Talk:Quasi-experiment

Merge?

Why do we have Quasi-experiment and Design of quasi-experiments? Should they be merged? Tayste (edits) 00:22, 9 August 2012 (UTC)

• Yes Thosjleep (talk) 00:06, 5 September 2012 (UTC)
• Yes there is almost total overlap71.208.252.27 (talk) 15:39, 6 November 2012 (UTC)
• Yeah, I dunno why there are separate pages either. I doubt anyone's going to object, so whoever wants to can probably just go ahead and do it. --V2Blast (talk) 03:19, 15 November 2012 (UTC)

Bnadrow (talk) 16:11, 21 February 2013 (UTC) I'm going to begin editing this page. I am a student at Clemson University working with Dr. June Pilcher's Senior Psychology class.

• Yes, I also believe that they should be merged. (Elena Loysha)

Merge complete --Thosjleep (talk) 13:22, 27 February 2013 (UTC)

Use of first person in Ethics

The ethics section is written in first person, for example "we divide households" and "we run a linear regression". Is this acceptable? alex3yoyo (talk) 02:35, 7 March 2013 (UTC)

Geared to strongly towards medical research?

Is the article possibly heard to strongly towards medical research? I'm referring specifically to talk of "treatment conditions" (or is this some neutral jargon that I'm not aware of?)

Also should there be a link to this article from "experiment" which doesn't seem to be aware of the existence of "quasi experiments", just "natural experiments". — Preceding unsigned comment added by 220.233.65.12 (talk) 08:58, 2 August 2014 (UTC)

Quasi-experiments: to be credible ALL records from start to finish must be disclosed

"Quasi Experiments are also effective because they use the "pre-post testing". This means that there are tests done before any data is collected to see if there are any person confounds or if any participants have certain tendencies. Then the actual experiment is done with post test results recorded. This data can be compared as part of the study or the pre-test data can be included in an explanation for the actual experimental data." -- from the article

"Cherry pick your data carefully, and don't forget to cover your tracks" -- translation — Preceding unsigned comment added by 71.102.112.169 (talk) 03:43, 7 October 2014 (UTC)

Inconsistent types of quasi-experiments

The list of quasi-experiments is very inconsistent, it mixes things like statistical techniques/methods (e.g. difference in differences) with proper quasi-experimental designs (e.g. Nonequivalent control groups design). That section clearly needs to be re-structured. — Preceding unsigned comment added by 193.146.15.253 (talk) 16:44, 3 February 2016 (UTC)

External links modified

Hello fellow Wikipedians,

I have just modified 2 external links on Quasi-experiment. Please take a moment to review my edit. If you have any questions, or need the bot to ignore the links, or the page altogether, please visit this simple FaQ for additional information. I made the following changes:

• Added {{dead link}} tag to http://www.osulb.edu/~msaintg/ppa696/696quasi.htm
• Added archive http://web.archive.org/web/20120916072343/http://ssmon.chb.kth.se/safebk/Chp_4.pdf to http://ssmon.chb.kth.se/safebk/Chp_4.pdf
• Added archive http://web.archive.org/web/20120328052718/http://ssmon.chb.kth.se/safebk/ to http://ssmon.chb.kth.se/safebk/

When you have finished reviewing my changes, please set the checked parameter below to true or failed to let others know (documentation at {{Sourcecheck}}).

This message was posted before February 2018. After February 2018, "External links modified" talk page sections are no longer generated or monitored by InternetArchiveBot. No special action is required regarding these talk page notices, other than regular verification using the archive tool instructions below. Editors have permission to delete these "External links modified" talk page sections if they want to de-clutter talk pages, but see the RfC before doing mass systematic removals. This message is updated dynamically through the template {{source check}} (last update: 5 June 2024).

• If you have discovered URLs which were erroneously considered dead by the bot, you can report them with this tool.
• If you found an error with any archives or the URLs themselves, you can fix them with this tool.

Cheers.—InternetArchiveBot (Report bug) 18:12, 20 July 2016 (UTC)

Context missing

The whole article is written assuming it refers to statistical analysis in medical research without ever saying that. I think a word of introduction about the field in which this term is used would be helpful.

I get redirected to this article when I look for "pseudoexperiment", which we do a lot in particle physics, but is something very different. Pkoppenb (talk) 09:34, 1 June 2018 (UTC)

Needs merge to more common term "natural experiment"

The following discussion is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

---

"Natural experiment" seems to be much more common and I'm not sure there's a distinction. – Closed Limelike Curves (talk) 20:05, 4 January 2026 (UTC)

Strongly oppose: As noted in the article natural experiment, it is key that there is random assignment (performed by natural processes). For example, it is essentially random whether children are born male, female, or intersex, even though this is not a characteristic that scientists can assign themselves--it's chosen by nature.

Under quasi-experiments, there is no random assignment. Treatment is assigned based on some factor other than randomness. For example, we may have reason to believe that there are parallel trends that enable the causal analysis. For example, in David and Card's famous 1994 study on minimum wages, they decide that New Jersey and eastern Pennsylvania's fast food locations are similar enough pre-treatment to roughly estimate the effects of a new minimum wage law in New Jersey (whereas Pennsylvania's remained constant) on those firms. Clearly, the treatment was not randomly assigned, neither by the scientists nor nature--it was specific to those fast food restaurants in New Jersey. But they seemed otherwise similar enough prior to and after the policy's implementation to warrant the parallel trends assumption.

These are completely different methods. They may share some core ideas, which is why they both evaluate causality, inspired by experimental design. But they are very different from each other. LaborHorizontal (talk) 15:52, 27 January 2026 (UTC)

Oppose. The natural experiment and the standard quasi-experiment are different methods as indicated by User:LaborHorizontal, although some would, with reason, say that the natural experiment is a kind of quasi-experiment. What sets the natural experiment apart from a the standard quasi-experiment (different treatments administered to pre-existing groups without involving random allocation of research participants to groups) is that the natural experiment mimics a true experiment's random assignment. But we don't merge the natural experiment and the experiment entries because we want readers to appreciate the nuanced differences. The natural experiment is much rarer than the true experiment in the research literature. Its rarity and the cleverness of the researchers who leverage it should be understood by readers without muddying their view. However, having cross-links among the concepts of natural experiment, experiment, and quasi-experiment is important. I added a link on the natural experiment entry to the quasi-experiment entry. I also added a link on the quasi-experiment entry to the natural experiment entry. There were already cross-links involving the entries for the experiment, on one hand, and quasi-experiment and natural experiment, on the other. Iss246 (talk) 16:28, 27 January 2026 (UTC)

Good points. I agree on illustrating the nuances even within experiments. LaborHorizontal (talk) 12:09, 28 January 2026 (UTC)

it is key that there is random assignment (performed by natural processes)

In both cases there is quasi-random assignment performed by natural processes. In Card-Krueger, restaurants were quasi-randomly assigned to either an increased minimum wage or to the preexisting minimum wage, based on a feature that was previously irrelevant to changes in employment (which side of the PA-NJ border the restaurant was on). In any case, this cuts against the argument that the two are distinct, since RSes frequently describe Card–Krueger using the words "natural experiment". – Closed Limelike Curves (talk) 23:23, 27 January 2026 (UTC)

I'd argue the PA-NJ case isn't "random" in the same way as a natural experiment as treatment assignment is correlated to each state's different characteristics—it's just that it's irrelevant due to the DiD design. But I'd consider a process like sex assignment perhaps the closest thing to true randomness (assuming it has no relation to any other factors), even better than computer-generated or pull-out-of-a-hat "random" assignment as these even are slightly affected by human biases (e.g., computer randomness is technically predictable by predetermined algorithms, preference of how to mix and pick the cards in the hat).

But maybe it's debatable what's "truly" vs "quasi" random. Interesting find on the Nobel site, which I'll have to look more into. My department has generally considered it non-natural but quasiexperimental, at least.

Anyway, I think the existing sources on both natural experiment and quasi-experiment clearly illustrate that the two are distinct concepts (probably better than I'm explaining). LaborHorizontal (talk) 11:57, 28 January 2026 (UTC)

I agree that the NY-NJ study is not the most illustrative example. A better example would the Vietnam draft lottery study described in the natural experiment entry. Here is a link: Natural_experiment#Vietnam_War_draft. Iss246 (talk) 18:14, 28 January 2026 (UTC)

I'd argue the PA-NJ case isn't "random" in the same way as a natural experiment as treatment assignment is correlated to each state's different characteristics

That's true, but is more a criticism of Card–Krueger's status as a valid quasi/natural experiment at all.

Anyway, I think the existing sources on both natural experiment and quasi-experiment clearly illustrate that the two are distinct concepts (probably better than I'm explaining).

Mostly I think they show the uses are inconsistent (e.g. the Nobel source I gave): some sources use them as synonyms whereas others draw slight/subtle distinctions between them. If I had to draw a line, I'd say I usually see natural experiment used to describe simple A/B tests with control and treatment groups, whereas quasi-experiment allows more complex designs like IV or regression discontinuity, but it's not at all consistent. I think a strong case for keeping these articles separate has to rest on practical arguments for why this distinction would be useful to have on Wikipedia (e.g. written for different fields or at different technical levels and the combined article would be too long). – Closed Limelike Curves (talk) 04:45, 1 February 2026 (UTC)

The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.