TC-06 Ecological fallacy check (aggregation bias check)
Decides whether a claim about people is supported by group-level data and produces a decision on whether the claim is group-level or individual-level.
What it does
Decides whether a claim about people is supported by group-level data. Artefact: a decision on whether the claim is group-level or individual-level. In Robinson's 1930 state data the correlation between illiteracy and foreign-born share was about minus 0.53, against about plus 0.12 for individuals.
When to use it
Use when a claim about individuals is drawn from regional, organisational or document-level aggregates.
When it does not fit
Fits claims about people drawn from regional, organisational or document-level aggregates. Misfit: individual-level data are available, and then they should be used. The reverse error was not covered.
Result format
The result states the claim, the level of the data (group or individual) and the level of the claim. It ends with a decision on whether the claim is group-level or individual-level and whether the group-level data supports it.
Requirements on the material
| Property | Requirement type | Requirement |
|---|---|---|
| Content type | Always applies | requires=tabular |
| Unit of analysis | Always applies | requires=dataset |
| Level of detail available | Always applies | requires=aggregate tables |
| Appraises source credibility | Applies in some cases | no |
Not restricted on: Structure, Variable types, Time structure, How the sample was chosen, Number of sources, Whether sources can be joined, Amount of material, Known data problems, Producer interest, Source dependence, Origin, Document genre, Language, Personal information, Security classification, Licence.
| Minimum material | A claim about individuals and the group-level data it was drawn from. |
|---|---|
| Additional sources needed | none |
| Not for (rules) | access_depth=record-level :: individual-level data are available and should be used instead |
| Not for (text) | The reverse error, where individual patterns are wrongly applied to groups, is not covered. |
Choosing and ordering
| Kind of technique | Verification technique, runs in step 6 |
|---|---|
| Phase tag | validate |
| Other phase tags | analyse |
| Kind of finding | contradiction; gap |
| Question answered | what |
| Strongest statement it supports | describe |
| Needs as input |
|
| Makes as output |
|
| Must exist first | None |
| Loop target | return to gather |
Library checks by stakes
Each stakes level includes the checks of the level below it. Checks that need a person apply at high stakes only, so the process leaves them out at lower stakes (decision 38). Repeatability: judgement dependent.
| Stakes | Check | Person involved | Counts as a different method |
|---|---|---|---|
| exploratory | spot check | optional | no |
| decision-informing | spot check | optional | no |
| decision-informing | alternative explanations check | none | no |
| high-stakes | spot check | optional | no |
| high-stakes | alternative explanations check | none | no |
| high-stakes | independent review | needed | no |
Failure modes and limits
| Model suitability | human judgement required |
|---|---|
| Typical model failure | Reads a group-level correlation as a statement about individuals, or fails to notice that the claim concerns people when the data are regional. |
| Error sources | input data; coder or rater judgement |
| Certainty limits | The decision depends on how the claim is worded and what the unit of the data is. Certainty drops when the data level is not documented. |
Related techniques
- pairs with TC-05 Subgroup and Simpson's paradox check (stratified reanalysis; confounding check; collapsibility check): subgroup reanalysis shows whether a pooled pattern reverses inside groups
- pairs with TC1-08 Confounder identification with directed acyclic graphs (adjustment sets, backdoor criterion): a causal graph states at which level the mechanism acts
Evidence and tag basis
Documented. Read on an encyclopaedia page citing Robinson (1950); the original was not opened. (Wikipedia: Ecological fallacy)
Tag basis: input profile inference; selection inference; sequencing inference; robustness inference. All tag values are inference unless a source is named.