Upload
siddharth-nath
View
2.590
Download
0
Tags:
Embed Size (px)
DESCRIPTION
Measuring Correlation - Spearman Rank C.C.
Citation preview
Spearman’s Rank C.C.Spearman’s Rank C.C.
Measuring CorrelationMeasuring Correlation
Beyond The ScattergraphBeyond The Scattergraph
For GCSE Mathematics you simply draw a line For GCSE Mathematics you simply draw a line of best fit through the data and talk about of best fit through the data and talk about “strong” or “weak” correlation in vague terms.“strong” or “weak” correlation in vague terms.
For GCSE Statistics you learn how to For GCSE Statistics you learn how to CALCULATE the amount of correlation so that CALCULATE the amount of correlation so that you can compare one scattergraph with another.you can compare one scattergraph with another.
You don’t even need to draw the scattergraph!You don’t even need to draw the scattergraph!
Spearman’s Rank C.C. FormulaSpearman’s Rank C.C. Formula
This formula is on the formula sheet so you don’t need to learn it!
This formula is on the formula sheet so you don’t need to learn it!
Perfect Negative
Correlation
No Correlation
Perfect Positive
Correlation
Interpretation of S.R.C.C.Interpretation of S.R.C.C.
Values close to 0 suggest no correlationValues close to 0 suggest no correlation 0.4 to 0.6 (0.5) is “weak correlation”0.4 to 0.6 (0.5) is “weak correlation” 0.7 and higher suggests “strong 0.7 and higher suggests “strong
correlation”correlation” This applies to negative values tooThis applies to negative values too There are no hard and fast rules about There are no hard and fast rules about
when “weak” becomes “strong” when “weak” becomes “strong” If If rr > 1 then you went wrong !!!!!! > 1 then you went wrong !!!!!!
Outline of ProcedureOutline of Procedure
Let’s say you are exploring Let’s say you are exploring nn heights and heights and weights in an investigationweights in an investigation
Rank the heights i.e. put them in order Rank the heights i.e. put them in order (1 = biggest, (1 = biggest, nn = smallest) = smallest)
Rank the weights Rank the weights (1 = biggest, (1 = biggest, nn = smallest) = smallest)
dd = difference between the two ranks for = difference between the two ranks for each personeach person
Square and add these differencesSquare and add these differences
Problems With Equal RanksProblems With Equal Ranks
What if two things are 3rd= ?What if two things are 3rd= ? One has to be third, the other fourth.One has to be third, the other fourth. To be fair, each takes the average:To be fair, each takes the average:
(3 + 4) ÷ 2 = 3.5(3 + 4) ÷ 2 = 3.5 What if three things are 5th= ?What if three things are 5th= ? Call them 5th, 6th and 7thCall them 5th, 6th and 7th Give each one the average rank:Give each one the average rank:
(5 + 6 + 7) ÷ 3 = 6(5 + 6 + 7) ÷ 3 = 6
Fertiliser v. Plant GrowthFertiliser v. Plant Growth
CropCrop AA BB CC DD EE
FertiliserFertiliser 12.812.8 17.117.1 8.38.3 6.76.7 10.210.2
YieldYield 103103 108108 8989 7575 105105
First, Rank The Data:First, Rank The Data:
CropCrop AA BB CC DD EE
FertiliserFertiliser 12.812.8 17.117.1 8.38.3 6.76.7 10.210.2
FertiliserFertiliserRANKRANK 22 11 44 55 33
YieldYield 103103 108108 8989 7575 105105
YieldYieldRANKRANK 33 11 44 55 22
Second, Find The Rank Differences:Second, Find The Rank Differences:
CropCrop AA BB CC DD EE
FertiliserFertiliser 12.812.8 17.117.1 8.38.3 6.76.7 10.210.2
FertiliserFertiliserRANKRANK 22 11 44 55 33
YieldYield 103103 108108 8989 7575 105105
YieldYieldRANKRANK 33 11 44 55 22
RankRankDifferenceDifference -1-1 00 00 00 11
Third, Square The Rank Differences:Third, Square The Rank Differences:
CropCrop AA BB CC DD EE
FertiliserFertiliser 12.812.8 17.117.1 8.38.3 6.76.7 10.210.2FertiliserFertiliserRANKRANK 22 11 44 55 33
YieldYield 103103 108108 8989 7575 105105
YieldYieldRANKRANK 33 11 44 55 22
RankRankDifferenceDifference -1-1 00 00 00 11
d^2d^2 11 00 00 00 11
Now Find “Sigma D Squared”Now Find “Sigma D Squared”
CropCrop AA BB CC DD EE
FertiliserFertiliser 12.812.8 17.117.1 8.38.3 6.76.7 10.210.2FertiliserFertiliserRANKRANK 22 11 44 55 33
YieldYield 103103 108108 8989 7575 105105YieldYield
RANKRANK 33 11 44 55 22
RankRankDifferenceDifference -1-1 00 00 00 11
d^2d^2 11 00 00 00 11
Finally Use The Formula:Finally Use The Formula:
n = 5 (there were 5 crops)
-> There is very strong correlation between the amount of fertiliser and the crop yield.
Here’s One For You To Try…Here’s One For You To Try…
AA 5050 175175 270270 375375 425425 580580 710710 790790 890890 980980
BB 1.801.80 1.201.20 2.002.00 1.001.00 1.001.00 1.201.20 0.800.80 0.600.60 1.001.00 0.850.85
These data show the distances in metres (A) of shops from the Museum in Barcelona and the prices in Euros (B) for a 50cl bottle of water.
Hypothesis: Shops closer to the museum charge more for their water than those further away.
Do the data support this hypothesis?
ProcedureProcedure
AA 5050 175175 270270 375375 425425 580580 710710 790790 890890 980980
RARA 1010 99 88 77 66 55 44 33 22 11
BB 1.801.80 1.201.20 2.002.00 1.001.00 1.001.00 1.201.20 0.800.80 0.600.60 1.001.00 0.850.85
RBRB 22 3.53.5 11 66 66 3.53.5 99 1010 66 88
Differences and SquaresDifferences and Squares
AA 5050 175175 270270 375375 425425 580580 710710 790790 890890 980980
RARA 1010 99 88 77 66 55 44 33 22 11
BB 1.801.80 1.201.20 2.002.00 1.001.00 1.001.00 1.201.20 0.800.80 0.600.60 1.001.00 0.850.85
RBRB 22 3.53.5 11 66 66 3.53.5 99 1010 66 88
d^2d^2 6464 30.2530.25 4949 11 00 2.252.25 2525 4949 1616 4949
ConclusionConclusion
Do the data support our hypothesis?