Practice Hidden Variables in Math

Use these practice problems to test your method after reviewing the concept explanation and worked examples.

Quick Recap

Quantities or factors that influence a mathematical or real-world situation but are not explicitly included in the current model or expression.

What's lurking behind the scenes that we forgot to account for?

Showing a random 20 of 50 problems.

Example 1

medium
A linear fit of y on x has slope 2. After adding an omitted variable z (correlated with both), the slope of x drops to 0.5. What does this reveal about the original model?

Example 2

challenge
In a regression y = b0 + b1 x, the true model is y = b0 + b1 x + b2 z + e, with z = c x + u. Show that omitting z makes the estimated x-coefficient converge to b1 + b2 c, and state the condition under which omission causes no bias.

Example 3

easy
A store finds taller shelves sell more, so stocks taller shelves. Sales drop. What hidden variable might have driven the original pattern?

Example 4

challenge
Across 5 years, a treatment's yearly success rate exceeds control's every year, yet pooled over all years control wins. Construct the hidden-variable condition (counts) that makes this possible and name the phenomenon.

Example 5

challenge
Two regions show the same average household income $50,000\$50{,}000, but region A has a Gini of 0.250.25 and region B has 0.550.55. What hidden variable does a single mean conceal?

Example 6

medium
A linear model y=3xy = 3x holds in summer. In winter, the same data show y=1.5xy = 1.5x. Identify the hidden variable and write the augmented model.

Example 7

medium
Region A and Region B both show a hospital treatment with 80% survival overall, yet within both mild and severe cases Hospital X beats Hospital Y. Which hidden variable produces this reversal (Simpson's paradox)?

Example 8

easy
The area formula A=lwA = lw has hidden units. If l=5l = 5 m and w=3w = 3 m, what is AA, and what hidden variable (units) must be tracked?

Example 9

easy
Fill in the blank: A variable that influences a relationship but is not included in the model is called a ____ variable.

Example 10

medium
A simple regression y=a+bxy = a + bx fits perfectly on training data with b=4b = 4. Adding a measured variable zz yields y=aโ€ฒ+0.5x+3.5zy = a' + 0.5 x + 3.5 z with z=xz = x. Explain what was hidden.

Example 11

hard
A regression of wages on years of education produces slope 0.100.10. After adding IQ, the slope drops to 0.060.06. Compute the share of the original slope attributable to IQ-correlated effects.

Example 12

easy
A formula gives a car's stopping distance as d=0.044v2+0.75vd = 0.044v^2 + 0.75v (metres, km/h). Identify the hidden variables that this formula ignores.

Example 13

medium
Two students score identically on a test, but one cheated. A model using only score predicts equal ability. What hidden variable invalidates the prediction?

Example 14

easy
A formula models a falling object's time using only height, ignoring air. For a feather it fails badly. What hidden variable matters?

Example 15

hard
A disease has prevalence 2%2\%; a test has sensitivity 95%95\% and specificity 90%90\%. Compute the posterior probability of disease given a positive test.

Example 16

medium
A 2x2 table compares treatment A vs B by gender. Within each gender A wins; pooled, B wins. Which hidden variable produces the reversal?

Example 17

medium
A regression coefficient changes when a new predictor is added. What does this typically indicate?

Example 18

medium
A test for a rare disease (1% prevalence) is 90% accurate. A patient tests positive. The 'hidden' factor inflating false positives is base rate. What is the approximate probability the patient actually has the disease?

Example 19

medium
A company finds employees with bigger offices earn more, and concludes office size raises pay. Name the most likely hidden variable and the true causal direction.

Example 20

hard
A test reports drug A is better in adults and in children separately, but worse overall. Construct counts (adult 90/100 vs 60/100, child 10/100 vs 30/100 for A vs B) and confirm.