Explaining: Spearman's Rank Correlation Coefficient

Рет қаралды 46,694

Geographer Online

Күн бұрын

Пікірлер: 15

@GlynisDance Жыл бұрын

Thank you - your explanation was completely clear and fresh, even though my brain feels utterly tired!

@patrickwise6297 4 жыл бұрын

Thank you - very helpful. Note, though the "14" width should have rank 7 and the "18" width should have rank 6 in the first Sunflower example.

@amantaybilash9495 4 жыл бұрын

Very good explanations. Thanks for sharing!

@sergio7470 3 жыл бұрын

Thank you for the clear explanation and getting to the point.

@AnnaelleD 4 жыл бұрын

Whoua! Great video. Thank you very much.

@hopestreet3452 4 жыл бұрын

Really clear explanation - thank you :)

@Dark6In6The6Eyes 3 жыл бұрын

Thank you, this was very clear.

@abishrantasaikia2393 3 жыл бұрын

Thank you so much. It was very helpful.

@Memzii Жыл бұрын

Thank you!

@kimballclark1392 Жыл бұрын

When there are repeated values I find a different formula where the D^2 gets terms added for each repeated value a 1/2(value^3 -value). For each value that is repeated a term is added to the numerator. Can you explain why your formula is different?

@tanoysaha9243 4 жыл бұрын

Can it be computed for all quantitative variables??

@emmanuelbanda7123 4 жыл бұрын

Nice

@najlaelhachem9159 4 жыл бұрын

Hi, thanks for these explanations. It's is very clear. I have two questions: First, would it be possible to use spearman's rank correlation with different rank numbers (eg: 7 ranks for the height and 10 for the width)?. Second, would it be possible to use the formula by ranking pvalues? Thanks plenty, Najla

@archidar1 4 жыл бұрын

1. I dont think that would be well defined. We cant even apply Pearsons correlation to unmatched data. I.e. it doesnt make sense to compare two data sets of different length. An extreme example is to compare 1 data value against no data value (e.g. height 1.7m, weight unknown). 2. Depends on whether the p-values are comparable. Some argue that p-values should only be seen as binary, either being greater than or less than the level of significance, and that the value of p-values should be ignored.