Math 2.0

Statistics Definition

Wilcoxon signed-rank test

Builds on Statistical hypothesis testIndependent and identically distributed random variablesBernoulli distribution

Definition

Let Z1,…,ZnZ_1, \dots, Z_n be i.i.d. real random variables with a continuous distribution that is symmetric about a number θ\theta, that is, Z1−θZ_1 - \theta and θ−Z1\theta - Z_1 have the same law. Let RiR_i be the rank of ∣Zi∣\abs{Z_i} among ∣Z1∣,…,∣Zn∣\abs{Z_1}, \dots, \abs{Z_n}, so that the smallest has rank 11. The signed-rank statistic is W+=∑i : Zi>0Ri.W^+ = \sum_{i \,:\, Z_i > 0} R_i. The Wilcoxon signed-rank test of H0 ⁣:θ=0H_0 \colon \theta = 0 rejects H0H_0 when W+W^+ is too large, too small, or either, according to the alternative. Under H0H_0, W+W^+ has the law of ∑k=1nkBk\sum_{k=1}^{n} k B_k, where B1,…,BnB_1, \dots, B_n are independent Bernoulli variables with parameter 1/21/2.

Remarks

Under H0H_0, E[W+]=n(n+1)/4E[W^+] = n(n+1)/4 and Var⁡(W+)=n(n+1)(2n+1)/24\operatorname{Var}(W^+) = n(n+1)(2n+1)/24, and W+W^+ is asymptotically normal. For matched pairs (Xi,Yi)(X_i, Y_i) one takes Zi=Yi−XiZ_i = Y_i - X_i.

Example: Z=(1.2,−0.4,2.5)Z = (1.2, -0.4, 2.5). The absolute values 1.2,0.4,2.51.2, 0.4, 2.5 have ranks 2,1,32, 1, 3, so W+=2+3=5W^+ = 2 + 3 = 5. Under H0H_0 each of the 88 sign patterns has probability 1/81/8, and only the sets of ranks {2,3}\set{2, 3} and {1,2,3}\set{1, 2, 3} give a sum at least 55. So the one-sided p-value is P(W+≥5)=2/8=1/4P(W^+ \ge 5) = 2/8 = 1/4.

Non-example: if the ZiZ_i have median 00 but a skewed law, the clause on symmetry fails. Then the null law above is wrong, and the test can reject often even though the median is 00; the sign test is valid in that case.

Inverting the test gives the Hodges–Lehmann estimator of θ\theta.

Reading path

To understand this page from the ground up, read these 56 pages in order. The foundations come first.

  1. Construction of the reals from Cauchy sequences concept
  2. Element definition
  3. Integers object
  4. Ordered field definition
  5. Proposition concept
  6. Random experiment concept
  7. Rational numbers object
  8. Real numbers object
  9. Set definition
  10. Ring of integers statement
  11. Disjunction definition
  12. Implication definition
  13. Predicate definition
  14. Bernoulli trial concept
  15. Ordered pair definition
  16. Union definition
  17. Sample space definition
  18. The rationals form a field statement
  19. Existential quantifier definition
  20. Universal quantifier definition
  21. Cartesian product definition
  22. Subset definition
  23. The rationals form an ordered field statement
  24. Uniqueness quantifier definition
  25. Binary relation definition
  26. Universal set definition
  27. Order on the reals definition
  28. Power set definition
  29. Complement definition
  30. Function definition
  31. Interval definition
  32. Injective function definition
  33. Preimage of a set definition
  34. Sigma-algebra definition
  35. Countable set definition
  36. Event definition
  37. Borel sigma-algebra definition
  38. Probability axioms concept
  39. Probability measure definition
  40. Probability distribution definition
  41. Probability space definition
  42. Conditional probability definition
  43. Random variable definition
  44. Discrete random variable definition
  45. Law of a random variable definition
  46. Independence of two events definition
  47. Probability mass function definition
  48. Identically distributed random variables definition
  49. Mutual independence of events definition
  50. Independent sigma-algebras definition
  51. Bernoulli distribution definition
  52. Independence of random variables definition
  53. Independent and identically distributed random variables definition
  54. Statistic definition
  55. Statistical hypothesis test concept
  56. Wilcoxon signed-rank test definition

Further reading

  1. Web https://en.wikipedia.org/wiki/Wilcoxon_signed-rank_test
About this page
Tag9RPYB7S4HM
URIm2://9RPYB7S4HM
Kinddefinition
WritingW0 stub
MSC62G10
DefinesWilcoxon signed-rank test
UsesStatistical hypothesis test, Independent and identically distributed random variables, Bernoulli distribution

Markdown · JSON · Edit on GitHub

What this page needs for W1

  • a source with a locator (0 of 1)

The writing levels say how complete the text of a page is. They do not measure its truth.

What links here

No page links to this page yet.