Strong Cross-Pillar

Relative vs Absolute Risk: The Number That Makes Headlines Lie

Name: Relative vs Absolute Risk: The Number That Makes Headlines Lie
Creator: Realised
Published: 2026-06-01

Summary

"X raises your risk 50%" is almost meaningless on its own: if the risk goes from 2-in-1,000 to 3-in-1,000, that's a 50% relative increase and a 0.1% absolute one — and the headline always quotes the big relative number while hiding the tiny absolute one, so the single most useful stats skill is to demand the absolute change before you react.

Why Strong

Tier 1 (Strong). The arithmetic is deterministic and the persuasion asymmetry is well-documented in the medical-communication literature. It sits under the Foundational hub (rct_vs_observational_evidence) as a Strong, concrete tool rather than a Foundational axiom.

Practical takeaway

• Always ask: "from what baseline, to what?" Convert any relative claim into an absolute change before reacting. No baseline given = treat the claim as incomplete.
• Translate to Number Needed to Treat/Harm (≈ 1 ÷ absolute change): "how many people, for how long, for one extra good/bad outcome?" — the most intuitive form.
• Match like with like: never compare a relative figure to an absolute one.
• Weigh benefit against cost/harm in absolute terms — a 1% absolute benefit may or may not be worth a drug's side effects and price; you can only judge with the real number.
• Apply it symmetrically — to scares (processed meat, alcohol, a pollutant) and to miracle claims (a supplement, a drug, a superfood).

Evidence detail

Why This Entry Exists

This is the most decision-relevant statistics skill there is, and almost nobody is taught it. Risk is reported in two ways — relative (the proportional change between groups) and absolute (the actual change in your chance) — and they can differ by orders of magnitude. Headlines, drug ads, and scare-pieces quote whichever number is more dramatic, which is almost always the relative one. Learn to ask "from what to what, in absolute terms?" and most health scares and miracle claims deflate to their real size.

It's a spoke of rct_vs_observational_evidence: even a perfectly-run trial can be honestly reported and still mislead if only the relative figure is given.

What bad thinking this protects against:
• "Bacon raises bowel-cancer risk 18%!" → panicking over a large relative number sitting on a small absolute baseline.
• "This drug cuts heart attacks by a third!" → starting a lifelong drug for a fraction-of-a-percent absolute benefit you'd weigh differently if you saw it.
• Comparing a relative number to an absolute one → apples-to-oranges fear or hope.

Evidence

1. The arithmetic (deterministic). If baseline risk is 2% and an exposure raises it to 3%:
• Relative risk increase = (3−2)/2 = 50% ("raises risk by half!").
• Absolute risk increase = 3%−2% = 1 percentage point.
• Number Needed to Harm ≈ 1/0.01 = 100 (100 people exposed for 1 extra case).
Same data, three honest framings — and the first sounds ~50× scarier than the third.

2. The classic real example — processed meat (Tier 1). The WHO/IARC figure: ~50 g/day of processed meat raises colorectal-cancer risk ~18% (relative). Baseline lifetime CRC risk is ~5%, so the absolute increase is on the order of ~1 percentage point. "18% higher" drove global headlines; "~1 in 100 extra over a lifetime" is the honest size.

3. The mirror case — benefits (Tier 1). A drug that takes 5-year event risk from 4% to 3% is a 25% relative reduction and a 1% absolute one (NNT = 100: treat 100 people 5 years to prevent 1 event). Whether that's worth the cost/side-effects is a different decision once you see the absolute number — which ads omit.

4. Why the relative number is the default (Tier 1, documented). Studies of medical reporting and drug advertising consistently find relative figures dominate because they're larger and more persuasive; presenting absolute risk or NNT changes how patients and clinicians decide. This is a known, exploited asymmetry.

Mechanism

Why the two diverge. Relative measures divide out the baseline; absolute measures keep it. When the baseline is small, a large proportional change is a tiny actual change (rare-disease scares). When the baseline is large, even a modest relative change can be a big absolute one. So the same relative figure means wildly different things depending on a baseline the headline usually omits.

Why it persuades. Bigger numbers feel more important; "50%" triggers more action than "1 in 1,000 extra." Marketers and journalists know this, so the framing is chosen for impact, not clarity. The fix is mechanical: always reconstruct the absolute change and the NNT/NNH.

RISKS AND CONTRAINDICATIONS (how this gets misused)

• Dismissing real risks because the absolute number is "small" — a 1% absolute lifetime risk across a population is many people; absolute framing aids personal decisions but isn't a reason to ignore population-level effects.
• Cherry-picking the framing yourself — using absolute to downplay things you like and relative to inflate things you dislike. Apply consistently.
• Ignoring cumulative/lifetime vs annual — make sure the baseline timeframe is clear (per-year vs lifetime risks differ hugely).

Controversy

There's little scientific controversy that absolute risk and NNT are the more informative frame — the "controversy" is why they're so rarely used, and that's about incentives, not statistics: dramatic relative numbers sell papers, drugs, and supplements. Realised's position is simply to always surface the absolute number.

Cross-Pillar Connections

• Hub (rct_vs_observational_evidence): the parent literacy skill — design tells you if a number is valid; this tells you if it's meaningful.
• publication_bias_and_evidence_distortion: even valid, well-framed numbers can come from a skewed literature.
• cui_bono_industry_funding_bias: who chose the relative framing, and why.
• surrogate_endpoints_vs_outcomes: a relative reduction in a surrogate is doubly removed from real benefit.

Industry bias note

Structural incentives the evidence base may reflect

The relative-risk frame is the favoured tool of both sides: pharma quotes relative risk reduction to sell drugs; scare-marketing and some wellness/clean-living content quote relative risk increase to sell fear (and the alternative product). The honest anchor — absolute risk and NNT — is what independent evidence-communication researchers (and good clinical guidelines) advocate. Demanding the absolute number is a one-move defence against both.

Sources (4)

Open in the Library: search, filter, every entry →