Blog / Why Base Rates Change What a Positive Assessment Threshold Means

Why Base Rates Change What a Positive Assessment Threshold Means

A threshold’s positive predictive value depends on who it is applied to, because the underlying frequency of the target condition changes the mix of true and false positives.

October 21, 2026 · 6 min read · By Cepingku Editorial Team

Any practice that cuts a continuous score into ‘meets a threshold’ and ‘does not meet a threshold’ can create a misleading shortcut. Sensitivity asks, among people for whom a target condition is truly present, what proportion receives a positive result. Positive predictive value (PPV) asks a different question: among positive results, what proportion truly has the target condition? The denominators differ, so one number cannot replace the other.

FDA’s Statistical Guidance on Reporting Results from Studies Evaluating Diagnostic Tests defines PPV as 100 × TP/(TP + FP): the proportion of true positives among all positive results. It defines sensitivity as 100 × TP/(TP + FN), specificity as 100 × TN/(FP + TN), and prevalence as the frequency of a condition in a specified group. Those are diagnostic-test definitions. The fictional tables below use their 2×2 arithmetic to explain conditional probability; they do not treat this site’s self-reflection assessments as diagnostic tests.

A fully fictional 1-percent base-rate table

Imagine 1,000 fictional participants. For teaching only, suppose a binary threshold has 80 percent sensitivity and 90 percent specificity for a fictional condition that can be checked independently. Let the condition’s base rate be 1 percent. Ten people have the condition and 990 do not. Among the ten, 10 × .80 = 8 are true positives and 2 are false negatives. Among the 990 without the condition, 990 × .90 = 891 are true negatives and 990 − 891 = 99 are false positives.

Fictional condition statusThreshold results
Condition present8 true positives; 2 false negatives; total 10.
Condition absent99 false positives; 891 true negatives; total 990.
All 1,000 participants107 positive results and 893 negative results.

Sensitivity in this table is 8 ÷ 10 = 80 percent. But PPV is 8 ÷ (8 + 99) = 8 ÷ 107, or about 7.48 percent. Of the 107 positive labels, eight are true positives. This does not make the stated sensitivity wrong; it is answering a different conditional question. With a low base rate, many people do not have the condition, so even a 10 percent false-positive rate among them produces 99 false positives.

Change only the base rate

Keep the same fictional sensitivity of 80 percent and specificity of 90 percent, again among 1,000 people, but change the base rate to 10 percent. There are now 100 people with the condition and 900 without it. The table has 80 true positives, 20 false negatives, 810 true negatives, and 90 false positives. PPV becomes 80 ÷ (80 + 90) = 80 ÷ 170, or about 47.06 percent.

Fixed fictional performanceBase rate, true positives, false positives, and PPV
80% sensitivity; 90% specificity1% base rate: 8 TP, 99 FP, PPV 7.48%.
80% sensitivity; 90% specificity10% base rate: 80 TP, 90 FP, PPV 47.06%.

The assumed threshold performance is identical in both tables. Only the frequency of the condition in the population changed. FDA’s public information on antibody testing likewise explains that predictive values are calculated from sensitivity, specificity, and an assumed prevalence, and that lower prevalence lowers PPV in that testing context. That does not make the figures above a health recommendation. It identifies the base rate as a condition of interpretation whenever a binary test claim is made.

What a threshold claim needs

Before accepting a claim about a positive threshold, ask what independently establishes the target condition; where the threshold was set; which population actually received it; what the base rate is in that population; and whether the full 2×2 counts and uncertainty are available. Without those facts, a ‘positive’ or ‘high’ label cannot establish the probability that an individual has a characteristic, and it cannot serve as a screening or major-decision label. A table from one population also does not automatically apply after the population, language, use, or base rate changes.

Career Decision Readiness is a self-reflection entry point for exploring how a person gathers information, makes choices, and turns a possible direction into next steps. The site does not claim a validated binary decision threshold, reference standard, sensitivity, specificity, base rate, or predictive value for this assessment. This article does not make it suitable for hiring, admission, diagnosis, treatment, or major career decisions.

Want to see where you land?

Take the Career Decision Readiness and get your results in minutes.

Take the Career Decision Readiness →

← Back to the blog