Evidence
There are good reasons to run a political simulation and there are reasons people give that the evidence does not support. Both are below, with the citation that stands behind each claim, because a case built on the real findings survives a skeptical colleague and a case built on the usual ones does not.
The claims, and what stands behind each one
| The claim | What the evidence actually shows |
|---|---|
| Students learn the content better than from a lecture | Mostly no, and that was settled for twenty years, but it is no longer flatly false. Krain and Lantis (2006) found different kinds of knowledge rather than more of it; Raymond (2010) found no significant difference in exam performance; Powner and Allendoerfer (2008) similar. Shanks and Zhang (2024) is the first well-powered challenge: a three-year natural experiment in which some sections of an introductory course got simulations and others got debates and discussions, 244 students, measured on weekly quizzes and a final exam rather than on self-report, finding gains for stronger and weaker students alike. One study at one institution does not overturn three, so claim this carefully or not at all |
| They remember it longer | Supported, and this is the strongest claim available. Wunische (2019): lecture students learned more at first, simulation students retained more over time. Levin-Banchik (2018): simulation plus an in-class debrief beat every other condition on an anonymous quiz three months later |
| Students find the subject more interesting | Supported in most studies and not in the largest one. Hendrickson (2021) found higher excitement and interest, surveying students on each active component separately. Clark and colleagues (2017) found the same direction for interest in EU politics. But Clark and Scherpereel (2024), four consecutive years of pre and post surveys across six institutions, found no significant change in interest in the EU or in politics generally; only discussion of international news moved. Worth claiming as likely, not as established |
| Students engage with the course more: they prepare, turn up, and take part | Supported, and it is the reason most teachers actually give. Baranowski and Weir (2010) found that students holding roles with real power reported more understanding and judged the exercise more worthwhile, which makes engagement something a design produces rather than something simulations have. Hendrickson (2021) found the same direction across three different active formats. Bernstein and Meizlish (2003) still found differences three years out |
| They understand how the institution actually works | Supported, with a caveat. Krain and Lantis (2006) found students gained a different kind of knowledge, how the process moves rather than more facts about it, and Levin-Banchik (2018) found that kind of knowledge still present three months later. No study shows it beats a lecture course covering the same ground |
| They build skills: speaking, negotiating, leading | Not supported. The largest in-discipline test, Clark and Scherpereel (2024), looked for skill gains across 924 responses and did not find them. Kartal (2021) explains why self-rated gains look larger than they are: students revise their opinion of their former selves downward. Oberle and colleagues (2020) put a number on the gap, finding that measured knowledge gain correlated only r = .18 with students’ own sense of having learned. Baranowski and Weir (2015) is blunter still, that most designs in this literature could not establish a claim like this one even if it were true |
| Students believe they can do the work | Supported as self-efficacy, which is not the same as skill. Hendrickson (2021) measured confidence in succeeding at the course and found it rose; Kensicki and colleagues (2022) found the same across two weeklong Model G20 summits. Both are self-reports, so claim the belief and not the ability |
| They build empathy, or understanding of opposing views | One tentative result. Clark and Scherpereel (2024) found position shifts in an alter-ego exercise. Worth pursuing, not worth promising |
| Students become more politically engaged citizens | Not supported. Bernstein and Meizlish (2003) found real differences three years after a congressional simulation and no difference in political participation. Engagement with a course is not the same thing as engagement with politics, and the evidence separates them |
| A single class period does what a three-day conference does | Not supported, and this is the finding most likely to change what you build. Leib and Ruppel (2020) compared a short in-class Security Council exercise, a regional Model UN and National Model UN delegations across 159 students: only the long, heavily prepared formats reached all of their intended outcomes, and the hardest goals were reached by 10 to 25 percent of students. Kensicki and colleagues (2022) point the same way. Length and preparation look like what separates a simulation that works from one that does not |
| It gives the class a shared reference point for the rest of the term | Untested. No study measures it. It is a reasonable design claim and should be offered as one, not as a finding |
| It levels the playing field for quieter students | The evidence does not run one way, and the one intervention anybody tried backfired. Engel, Pallas and Lambert (2021): women participated less relative to their numbers, the gap cost them grades, and a short intervention meant to close it made it worse. Coughlin (2013) found the same pattern at a conference; Kaftan and Linantud (2025) found everyone’s grades rose and no gap closed. But Davesa and Piros (2019), surveying about 300 participants across two multi-institution EU simulations, found women among the better performers. The studies differ in what they measure, participation and grades inside one course against self-reported outcomes after a conference. Promise neither outcome |
Read down that second column and the honest case is narrow but solid: retention, engagement, confidence, and a kind of understanding that a lecture reaches differently, with interest likely and contested. That is enough. The claims that fail are the ones nobody needed.
One design finding runs underneath the whole table and is worth more than any single claim on it: what students keep comes from the exercise plus a structured conversation afterward, and length and preparation matter more than which format you pick. A long simulation with no debrief is a worse buy than a short one with a good one.
Where these citations come from
Every study named above is set out in full on What the Research Says, grouped by the question it answers rather than by date, with what each study’s design can and cannot establish, the methodological warning that should travel with any survey result about skills, and the one widely quoted effect size that is mostly medical education. The whole collection is also there as a printable reading list.
Every citation here was checked against publisher or repository records. If you find an error, or a study we have missed, please say so.
