# Quasi-Experimental and Non-Experimental Designs

Psychology I · Research Techniques · https://tryals.app/learn/psychology-i/quasi-experimental-and-non-experimental-designs

## When Random Assignment Fails

Not every question can be studied with a true experiment. Researchers cannot ethically or practically assign people to smoke, experience trauma, or be born male or female.

**Quasi-experimental designs** compare groups that differ naturally, like a pre-existing condition, **WITHOUT random assignment**. A **natural experiment** compares regions before and after an uncontrolled policy change. They are informative, but weaker at ruling out confounds than true experiments.

**Non-experimental designs** involve zero variable manipulation.

| Design | Manipulation? | Random Assignment? | Can Establish Causation? |
| --- | --- | --- | --- |
| True Experiment | Yes | Yes | Yes |
| Quasi-Experiment | Sometimes | No | Weakly, at best |
| Correlational | No | No | No |
| Case Study | No | No | No |

## Non-Experimental Designs

**Correlational studies** measure the relationship between two variables as they naturally occur, such as hours of sleep and reported mood. 

**Critical Pitfall:** Treating a strong correlation as proof of causation. A correlation could mean variable A causes B, B causes A, or a third variable (like chronic stress) causes both.

**Case studies** examine a single individual or small group in great depth, perfect for rare phenomena like an unusual brain injury. **Archival research** analyzes existing records or data collected for another purpose entirely.

## Practice questions

7 of this lesson's 12 practice questions, with answers. The full set is in the app.

### 1. Ice-cream sales and drowning deaths rise and fall together across the year. Nobody thinks ice cream causes drowning. Why does this obvious case generalise into a rule that applies to correlations where the third factor is NOT obvious?

A. Because this correlation is spurious and others are not.
B. Because seasonality should be checked first in any study.
C. Because variables that rise together are never causal.
D. Because the logical structure is identical either way.

**Answer:** D. Because the logical structure is identical either way.

**Why:** The example is chosen because the confounder is easy to see, and that obviousness makes the lesson feel narrow. Strip out the content and the structure is bare. When a study reports that breakfast-eaters weigh less, the absence of an obvious confounder is a fact about your imagination, income and health consciousness are sitting right there.

Page: https://tryals.app/practice/psychology-i/quasi-experimental-and-non-experimental-designs/ice-cream-sales-and-drowning-deaths-rise-and-fall-together-across-the

### 2. A large and statistically robust correlation is enough to establish which variable causes the other.

**Answer:** False

**Why:** False. Size and significance say nothing about direction or about a lurking common cause. Ice-cream sales and drownings correlate strongly through summer heat. Only intervention, randomised assignment, settles causation, which is why the correlational/experimental distinction is the backbone of methods teaching.

Page: https://tryals.app/practice/psychology-i/quasi-experimental-and-non-experimental-designs/a-large-and-statistically-robust-correlation-is-enough-to-establish

### 3. Comparing two regions before and after a policy the researcher did not control is quasi-experimental, not experimental. What is missing?

A. A large enough sample to detect an effect
B. Random assignment, so the regions are comparable
C. A control group, which quasi-experiments never have
D. A measurable outcome variable

**Answer:** B. Random assignment, so the regions are comparable

**Why:** Randomisation balances unmeasured differences. Without it, correlation with the policy is confounded with whatever made those regions different in the first place.

Page: https://tryals.app/practice/psychology-i/quasi-experimental-and-non-experimental-designs/comparing-two-regions-before-and-after-a-policy-the-researcher-did

### 4. Sort each design by whether it can, on its own, establish causation.

**Answer:**

- Can establish causation: True experiment with random assignment
- Cannot establish causation on its own: Correlational study, Case study of a single individual, Quasi-experiment without random assignment, Archival analysis of existing records

**Why:** Only the true experiment, through random assignment, controls confounds well enough to support causation. Correlational, case, quasi-experimental, and archival designs can suggest relationships but cannot, by themselves, establish which variable causes which.

Page: https://tryals.app/practice/psychology-i/quasi-experimental-and-non-experimental-designs/sort-each-design-by-whether-it-can-on-its-own-establish-causation

### 5. A quasi-experiment compares crime rates in a city before and after a new law, which the researcher had no power to assign. Why does the missing random assignment weaken the causal claim, when the law itself was a genuine change imposed on everyone?

A. Because before-and-after always shows some change.
B. Because a researcher must administer the treatment.
C. Because the periods may differ in ways beyond the policy.
D. Because the sample was not randomly selected.

**Answer:** C. Because the periods may differ in ways beyond the policy.

**Why:** Randomisation makes groups equivalent on average across everything, including what nobody measured, and a before-and-after design differs in every respect that changes over time. The finding is ambiguous rather than merely uncertain. It is not worthless: some questions cannot be randomised, and the remedy is ruling out rivals explicitly with comparison cities and pre-trends.

Page: https://tryals.app/practice/psychology-i/quasi-experimental-and-non-experimental-designs/a-quasi-experiment-compares-crime-rates-in-a-city-before-and-after-a

### 6. You are handed a study reporting that a variable X is associated with an outcome Y, and asked whether X causes Y. Order the questions to ask, from the one that settles the most to the one that settles the least.

**Answer:**

1. Did the researcher assign participants to levels of X at random?
2. If not, did the researcher at least control who received X and when?
3. If not, was X observed as it varied naturally, with the groups differing in other ways too?
4. Could a third variable plausibly drive both X and Y?
5. Could Y in fact be causing X, rather than the other way round?

**Why:** The first question is first because a yes settles almost everything. Random assignment makes the groups equivalent on all other variables at once, which is why a true experiment does not need you to think of every possible confounder - it neutralises the ones you never considered.

Each subsequent step is a weaker version of the same protection, and when the answer to all three design questions is no, you are left doing by hand what randomisation would have done automatically: naming plausible third variables and arguing them away one at a time, and checking the direction of the arrow.

The reverse-causation question sits last not because it matters least but because it is the one people forget. Screen time is associated with poor sleep, and it is entirely plausible that not sleeping is what leaves someone scrolling at 2 a.m. A correlation is symmetric; the causal story you find natural is not.

Page: https://tryals.app/practice/psychology-i/quasi-experimental-and-non-experimental-designs/you-are-handed-a-study-reporting-that-a-variable-x-is-associated-with

### 7. You cannot randomly assign people to a hurricane, yet the health effects of disasters are studied causally. What makes a **natural experiment** more than an observational study?

A. Nothing, a natural experiment is an observational study with a more persuasive name
B. That natural events affect everyone equally, removing selection effects
C. That the sample is larger, which compensates for the lack of assignment
D. That something outside the researcher's control assigned exposure

**Answer:** D. That something outside the researcher's control assigned exposure

**Why:** What matters is whether assignment was independent of the outcome, and a storm path can be as indifferent to your health as a coin. It is weaker than a trial, you must argue the exposure really was as-if random, and far stronger than comparing people who chose their circumstances.

Page: https://tryals.app/practice/psychology-i/quasi-experimental-and-non-experimental-designs/you-cannot-randomly-assign-people-to-a-hurricane-yet-the-health