confidence interval
A range of values, computed from data, that is likely to contain the true parameter with a stated level of confidence.
As defined in How to Avoid Lying With Statistics (with Jeremy Weber)
Other definitions
A range of values within which the true value is expected to fall at a stated probability; wider intervals mean more uncertainty.
Data, Decisions, and Basketball
Where it comes up · 16
- How to Think So Clearly People Assume You’re A Genius2 claims
- Nature, Nurture, and Identical Twins (with David Bessis)2 claims
- Brian Nosek on the Reproducibility Project2 claims
- How to Avoid Lying With Statistics (with Jeremy Weber)2 claims
- Rebecca Lynn - Finding Your Customers - [Invest Like the Best, EP.309]2 claims
- The "Profound" Impact of Rate Cuts Have BEGUN1 claim
- Sink or Swim: How to Combine Quant and Traditional Asset Management Techniques1 claim
- Data, Decisions, and Basketball1 claim
and 8 more
Most important claims · 10 of 21Search all claims for “confidence interval” →
- Across the 100 replications, replication effects were on average half the magnitude of the original effects: 97% of original studies reported statistically significant results but only 36% of replications did, and only 39% of effects were subjectively rated as having replicated.Brian Nosek on the Reproducibility Project
- GiveWell's 'Change Our Mind' contest, offering prizes to people who found errors in its public analysis, drew more than 50 high-quality submissions including from prestigious academics, surfaced small errors across programs, and—most importantly—revealed GiveWell had been insufficiently transparent about the uncertainty inherent in its analysis.Elie Hassenfeld on GiveWell
- Decisions under uncertainty should be based on payoffs and judgment rather than true-false statements, because statistics such as a 95 percent confidence interval treat truth as binary, while the ancients treated probability as something one approves of regardless of the odds.Nassim Taleb on the Financial Crisis 03/23/2009
- The FDA had set the approval bar at 50% efficacy with sufficiently narrow confidence intervals, but the mRNA vaccine trials delivered 95% efficacy—a level matched by only one other vaccine in history.256 - A Contagion Of Bad Ideas
- When the subjective wellbeing data in the single long-term deworming study is taken literally, deworming shows no clear positive effect on well-being and could even appear negative, though the confidence intervals are too wide to detect any substantial effect—meaning there is not strong evidence that this widely-praised intervention improves well-being.446 - How To Do The Most Good
- Because there is no single unambiguous way to define a successful replication, the project used five different criteria—including whether the replication reached statistical significance (p<.05), whether the original effect size fell in the replication's 95% confidence interval, effect-size comparison, subjective ratings, and combined significance—so the same result can count as a success on some criteria and a failure on others.Brian Nosek on the Reproducibility Project
- In the WHO 2000 World Health Report ranking, the differences between countries ranked number 3 and number 30 were not statistically significant -- meaning the difference could be explained by random variation -- yet they were still reported as ordered rankings, which misleads the public.Scott Atlas on American Health Care
- Companies frequently conclude that marketing campaigns did or did not work based on flawed testing, such as too-small sample sizes, inconsistent placement, or poor copy, so the conclusions are often unreliable.Rebecca Lynn - Finding Your Customers - [Invest Like the Best, EP.309]
- Monthly payroll reports should not be trusted as reliable signals of economic direction because they contain large confidence intervals and have become less reliable due to QCW and regular benchmark revisions that alter results months after initial publication.The "Profound" Impact of Rate Cuts Have BEGUN
- Hereditarian critics argued that using later-vintage data yields a heritability confidence interval including negative values, claiming this is why it could not be published—but a negative heritability is meaningless and simply indicates the study does not work, which is a fact that should be disclosed rather than hidden; also, swapping in a different vintage of control data violates the logic of a control arm, which requires the same vintage as the treatment data.Nature, Nurture, and Identical Twins (with David Bessis)