Kendall's Tau Test

From ZCubes Wiki
Jump to navigation Jump to search
KENDALLSTAUTEST(Range1, Range2, alpha, NewTableFlag)
  • is the array of x values.
  • is the array of y values.
  • is the value from 0 to 1.
  • is either TRUE or FALSE. TRUE for getting results in a new cube. FALSE will display results in the same cube.

DESCRIPTION

  • It is a statistic test used to measure the ordinal association between two measured quantities.
  • It is a measure of rank correlation: the similarity of the orderings of the data when ranked by each of the quantities.
  • Kendall correlation between two variables will be high when observations have a similar rank.
  • It will be low when observations have a dissimilar rank between the two variables.

Let (x1, y1), (x2, y2), …, (xn, yn) be a set of observations of the joint random variables X and Y respectively, such that all the values of and are unique.

concordant if  &  or  & 
discordant if  &  or Failed to parse (MathML with SVG or PNG fallback (recommended for modern browsers and accessibility tools): Invalid response ("Math extension cannot connect to Restbase.") from server "https://wikimedia.org/api/rest_v1/":): {\displaystyle (x_i < x_j)}
 & Failed to parse (MathML with SVG or PNG fallback (recommended for modern browsers and accessibility tools): Invalid response ("Math extension cannot connect to Restbase.") from server "https://wikimedia.org/api/rest_v1/":): {\displaystyle (y_i > y_j)}

neither if Failed to parse (MathML with SVG or PNG fallback (recommended for modern browsers and accessibility tools): Invalid response ("Math extension cannot connect to Restbase.") from server "https://wikimedia.org/api/rest_v1/":): {\displaystyle (x_i = x_j)}
 or Failed to parse (MathML with SVG or PNG fallback (recommended for modern browsers and accessibility tools): Invalid response ("Math extension cannot connect to Restbase.") from server "https://wikimedia.org/api/rest_v1/":): {\displaystyle (y_i = y_j)}
 (i.e. ties are not counted).

The Kendall's Tau statistic is:

Failed to parse (MathML with SVG or PNG fallback (recommended for modern browsers and accessibility tools): Invalid response ("Math extension cannot connect to Restbase.") from server "https://wikimedia.org/api/rest_v1/":): {\displaystyle \tau=1-\frac{4D}{n(n-1)}} .
  • C is the number of concordant pairs.
  • D is the number of discordant pairs.

RESULT

  • If number of values in a set is <15, critical tables are used to calculate Failed to parse (MathML with SVG or PNG fallback (recommended for modern browsers and accessibility tools): Invalid response ("Math extension cannot connect to Restbase.") from server "https://wikimedia.org/api/rest_v1/":): {\displaystyle \tau} .
  • If number of values in a set is >=15, Normal approximation is used for calculation.
* If Failed to parse (MathML with SVG or PNG fallback (recommended for modern browsers and accessibility tools): Invalid response ("Math extension cannot connect to Restbase.") from server "https://wikimedia.org/api/rest_v1/":): {\displaystyle \tau}
 > critical value from the Kendall's Tau Critical table, then reject the null hypothesis that there is no correlation.
* else if, Failed to parse (MathML with SVG or PNG fallback (recommended for modern browsers and accessibility tools): Invalid response ("Math extension cannot connect to Restbase.") from server "https://wikimedia.org/api/rest_v1/":): {\displaystyle \tau}
 <critical value, correlation exists.

EXAMPLE

Spreadsheet
A B
1 80 5
2 78 23
3 60 25
4 53 48
5 85 17
6 84 8
7 73 4
8 79 26
9 81 11
10 75 19
11 68 14
12 72 35
13 58 29
14 92 3
15 65 24

=KENDALLSTAUTEST(A1:A15,B1:B15, 0.05, true)

KENDALL'S TAU TEST USING NORMAL APPROXIMATION
RANGE1 SORT RANGE2 SORT CONCORDANT DISCORDANT
53 48 0 14
58 29 1 12
60 25 2 10
65 24 2 9
68 14 5 5
72 35 0 9
73 4 7 1
75 19 2 5
78 23 1 5
79 26 0 5
80 5 3 1
81 11 1 2
84 8 1 1
85 17 0 1
92 3 0 0
SUMMARY
VARIABLE RESULT
COUNT 15
C 105
SUM CONCORDANT 25
SUM DISCORDANT 80
KENDALL'S TAU -0.52381
STDERROR 0.19245
Z-VALUE -2.7218
ZCRITICAL 1.95996
P-VALUE 0.00649
RESULT REJECT NULL HYPOTHESIS, NO CORRELATION
  • CONCLUSION: REJECT NULL HYPOTHESIS, NO CORRELATION