6.0 The selective or correlational method

The selective or correlational method

EX POST FACTO

After the fact

After the fact (EX POST FACTO)

• observation of what has already happened

• help us look at group differences & correlation (variables)

In ex post effect researchers

• reverse searching (explore factors that led to differences)

• takes effect (DV), examines data retrospectively → establishes causes, relationships or associatio. ns & meaning.

• not possible to control variable before or during

EX POST FACTO STUDIES → researchers can report

• what has happened or what is happening

• IV cannot be manipulated

• inferring causes tentatively

• causation cannot be demonstrated.

What is an Ex Post Facto Design?

• studies existing differences between groups because the IV cannot be manipulated.
  

Ex Post Facto Design is used when

• Used when random assignment or experimental control is impossible, unethical, or impractical
  (e.g., smoking, gender, trauma, socioeconomic status).

What is an Ex Post Facto (Causal-Comparative) Design

• Used when → IV has already occurred or cannot be manipulated (e.g., gender, smoking, trauma).

• Researchers compare naturally existing groups that differ on one key condition.

• Resembles an experiment but without random assignment & without control over the IV → causal claims are tentative.

limitations of Ex Post Facto Designs

• other variables may differ between groups → cannot confidently infer cause & effect.

researchers try to figure out cause-&-effect in an ex post facto design, even though they cannot manipulate the independent variable.

• List possible causes .

• Check the data

• Rule out (“eliminate”) not supported causes

• Make sure sampling and comparisons are fair (keep controlled) → so group differ only on the variable of interest.

8 stages of EX POST FACTO PROCEDURES

1. Define the problem & survey the literature

2. State the hypotheses → the assumptions or premises on which the hypotheses & research procedures are based.

3. Select the subjects (sampling) & identify the methods for collecting the data

4. Identify the criteria & categories for classifying data: → fit purposes, unambiguous, enable relationships & similarities

5. Gather data from always present factors & discard data from factors are not always present

6. Gather data on always present factors in which the given outcome does not occur

7. Compare the two sets of data (stage 6 & stage 5)→  infer the causes that are responsible for the occurrence or non-occurrence of the outcome.

8. Analyse, interpret & report the findings.

APPROACHES IN EX POST FACTO RESEARCH

• Prospective

• Retrospective

Prospective

• Participants differ on an independent variable

• → later studied how they differ on the dependent variable

•Retrospective

• Participants differ on the dependent variable

• → later studied how they differed on a range of independent variables

Retrospective Studies

• co-relational (or “causal”) study.

• identifies the antecedents of a present condition

• involves the collection of two sets of data (retrospective relationship between X → O)

• value lies in explanatory/ suggestive character

limitations in RETROSPECTIVE STUDIES

• difficult to determine what causes what: whether A causes B or B causes A.

Three types of retrospective designs

• Simple retrospective designs

• Retrospective design with quasi-control group

• Single-group retrospective design

Simple retrospective designs

• Researcher starts with the outcome (DV) & then looks back for possible causes (IV).

• Only one group is studied.

• Used to look for patterns or associations, not causes.

Why are Retrospective Designs Weak?

• The DV does not vary → cannot establish covariation.

• No control group → no comparison.

• Results can only show coincidences/correlations, not causation.

• High risk of bias in interpreting patterns.

Retrospective Design with Quasi-Control Group

• Researchers start with a group that has already experienced the outcome (DV).

• Then they find a similar group without that outcome (quasi-control).

• They compare groups to look for differences in potential causes (IVs).

• Improves comparison, but still no random assignment, so no strong causal inference.

What does adding a quasi-control group improve?

• It allows comparison between groups → can test for covariation.

• Reduces some alternative explanations.

• BUT, because groups were not randomly formed, group differences may still exist.
  → Causal conclusions remain tentative.

Single-Group Retrospective Design

• Uses one group only (no control group).

• The DV is measured first, then possible IVs are identified by looking backward.

• Uses correlational analysis to find patterns.

• Cannot infer cause & effect → only associations.

hat are the key features & limitations of single-group retrospective designs?

• One group → DV and IV vary across participants.

• Very large sample often needed for representativeness.

• No comparison group, no manipulation, no random assignment.
  → Results are correlational, not causal.