250 years of US history: how has the country's interaction with the world changed over the years?

Idea: Peter Hille

Data research: Peter Hille & Gianna-Carina Grün

Data analysis and visualization: Gianna-Carina Grün

Interviews: Peter Hille

Writing: Gianna-Carina Grün

Video: Peter Hille

Data sources

Military Intervention Project (MIP): Military Interventions until 2019

The latest available dataset (2022) was downloaded from the Military Intervention project's website at Tufts University.

Before analysis, data was modified in two ways:

• we noticed the spelling of North America was off in few cases, this was unified
• World War II was only assigned to the region Asia in the UNRegT column. We manually changed that in the aggregated data for the visualization, to also display for Europe and Africa
• We filtered on the RemoveCase column to exclude all entries with a value of 1 and furthermore filtered on the US HiHost column all entries with a value of 1. After filtering, 392 entries remain and entered into the analysis stage. All code for the analysis can be found in the MIP-military-interventions.ipynb notebook.

Objectives: For analysing the objective over time, we relied on the column ObjectiveCode that is not explicitly mentioned in the codebook, but matches the verbatim description of the Objective column. To simplify our visualization of the data, we re-coded entries in category 6.1 and 6.2 to instead count toward the umbrella category 6. While in the original dataset, one row corresponds to one intervention that can be assigned several objectives, for analysis we de-aggregated the objectives, so each event appears as many times in the dataset as it has objectives assigned -- to make counting the prevalence of individual objectives over time possible.

Eras: When investigating objectives over time, we decided to use the Era column that already classifies the years of intervention into different timeframes. The codebook mentions seven different eras, however, in the dataset there are only 6 present. In order to match eras back to actual years, we inspected the dataset to spot in which years the value in the era column changes.

Measuring American Diplomacy: UN-registered treaties until 2021

The dataset on treaties from Measuring American Diplomacy were downloaded from the referenced dataverse site. The presidents and their party affiliation were taken from GovTrack, in order to compare between administrations and whether party affiliation made a difference for the number of UN-registered treaties being signed.

Nira Data: Perception of the US abroad, 2022 vs 2026

Data on how other countries see the U.S. was provided upon request from Nira Data (Fred De Veaux and Nico Jaspers (2026): Global Country Perceptions Database). For recreating the comparison chart from their website, we used data from the 2026 Democracy Perception Index (column BU / Q24-Qn) and compared it to the 2022 version (colum AW/ Q23).

For 2026, the survey was conducted between March 19th and April 21st. For 2022, between March 30th and May 10th. The survey was targeting online-connected respondents and for 2026 was mobile optimized. The survey size for 2026 had a total sample size of n=94,146 acorss 98 countries, the 2022 survey of n=52,785 across 53, with an average of 1K respondents per country, although these vary (see next paragraph). For 2026, the average margin of error (95% confidence) was given at "approximately ±4.7 percentage points". For 2022, no such information was given.

For the 2022 version, the sample size was generally higher, with Morocco, Algeria, Pakistan and Israel having the fewer than the average 1K respondents, but still more than 700 each. For the 2026 survey, the sample size in some countries was a lot lower, among them Switzerland (n= 280), Austria (397), Ireland (562), Russia (385) and Israel (483) with the lowest number of respondents. Upon request, Nira Data explained that for these countries in particular the above mentioned margin of error would apply. Since the differences between 2022 and 2026 are larger than the potential margin of error, we concluded the overall trend will still hold true despite the low sample size in 2026.

Gallup: How Americans see the US

Data was used without further alteration as included on the Gallup Website. Gallup confirmed upon request that each datapoint builds upon a survey of around 1K people. In 2026 these surveys were conducted by telephone (20% landline, 80% cell phone). The margin of sampling error was given at ±4 percentage points at the 95% confidence level. All reported margins of sampling error include computed design effects for weighting.