Nonparametric Tests

Page's L Test Table: Exact Critical Values

Exact critical values for Page's trend test, used for ordered alternatives in repeated-measures designs.

  • k = 3 to 8
  • n = 2 to 20 blocks
  • Exact calculator

Page's L Table

Jump to a significance level below. Rows: number of blocks n · Columns: number of treatments k · Cells: critical L. Reject H₀ if L ≥ the cell.

α = 0.05 (one-tailed)

α = 0.05 (one-tailed)
Number of treatments (k) →
n (blocks) ↓
345678
22858103166252362
34184150244370532
454111197321487701
566137244397603869
6791632914747191037
7911893385508351204
81042143846259501371
911624043170110651537
1012826647777711811704
1114129252385212951870
1215331757092814102036
13165343616100315252201
14178369662107816392367
15190394708115317542532
16202420754122918682697
17215446800130419832863
18227471846137920973028
19239497892145422113193
20251522938152923253358

α = 0.01 (one-tailed)

α = 0.01 (one-tailed)
Number of treatments (k) →
n (blocks) ↓
345678
2—60106173261375
34287155252382550
455114204331501722
568141251409620893
6811672994867371063
7931933465638551232
81062203936409721401
911924644171710881569
1013127248779312051737
1114429853486913211905
1215632458194614372072
13169350628102215522240
14181376674109816682407
15194402721117417842573
16206428767124918992740
17218453814132520142907
18231479860140121293073
19243505906147722443239
20256531953155223603406

α = 0.001 (one-tailed)

α = 0.001 (one-tailed)
Number of treatments (k) →
n (blocks) ↓
345678
2——109178269388
3—89160260394567
456117210341516743
570145259420637917
6831723074997571090
7961983555778761262
81092254036569951433
912125245173311131603
1013427849981112311774
1114730554688813481943
1216033159396514652113
13172357641104215832282
14185384688111916992450
15197410735119618162619
16210436782127219332787
17223462829134920492955
18235488876142521653123
19248514922150222813291
20260540969157823983459

Tip: click any value to highlight its row and column.

How to read this table

  1. Rank the k treatments within each block, with treatments listed in the hypothesised order.
  2. Add up the ranks Rⱼ for each treatment and compute L = Σ j·Rⱼ, where j is the treatment's position in the order.
  3. Reject H₀ if L is greater than or equal to the table value.

Worked example

Five subjects each receive four increasing doses, and the rank sums in dose order give L = 137. Row n = 5, column k = 4 of the α = 0.05 table gives 137, so the ordered trend is significant.

Click the underlined links to highlight the value in the table.

Calculator

For values between the rows and columns of the table. Results update as you type.

Formula and how the values were computed

L = Σⱼ j·Rⱼ  ·  E(L) = nk(k + 1)²/4  ·  Var(L) = nk²(k + 1)(k² − 1)/144

Values come from the exact null distribution of L, obtained by enumerating the k! orderings within one block and convolving over n independent blocks. Each entry is the smallest L with P(L ≥ l) ≤ α.

Frequently asked questions

How is Page's test related to Friedman's?

Both rank treatments within blocks. Friedman's test looks for any difference; Page's test concentrates its power on one specified ordering.

What if my n or k is not in the table?

The calculator gives exact p-values for k up to 8 and up to 60 blocks, and a normal approximation beyond that.

Related calculators and tutorials