mastering ‘Metrics should recognize these data. They’re featured prominently in that book and the authors’ discussion of random assignment and experiments.
mm_randhie These are data from the RAND Health Insurance Experiment (HIE).People who have read Mastering ‘Metrics should recognize these data. They’re featured prominently in that book and the authors’ discussion of random assignment and experiments.
mvprod Data, largely from Organisation Internationale des Constructeurs d’Automobiles (OICA), on motor vehicle production in various countries (and the world totals) from 1950 to 2019 at various intervals. Tallies include production of passenger cars, light commercial vehicles, minibuses, trucks, buses and coaches.
pwt_sample These are some macroeconomic data for 21 select (rich) countries. I’ve used these data before to discuss issues of grouping and skew in cross-sectional data.
quartets These are four x-y data sets, combined into a long format, which have the same traditional statistical properties (mean, variance, correlation, regression line, etc.). However, they look quite different.
recessions Data on U.S. recessions, past to present. Data include information on contraction, expansion, and cycle.
sealevels These data describe how sea level has changed over time, in both relative and absolute terms. Absolute sea level change refers to the height of the ocean surface regardless of whether nearby land is rising or falling.
so2concentrations This data set contains yearly observations by the Environmental Protection Agency on the concentration of sulfur dioxide in parts per billion, based on 35 sites. I use this for in-class illustration. Note that the national standard is 75 parts per billion.
steves_clothes I cobbled together this data set of the professional clothes (polos, long-sleeve dress shirts, pants) in my closet, largely for illustration on the origins of apparel in the U.S. for an intro lecture on trade.
sugar_price This is primary commodity price data for sugar globally, in the United States, and in Europe for every month from 1980 to (roughly) the present. Prices are nominal U.S. cents per pound and are not seasonally adjusted (“NSA”).
therms A data set on thermometer ratings for Donald Trump and Barack Obama in 2020. I use these data for in-class illustration of central limit theorem. Basically: the sampling distribution of a population is normal, even if the underlying population is decidedly not.
uniondensity Cross-national data on relative size of the trade unions and predictors in 20 countries. This is a data set of interest to replicating Western and Jackman (1994), who themselves were addressing a debate between Wallerstein and Stephens on which of two highly correlated predictors explains trade union density.
usa_chn_gdp_forecasts This is a toy data set to examine the time in which we should expect China to overtake the United States in total gross domestic product (GDP), given current trends. It includes an OECD long-term GDP forecast from 2014, and forecasts from the forecast and prophet packages in R.
usa_computers This is a simple and regrettably incomplete time-series on the percentage of U.S. households with access to a computer, by year.
usa_states A simple data set from,, state.region, and state.division (+ District of Columbia). I’d rather just have all these in one place.
wvs_ccodes A simple data set that syncs World Values Survey country codes (s003) with corresponding country codes from the Correlates of War state system membership data.
wvs_immig A data set on attitudes about immigration for all observations in the third to sixth wave of the World Values Survey. I use these data for in-class illustration.
wvs_justifbribe A data set on attitudes about the justifiability of bribe-taking for all observations in the third to sixth wave of the World Values Survey. I use these data for in-class illustration about seemingly interval-level, but information-poor measurements.
yugo_sales A data set on Yugo sales against two competing models in the United States from 1985 to 1992.