When Two Things Move Together, What Then?

Introduction

Many adult decisions turn on one question: did one thing actually cause another, or did they merely move together? A correlation means two things are associated; causation means changing one thing would change the other. The distinction matters because everyday choices about health, money and work often rely on causal claims: “this supplement improved my sleep”, “this fund manager creates higher returns”, or “this new policy boosted productivity”. Correlations can be useful clues, but they are not proof. Better thinking starts by asking about comparison groups, confounders and alternative explanations before acting on a pattern.

The practical skill is not cynicism. It is disciplined curiosity. Causal inference researchers make the same point in more formal language: observational data can help answer causal questions, but only when the reasoning about possible causes, biases and missing variables is made explicit. [Sage Journals]journals.sagepub.comSage JournalsThinking Clearly About Correlations and Causationby JM Rohrer · 2018 · Cited by 1429 — Correlation does not imply causation…

The Trap: A Pattern Becomes a Story

Correlation is persuasive because the mind likes tidy stories. If a person starts drinking green tea and then loses weight, it is tempting to credit the tea. If a company switches to hybrid working and sales rise, it is tempting to credit the policy. If a stock rises after a chief executive gives a confident interview, it is tempting to treat the interview as the cause. In each case, the pattern may be real while the explanation is wrong.

The most common error is ignoring the missing comparison: what would have happened without the tea, the workplace policy or the interview? Epidemiology makes this point sharply. The US Centers for Disease Control and Prevention describes the comparison group as a key feature of analytic epidemiology, because a claim about exposure and outcome cannot be assessed by looking only at people who had the exposure. [CDC Archive]archive.cdc.govCDC ArchiveSection 7: Analytic EpidemiologyThe key feature of analytic epidemiology is a comparison group. Consider a… Epidemiologic s…

Confounding is the second trap. A confounder is a third factor that is related both to the supposed cause and to the outcome. In health research, for example, people who choose a particular treatment, diet or supplement may also differ in income, existing health, education, exercise, access to care or risk tolerance. Those differences can make the exposure look beneficial or harmful even when it is not the true driver. [PMC]pmc.ncbi.nlm.nih.govPMCMethodological issues of confounding in analyticalPMCMethodological issues of confounding in analytical

A famous adult-decision example is menopausal hormone therapy. Earlier observational studies suggested possible cardiovascular benefit, but randomised evidence from the Women’s Health Initiative and later reviews changed the interpretation: hormone therapy did not show protective cardiovascular effects and increased risks such as stroke and venous thromboembolism. [PMC]pmc.ncbi.nlm.nih.govOpen source on nih.gov. The lesson is not that observational evidence is useless. It is that people who receive or choose an intervention may differ from those who do not, and those differences can distort causal judgement.

Questions That Test a Causal Claim

A good causal question changes the shape of a decision. Instead of asking, “Did these two things happen together?”, ask, “What would make this explanation more likely than the alternatives?” That shift is the mechanism that transfers beyond school: it turns critical thinking into a repeatable adult habit.

Useful questions include:

• What is the comparison group? A claim is weak if it only describes people, teams or investments that received the treatment, policy or exposure.
• What came first? A cause must occur before the effect. If the timing is unclear, the causal story is already shaky.
• What else could explain the result? Look for third factors such as age, income, seasonality, market conditions, workload, prior health or selection effects.
• Was the exposure chosen or assigned? Randomised controlled trials are often powerful for causal claims because randomisation reduces many pre-existing differences between groups. [PMC]pmc.ncbi.nlm.nih.govOpen source on nih.gov.
• Could the data be missing the people who failed? Workplace dashboards, investment backtests and health testimonials often overrepresent survivors and satisfied users.
• Would the claim survive in a new setting? A pattern found in one team, market cycle or patient group may not transfer elsewhere.

This does not mean only randomised trials count. In many adult decisions, randomisation is impossible, unethical, expensive or too slow. Observational studies are often necessary, especially in public health, social outcomes and business settings. The stronger approach is to treat causal claims from observational data as conditional: credible when the design, comparison and assumptions are strong; fragile when they are hidden. [PMC]pmc.ncbi.nlm.nih.govOpen source on nih.gov.

Health: When a Headline Overstates the Cause

Health headlines are fertile ground for causal mistakes because they often compress complex evidence into a simple instruction. “People who eat X have lower risk of Y” can become “X prevents Y”, even when the study only found an association. Research on health communication has found that causal claims in press releases and news can be brought into closer alignment with the underlying evidence without necessarily making the news less interesting. [PMC]pmc.ncbi.nlm.nih.govPMCClaims of causality in health news: a randomised trialPMCClaims of causality in health news: a randomised trial

A practical reader should therefore inspect the study design before changing behaviour. An observational study can suggest that a behaviour is linked with an outcome, but it may not settle whether the behaviour caused the outcome. People who eat more vegetables, sleep longer or use a preventive medicine may differ in many other ways from those who do not. [PMC]pmc.ncbi.nlm.nih.govPMCObservational StudiesPMCObservational Studies

The adult thinking move is to avoid both extremes. Do not dismiss every association, because many real causal discoveries begin as observed patterns. But do not treat every association as a prescription. Stronger confidence comes when several things line up: plausible mechanism, correct timing, consistent findings, good comparison groups, dose-response evidence where relevant, and research designs that reduce confounding.

Finance: When Backtests Pretend to Explain the Future

Finance makes correlation especially tempting because charts look precise. An investment strategy may appear to outperform when tested on historical data, or two assets may appear to move together in a stable way. But a historical pattern is not automatically a causal engine. It may reflect a specific market regime, data mining, omitted variables or luck.

Investor education bodies warn against treating past performance as a predictor. IOSCO’s investment risk education report lists the message that historical performance is a guide, not a predictor, and stresses that investors need to understand risk, diversification and their own tolerance before investing. [IOSCO]iosco.orgOpen source on iosco.org.

The causal question in finance is: what mechanism would make this pattern persist after costs, competition and changing conditions? A fund’s recent outperformance might come from skill, but it might also come from exposure to a booming sector, higher hidden risk, leverage, survivorship bias or simple chance. In quantitative investing, recent discussion has highlighted model risk when backtests and factor models rely too heavily on association without a convincing causal structure. [CFA Institute Research and Policy Center]rpc.cfainstitute.orgbacktests causality and model risk in quantitative investingbacktests causality and model risk in quantitative investing

For personal decisions, this translates into caution about confident explanations. “This asset rose when inflation rose” is not the same as “this asset will protect me from inflation next time”. A better question is whether the proposed cause still makes sense under different interest rates, policy choices, liquidity conditions and investor behaviour.

Work: When Dashboards Reward the Wrong Explanation

Workplace data can make weak causal claims look managerial and objective. A dashboard may show that employees who attend more training sell more, that teams using a certain software close tickets faster, or that people who work in the office more often receive higher performance ratings. Each pattern may be useful, but none automatically proves that the training, software or office attendance caused the result.

Selection effects are common. High performers may be more likely to attend optional training. Better-managed teams may adopt new software earlier. Employees who are more visible may receive higher ratings because managers notice them more, not because office presence directly improves output. Harvard Business Review has warned leaders that data-driven decisions can go wrong when organisations confuse correlation with causation and fail to ask how the data was generated. [Harvard Business Review]hbr.orgHarvard Business Review Leaders: Stop Confusing Correlation with CausationHarvard Business Review Leaders: Stop Confusing Correlation with Causation

The better workplace habit is to slow the jump from “associated with success” to “therefore we should mandate it”. Managers can ask whether comparable teams were used, whether the result appeared before or after the change, whether other changes happened at the same time, and whether a small trial or phased rollout could produce cleaner evidence. Even when a perfect experiment is impossible, a better comparison can prevent an expensive policy built on a misleading pattern.

The Payoff: Better Decisions Under Uncertainty

Learning the correlation-causation distinction does not make adult decisions perfectly scientific. It makes them less gullible. The point is to notice when a story has outrun the evidence and to ask what would make the causal claim stronger.

The most transferable habit is simple: whenever two things move together, pause before naming the cause. Ask what happened to a comparable group, what third factor might explain the link, whether the timing fits, and what alternative story could produce the same pattern. That routine works across health choices, financial decisions and workplace judgement because the same mechanism is at stake each time: turning an attractive association into a tested explanation before acting on it.

Amazon book picks

Further Reading

Books and field guides related to When Two Things Move Together, What Then?. Use these as the next step if you want deeper reading beyond the article.

Book

The Book of Why

By Judea Pearl, Dana Mackenzie

Directly explains the difference between correlation and causation and how to reason about causal claims.

See on Amazon

Book

The Demon-Haunted World

By Carl Sagan, Ann Druyan

Rating: 4.5/5 from 43 Google Books ratings

Demonstrates practical critical thinking, evaluating evidence and questioning unsupported claims.

See on Amazon

Book

Calling Bullshit

By Carl T. Bergstrom, Jevin Darwin West

Shows readers how to evaluate statistical claims, misleading correlations, and evidence.

See on Amazon

Book

The Art of Statistics

By David Spiegelhalter

Explains how to interpret data, uncertainty, and evidence behind causal claims.

See on Amazon

Browse more on Amazon: The Book of Why The Demon Haunted World Calling Bullshit

As an Amazon Associate I earn from qualifying purchases.

eBay marketplace picks

Marketplace Samples

Live-tested eBay searches with available results related to this page.

Shop location

Using USA USA USAUKCanadaAustraliaIreland

Example eBay listing

Educational Saturn Statistics Wall Sticker Art Mural Print Poster Room Decor QM7

Search eBay.co.uk: statistics educational poster

Browse similar on eBay.co.uk

Example eBay listing

Educational Mars Statistics Wall Sticker Art Mural Print Poster Room Decor QM4

Search eBay.co.uk: statistics educational poster

Browse similar on eBay.co.uk

Example eBay listing

Educational Earth Statistics Wall Sticker Art Mural Print Poster Room Decor QM1

Search eBay.co.uk: statistics educational poster

Browse similar on eBay.co.uk

Browse more on eBay.co.uk

Example items shown for inspiration; availability and pricing can change. Branchoria may earn a commission if you purchase through outbound eBay links.

Endnotes

1. Source: archive.cdc.gov
   Link: https://archive.cdc.gov/www_cdc_gov/csels/dsepd/ss1978/lesson1/section7.html
   Source snippet

   CDC ArchiveSection 7: Analytic EpidemiologyThe key feature of analytic epidemiology is a comparison group. Consider a... Epidemiologic s...

2. Source: pmc.ncbi.nlm.nih.gov
   Title: PMCMethodological issues of confounding in analytical
   Link: https://pmc.ncbi.nlm.nih.gov/articles/PMC3755849/

3. Source: pmc.ncbi.nlm.nih.gov
   Link: https://pmc.ncbi.nlm.nih.gov/articles/PMC3731075/

4. Source: pmc.ncbi.nlm.nih.gov
   Link: https://pmc.ncbi.nlm.nih.gov/articles/PMC6235704/

5. Source: pmc.ncbi.nlm.nih.gov
   Link: https://pmc.ncbi.nlm.nih.gov/articles/PMC8020490/

6. Source: pmc.ncbi.nlm.nih.gov
   Title: PMCClaims of causality in health news: a randomised trial
   Link: https://pmc.ncbi.nlm.nih.gov/articles/PMC6521363/

7. Source: pmc.ncbi.nlm.nih.gov
   Title: PMCObservational Studies
   Link: https://pmc.ncbi.nlm.nih.gov/articles/PMC10589119/

8. Source: pmc.ncbi.nlm.nih.gov
   Link: https://pmc.ncbi.nlm.nih.gov/articles/PMC5118066/

9. Source: iosco.org
   Link: https://www.iosco.org/library/pubdocs/pdf/ioscopd505.pdf

10. Source: cdc.gov
    Link: https://www.cdc.gov/field-epi-manual/php/chapters/design-conduct-analyze-field-studies.html

11. Source: cdc.gov
    Link: https://www.cdc.gov/field-epi-manual/php/chapters/analyze-interpret-data.html

12. Source: hsph.harvard.edu
    Link: https://hsph.harvard.edu/research/causalab/courses/

13. Source: hsph.harvard.edu
    Link: https://hsph.harvard.edu/research/causalab/

14. Source: hbsp.harvard.edu
    Title: H06OK3 PDF ENG
    Link: https://hbsp.harvard.edu/product/H06OK3-PDF-ENG

15. Source: youtube.com
    Title: Correlation vs Causation Explained: Why Patterns Can Mislead Us
    Link: https://www.youtube.com/watch?v=ofweqU0Lz4I
    Source snippet

    CRITICAL THINKING - Fundamentals: Correlation and Causation...

16. Source: youtube.com
    Title: CRITICAL THINKING
    Link: https://www.youtube.com/watch?v=U-_f8RQIIiw
    Source snippet

    Tools Podcast S2E2 | Correlation Does Not Mean Cause...

17. Source: journals.sagepub.com
    Link: https://journals.sagepub.com/doi/10.1177/2515245917745629
    Source snippet

    Sage JournalsThinking Clearly About Correlations and Causationby JM Rohrer · 2018 · Cited by 1429 — Correlation does not imply causation...

18. Source: ncbi.nlm.nih.gov
    Link: https://www.ncbi.nlm.nih.gov/sites/books/NBK154456/

19. Source: rpc.cfainstitute.org
    Title: backtests causality and model risk in quantitative investing
    Link: https://rpc.cfainstitute.org/blogs/enterprising-investor/2026/backtests-causality-and-model-risk-in-quantitative-investing

20. Source: rpc.cfainstitute.org
    Title: the factor mirage how quant models go wrong
    Link: https://rpc.cfainstitute.org/blogs/enterprising-investor/2025/the-factor-mirage-how-quant-models-go-wrong

21. Source: hbr.org
    Title: Harvard Business Review Leaders: Stop Confusing Correlation with Causation
    Link: https://hbr.org/2021/11/leaders-stop-confusing-correlation-with-causation

22. Source: pmc.ncbi.nlm.nih.gov
    Link: https://pmc.ncbi.nlm.nih.gov/articles/PMC7737849/

23. Source: pmc.ncbi.nlm.nih.gov
    Link: https://pmc.ncbi.nlm.nih.gov/articles/PMC12499922/

24. Source: pmc.ncbi.nlm.nih.gov
    Link: https://pmc.ncbi.nlm.nih.gov/articles/PMC8786092/

25. Source: pmc.ncbi.nlm.nih.gov
    Link: https://pmc.ncbi.nlm.nih.gov/articles/PMC3326437/

26. Source: pmc.ncbi.nlm.nih.gov
    Link: https://pmc.ncbi.nlm.nih.gov/articles/PMC2266743/

27. Source: linkedin.com
    Link: https://www.linkedin.com/posts/amycedmondson_where-data-driven-decision-making-can-go-activity-7229856878871126016-EN20

Additional References

1. Source: youtube.com
   Link: https://www.youtube.com/watch?v=iMbcMMe0D_Y
   Source snippet

   This video is about how causal models (which use causal networks) allow us to infer causation from correlation, proving the common refrai...

2. Source: youtube.com
   Title: Correlation CAN Imply Causation! | Statistics Misconceptions
   Link: https://www.youtube.com/watch?v=HUti6vGctQM
   Source snippet

   Correlation vs Causation Explained: Why Patterns Can Mislead Us - YouTube Correlation vs Causation Explained: Why Patterns Can Mislead Us...

3. Source: youtube.com
   Title: Critical Thinking Tools Podcast S2E2 | Correlation Does Not Mean Cause
   Link: https://www.youtube.com/watch?v=QQfPhzEq7iA
   Source snippet

   Correlation Doesn't Equal Causation: Crash Course Statistics #8...

4. Source: youtube.com
   Title: Correlation Doesn’t Equal Causation: Crash Course Statistics #8
   Link: https://www.youtube.com/watch?v=GtV-VYdNt_g
   Source snippet

   Correlation CAN Imply Causation! | Statistics Misconceptions...

5. Source: sec.gov
   Link: https://www.sec.gov/investor/locinvestorbehaviorbib.pdf

6. Source: researchgate.net
   Link: https://www.researchgate.net/publication/333135007_Claims_of_causality_in_health_news_A_randomised_trial

7. Source: researchgate.net
   Link: https://www.researchgate.net/publication/308927020_Observational_Research_Rigor_Alone_Does_Not_Justify_Causal_Inference

8. Source: researchgate.net
   Link: https://www.researchgate.net/publication/322778777_Thinking_Clearly_About_Correlations_and_Causation_Graphical_Causal_Models_for_Observational_Data

9. Source: teachepi.org
   Link: https://www.teachepi.org/wp-content/uploads/OldTE/documents/courses/bfiles/The%20B%20Files_File1_HRT_Final_Complete.pdf

10. Source: efama.org
    Link: https://www.efama.org/sites/default/files/files/EFAMA_Investor_Education_Report.pdf