If, in a series. there are m items whose ranks are common, then which factor is added for correction for each repeating value in both the series?

Question

Prepp Team · Accepted Answer

Understanding Tied Ranks in Rank Correlation

In statistics, when we calculate rank correlation (like Spearman's rank correlation coefficient) between two variables, we assign ranks to the observations for each variable. Ideally, each observation gets a unique rank. However, sometimes two or more observations have the same value. In such cases, these observations are given the same rank, which is usually the average of the ranks they would have received individually if there were no ties. This situation is known as having 'tied ranks' or 'common ranks'.

When tied ranks occur, the standard formula for rank correlation needs a correction because the presence of ties affects the sum of the squared differences between ranks. The question specifically asks about the correction factor added for each repeating value (tied group) in the series when calculating rank correlation, and it mentions 'm items whose ranks are common'. Here, 'm' represents the number of items or observations that share the same rank within a particular tie group.

Correction Factor Formula for Tied Ranks

For each group of tied ranks, a correction factor is calculated based on the number of observations within that group. If there are 'm' items that have the same rank, the standard correction factor added for this specific group of tied ranks is given by a formula involving 'm'.

Let's examine the options provided for the correction factor for a group of 'm' tied ranks:

$\dfrac{1}{12}m$
$\dfrac{1}{12}m(m-1)(m+1)$
$\dfrac{1}{12}(m-1)(m+1)$
$\dfrac{1}{12}m(m-1)$

The standard formula for the correction factor for a group of 'm' tied observations is:

$\dfrac{m^3 - m}{12}$

We can factor the numerator:

$m^3 - m = m(m^2 - 1)$

Using the difference of squares formula ($a^2 - b^2 = (a-b)(a+b)$), we can factor $m^2 - 1$ as $(m-1)(m+1)$.

So, the numerator becomes $m(m-1)(m+1)$.

Therefore, the correction factor formula for a group of 'm' tied ranks is:

$\dfrac{m(m-1)(m+1)}{12}$

This formula is added to the sum of squared differences in the numerator of the Spearman's rank correlation coefficient formula when ties are present. Each group of tied ranks contributes this value to the total correction.

Analyzing the Options

Comparing the derived formula with the given options:

Option 1: $\dfrac{1}{12}m$ - This is not the correct correction factor.
Option 2: $\dfrac{1}{12}m(m-1)(m+1)$ - This matches the derived formula $\dfrac{m(m-1)(m+1)}{12}$.
Option 3: $\dfrac{1}{12}(m-1)(m+1)$ - This is missing the 'm' term in the numerator.
Option 4: $\dfrac{1}{12}m(m-1)$ - This is not the correct factor.

Thus, the factor added for correction for each repeating value (tied group of 'm' items) in both the series is $\dfrac{1}{12}m(m-1)(m+1)$.

Conclusion on Tied Ranks Correction

When dealing with rank correlation and you encounter groups of observations with common ranks (tied ranks), a correction needs to be applied to the formula. For each group of 'm' tied items, the correction factor $\dfrac{m(m-1)(m+1)}{12}$ is calculated and summed up for all tie groups in both series. This total correction factor is then used in the adjusted rank correlation formula.

Term	Description
Rank Correlation	A non-parametric measure of association between two variables based on their ranks.
Tied Ranks	When two or more observations have the same value and are assigned the same rank (usually the average rank).
Correction Factor	A value added to the calculation to account for the effect of tied ranks on the correlation coefficient.
m	The number of observations within a specific group of tied ranks.

Revision Table: Rank Correlation and Tied Ranks

Concept	Key Point	Formula/Method
Spearman's Rank Correlation ($\rho$)	Measures monotonic relationship.	$\rho = 1 - \dfrac{6 \sum d^2}{n(n^2-1)}$ (without ties)
Dealing with Tied Ranks	Assign average rank to tied values. Apply correction factor.	For a tie group of size 'm', correction is $\dfrac{m(m^2-1)}{12}$ or $\dfrac{m(m-1)(m+1)}{12}$.
Adjusted Spearman's $\rho$ (with ties)	Involves summing correction factors for all tie groups.	More complex formula involving the correction factor in the denominator or adding it to $\sum d^2$.

Additional Information on Rank Correlation Calculations

When calculating Spearman's rank correlation with ties, the formula is adjusted. One common adjustment involves modifying the denominator. The formula becomes:

$\rho = \dfrac{\sum x^2 + \sum y^2 - \sum d^2}{2 \sqrt{\sum x^2 \sum y^2}}$

Where $\sum d^2$ is the sum of squared differences between ranks, $\sum x^2 = \dfrac{n^3-n}{12} - C_x$, and $\sum y^2 = \dfrac{n^3-n}{12} - C_y$.

$C_x$ is the sum of correction factors for all tied groups in the first series, and $C_y$ is the sum of correction factors for all tied groups in the second series. For each group of 'm' tied values in a series, the correction factor is $\dfrac{m(m^2-1)}{12}$.

Alternatively, a simpler approach for calculation is to calculate the correlation coefficient using the standard formula $\rho = 1 - \dfrac{6 \sum d^2}{n(n^2-1)}$ and compare it to the coefficient calculated using the adjusted method with corrections. The correction factors ensure that the denominator properly reflects the reduced variability in ranks due to ties.

The correction factor $\dfrac{m(m^2-1)}{12}$ comes from considering the variance of ranks. When there are no ties, the sum of squares of ranks has a certain value. When there are ties, this sum of squares is reduced, and the correction factor accounts for this reduction in variance introduced by the ties.

Prepp Team · Answer

$\dfrac{1}{12}m$

Prepp Team · Answer

$\dfrac{1}{12}(m-1)(m+1)$

Prepp Team · Answer

$\dfrac{1}{12}m(m-1)$

Mathematics	Statistics
3	6
5	4
8	9
4	8
7	1
10	2
2	3
1	10
6	5
9	7