Correlation

Kendall's Tau Table: Exact Critical Values of τ

Exact critical values of Kendall's tau, the rank correlation based on concordant and discordant pairs.

  • n = 4 to 100
  • Exact values
  • Exact calculator

Kendall's tau Table

Rows: sample size n · Columns: α (one-tailed on the first header row, two-tailed on the second) · Cells: critical τ. Reject H₀ if |τ| ≥ the cell.

Kendall's Tau Table (Critical Values)
One-tailed α →0.050.0250.010.0050.001
Two-tailed α →
n ↓
0.10.050.020.010.002
41.000————
50.8001.0001.000——
60.7340.8670.8671.000—
70.6200.7150.8100.9051.000
80.5720.6430.7150.7860.858
90.5000.5560.6670.7230.834
100.4670.5120.6000.6450.778
110.4190.4910.5640.6000.710
120.3940.4550.5460.5760.667
130.3590.4360.5130.5650.642
140.3630.4070.4730.5170.605
150.3340.3910.4670.5050.581
160.3170.3840.4340.4840.567
170.3090.3680.4270.4710.545
180.2950.3470.4120.4510.530
190.2870.3340.3920.4390.509
200.2740.3270.3790.4220.495
210.2670.3150.3720.4100.486
220.2650.3080.3600.3940.472
230.2570.2970.3520.3920.455
240.2470.2900.3410.3770.450
250.2400.2870.3340.3670.440
260.2370.2800.3300.3600.428
270.2310.2710.3220.3570.419
280.2280.2650.3130.3440.413
290.2220.2620.3110.3400.404
300.2190.2560.3020.3340.394
310.2130.2520.2950.3250.390
320.2100.2460.2910.3230.380
330.2050.2430.2880.3150.375
340.2020.2380.2800.3120.369
350.1970.2340.2780.3050.362
360.1940.2320.2740.3020.359
370.1930.2290.2680.2980.352
380.1900.2240.2640.2920.346
390.1880.2200.2610.2880.342
400.1850.2180.2570.2850.339
410.1810.2150.2540.2810.335
420.1780.2130.2500.2760.329
430.1770.2100.2470.2740.325
440.1740.2080.2440.2690.322
450.1720.2050.2410.2670.318
460.1700.2020.2390.2640.315
470.1680.1990.2360.2600.310
480.1670.1970.2330.2580.307
490.1640.1960.2300.2540.303
500.1630.1920.2280.2510.300
510.1610.1910.2260.2490.298
520.1590.1890.2240.2460.295
530.1570.1880.2210.2440.291
540.1560.1860.2190.2420.288
550.1550.1830.2170.2400.285
560.1520.1810.2150.2370.282
570.1520.1800.2120.2350.280
580.1500.1780.2100.2320.277
590.1480.1760.2090.2300.275
600.1470.1750.2070.2290.273
650.1410.1680.1990.2190.261
700.1360.1620.1910.2100.252
750.1310.1560.1840.2030.243
800.1260.1500.1780.1970.235
850.1230.1460.1720.1900.227
900.1190.1410.1670.1850.221
950.1150.1370.1620.1800.215
1000.1120.1340.1580.1750.209

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

How to read this table

  1. Count the concordant (C) and discordant (D) pairs. S = C − D and τ = S / (n(n − 1)/2).
  2. Find the row for n and the column for α.
  3. The association is significant if |τ| is greater than or equal to the table value. To work with S instead, multiply the table value by n(n − 1)/2.

Worked example

Ten patients are ranked by two raters, giving τ = 0.56 (S = 25). Row 10, two-tailed α = 0.05 gives 0.512, so the agreement is significant at the 5% level.

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

τ = (C − D) / (n(n − 1)/2)  ·  Var(S) = n(n − 1)(2n + 5)/18 under H₀

Values come from the exact null distribution of the number of discordant pairs, which is the distribution of inversions in a random permutation, computed by recursion for every n. Each entry is the smallest attainable τ with P(τ ≥ t) ≤ α.

Frequently asked questions

Kendall's τ or Spearman's rₛ?

Both test the same hypothesis and usually agree. τ is smaller in magnitude (roughly two-thirds of rₛ), has a direct interpretation as a difference between the probabilities of concordance and discordance, and handles ties more gracefully.

What if there are ties?

Use τ-b, which adjusts the denominator for ties. The exact table assumes no ties, so use the normal approximation with a tie-corrected variance when ties are common.

Related calculators and tutorials