Computational Skills for Multivariable Thinking in Introductory Statistics

Figures & data

Table 1 Computational skills for multivariable thinking in introductory courses.

Skill	Supporting tasks
Summarize relevant conditional distributions of data	Filter outliers, stratify data, calculate group means/proportions/standard deviations, and calculate other basic functions of existing variables.
Visualize multiple variables simultaneously	Create boxplots of a quantitative variable for different levels of two or more categorical variables. Create scatterplots to depict the relationships between two quantitative variables and other variables of interest.
Fit multiple regression models	Fit multiple regression models using software, obtain confidence intervals on parameters, and perform model diagnostics.

Table 2 Summary statistics for confirmed COVID-19 cases as of October 1, 2020 by region.

Region	Cases per 100,000	Population	Countries
	inhabitants	(millions)
East Asia & Pacific	40.4	2281.6	23
Europe & Central Asia	633.0	913.2	55
Latin America & Caribbean	1456.8	646.4	41
Middle East & North Africa	516.3	456.7	21
North America	2033.6	365.9	3
South Asia	392.5	1835.8	8
Sub-Saharan Africa	103.8	1106.7	48

Fig. 1 Reported confirmed cases per 100,000 inhabitants for countries with at last five million inhabitants and 50 confirmed cases per 100,000 using the tidycovid19 package (Gassen 2020). Labeled countries are in the top ten in terms of gross domestic product.

Fig. 2 Confirmed cases (cumulative) as of June 6, 2020 using the tidycovid19 package (Gassen 2020).

Fig. 3 Confirmed cases per 100,000 inhabitants, gross domestic product per capita, and COVID-19 testing rate by country on May 10, 2020. Labeled countries have one of the world’s ten largest economies.

Fig. 4 Mosaic plot depicting participants by political party and whether they believe social media companies censor political viewpoints they find objectionable.

Table 3 Number of participants (percent of column) by political party and whether they believe social media companies censor political viewpoints they find objectionable (Pew Research Center 2018).

Do social media sites censor?	Democrat	Republican
Yes	981 (58.5%)	1055 (88.4%)
No	696 (41.5%)	139 (11.6%)

Table 4 Number of participants (percent of column) by political party, educational attainment, and whether they believe social media companies censor political viewpoints they find objectionable.

College graduate = “Yes”			College graduate = “No”
Do social media sites censor?	Democrat	Republican	Do social media sites censor?	Democrat	Republican
Yes	535 (52.8%)	494 (87.0%)	Yes	446 (67.3%)	561 (89.6%)
No	479 (47.2%)	74 (13.0%)	No	217 (32.7%)	65 (10.4%)

Fig. 5 Mosaic plot depicting participants by political party, college graduate status, and whether they believe social media companies censor political viewpoints they find objectionable.

Table 5 Odds ratios (95% C.I.) for believing social media sites censor political viewpoints they find objections comparing Republicans to Democrats and college graduates to noncollege graduates for various models.

	Political party	College grad
Model	(Rep/Dem)	(Yes/No)
One-variable	5.38	0.52
	(4.41, 6.61)	(0.44,0.62)
Two-variable	5.14	0.59
	(4.20,6.31)	(0.50,0.71)
Age, race, sex, income adjusted	5.40	0.70
	(4.38,6.70)	(0.57,0.85)

Gassen, J. (2020), “tidycovid19,” available at https://github.com/joachim-gassen/tidycovid19.

 Google Scholar

Pew Research Center (2018), “American Trends Panel Wave 35,” available at https://www.pewresearch.org/internet/dataset/american-trends-panel-wave-35/.

 Google Scholar