An example of the result you get: a score, your time, and each skill area marked.
What the test covers
Statistics questions for analyst, data and research roles, and for a first statistics course, keep to the same ground: summarize a sample, work a probability, use a distribution, test a claim, fit a line. The test takes two problems from each of the five levels of Statisticsdrill, picked at random and with new data each time.
Describing data
Mean, median and standard deviation of a small sample, z-scores, percentiles, the interquartile range and outliers. x̄, s, z-score, IQR
Binomial, Poisson, normal, uniform and exponential probabilities, sampling distributions and the standard error. Binomial, Poisson, normal, standard error
Confidence intervals, sample size, t, z and chi-square tests, ANOVA and power. Hypothesis tests ask for the number and the decision. t test, p-value, confidence interval, chi-square
These are problem types from the same pool as the test. In the test each one comes with different data, and you type one number into an answer box. On a hypothesis test you also choose between Reject H0 and Fail to reject H0.
Q1Sample standard deviation
Describing data
A random sample of 6 minutes to resolve a support ticket (minutes): 46, 35, 28, 23, 26, 17. What is the sample standard deviation? Round to 2 decimal places.
minutes
1
46
2
35
3
28
4
23
5
26
6
17
Show the answer
Formulas = √( Σ(x - x̄)² / (n - 1) )
Step 2x̄ = 175 / 6 = 29.1667
Step 3Σ(x - x̄)² = 514.8333
Step 4s² = 514.8333 / 5 = 102.9667
Step 5s = √102.9667 = 10.1472
Excel=STDEV.S(A2:A7)
Answer10.15
STDEV.P would divide by n instead, which is only right when the data is the whole population.
Suppose a screening test checks patients for a condition. 0.5% of patients are affected. The screen flags 85% of affected patients (sensitivity) and correctly clears 99% of the others (specificity). A patient is flagged. What is the probability that this patient is really affected? Round to 4 decimal places.
Test H0: μ = 68 against H1: μ ≠ 68 for the average order value (dollars). A random sample of 11 gives x̄ = 69.33 and s = 29.5. Find the p-value of the one-sample t test, then decide at α = 0.05. Round to 4 decimal places.
Quantity
Value
1
μ0
68
2
x̄
69.33
3
s
29.5
4
n
11
5
α
0.05
Show the answer
Formulat = (69.33 - 68) / (29.5 / √11) = 0.1495
Step 2df = 10
Step 3p = 2 × P(T > |t|) = 2 × 0.442054 = 0.8841
Step 4p = 0.8841 is above α = 0.05, so fail to reject H0.
Excel=T.DIST.2T(ABS(0.149529), 10)
Answer0.8841
DecisionFail to reject H0
Find t = (x̄ - μ0) / (s / √n), then the two-tailed area beyond |t| with n - 1 degrees of freedom.
Each question is worth one point. Your number is checked on the server against the exact answer for the data you were given, rounded as the question states, with a small tolerance in the last decimal place. Counts must be exact.
On a hypothesis test both parts must be right: the statistic or p-value, and the decision to reject or fail to reject H0 at the stated α.
You get two checks per question. A check only says right or wrong: there is no hint about the direction or the likely slip during the test. After a second wrong check the question is marked wrong and the test moves on. A question you skip, or have not solved when the 15 minutes end, counts as wrong too. The result shows your score out of 10, your time, and each topic as strong or to review.
What your score means
0
1
2
3
4
5
6
7
8
9
10
0 to 4Foundations first
Start with Describing data and Probability, then come back. Distributions and Inference build on them.
5 to 7Working knowledge
The core is there. The topics you missed are the ones to drill before a timed statistics test.
8 to 10Interview-ready on this set
You solved problems from nearly every level under the clock. Drill the one or two topics you missed and you are in good shape.
These bands describe this test only. Employers and schools set their own pass marks, and we do not know them.
Take the test
10 questions from 5 skill areas, 15 minutes on the clock. It starts right here, with no account and no email.
Hints and solutions are off while the clock runs.
Two checks per question, right or wrong only. A second wrong check or a skip counts as wrong.
Your answers survive a page reload.
The mistakes that cost the most points
Most wrong answers in statistics come from a short list of slips. Check for them before a timed test.
Dividing by n instead of n - 1. A sample standard deviation divides by n - 1; dividing by n is only right when the data is the whole population.
One tail when the test has two. For H1 with a not-equal sign the p-value is the area in both tails, so it is twice the one-tail area.
Treating a large p-value as proof. A p-value above α means you fail to reject H0. It does not show that H0 is true.
Ignoring the base rate. When a condition is rare, most positive screening results are false alarms even with an accurate test. Bayes' rule needs the prevalence.
Using z where t belongs. With a small sample and an unknown population standard deviation, the critical value comes from the t distribution with n - 1 degrees of freedom.
The formats a statistics test for an interview comes in
Concept questions: what a p-value means, when a confidence interval is valid, how to explain a regression slope.
Worked problems with a number at the end, usually timed. This test is that format, because knowing a definition and getting the number under a clock are different skills.
A take-home analysis in a spreadsheet, R, Python or SQL. The same statistics, with the data cleaning and the write-up added.
A/B test cases for analyst and product roles: sample size, lift, significance. The Inference and Relationships levels cover these.
Questions
Is the statistics practice test really free?
Yes. The test is free and you do not need a card. Free visitors get one attempt a day; Pro members can retake it any time.
Do I need an account, tables or software?
No account is needed; the test starts in one click. You do need something to compute with: a calculator for most questions, and a spreadsheet, statistical calculator or tables for p-values and critical values.
How long does the test take?
Up to 15 minutes. There are 10 questions on one clock, and the test ends when you finish the last question or the time runs out.
How is the test scored?
One point per question, 10 in total. Your number is checked on the server against the exact answer for your data, with a small tolerance for rounding, and on hypothesis tests the reject or fail-to-reject decision must be right too. You get two checks per question and a check only says right or wrong; after a second wrong check the question is marked wrong and the test moves on. Skipped or unsolved questions count as wrong. You also get a breakdown by topic.
Is it like the statistics test employers use?
It asks the kinds of worked problems that statistics tests for analyst and data roles use, under a clock. It is not a copy of any employer's or testing company's test, and it has no multiple-choice concept questions, coding tasks or case write-ups.
Will I see the same problems twice?
Not with the same data. Every problem type is a template that generates a new sample or new inputs on each attempt, so a retake is a different test even when a problem type repeats.
Can I retake the test?
Yes. Free visitors can retake it the next day; Pro removes the limit.
What comes after the test?
Your result links straight to the problem types for the topics you missed. Your first 15 tasks on Statisticsdrill are free without an account, and questions you solve in the test count toward them. Pro unlocks all 87 problem types, every reference solution, timed mode and the certificate.
Find out where you stand
Ten problems, 15 minutes, a score by topic. Free, and no sign-up to start.
Statisticsdrill