# Regression

This scratchpad let's you see the build up to the correlationi coefficient. If there's a strong association then each pair of z-scores will be generally in the same direction (pos/neg) and the same quantity (small/medium/large). The correlation is simply the average of the z-score products.

#### Raw Scores and Z-Products (ZP)

##### ZP
.00
([s1v1]-[mean1])/[SD1]
.00
([s2v1]-[mean1])/[SD1]
.00
([s3v1]-[mean1])/[SD1]
.00
([s4v1]-[mean1])/[SD1]
.00
([s5v1]-[mean1])/[SD1]
.00
([s6v1]-[mean1])/[SD1]
.00
([s1v2]-[mean2])/[SD2]
.00
([s2v2]-[mean2])/[SD2]
.00
([s3v2]-[mean2])/[SD2]
.00
([s4v2]-[mean2])/[SD2]
.00
([s5v2]-[mean2])/[SD2]
.00
([s6v2]-[mean2])/[SD2]
.00
[Z11]*[Z21]
.00
[Z12]*[Z22]
.00
[Z13]*[Z23]
.00
[Z14]*[Z24]
.00
[Z15]*[Z25]
.00
[Z16]*[Z26]
([s1v1]+[s2v1]+[s3v1]+[s4v1]+[s5v1]+[s6v1])/6
([s1v1]-[mean1])**2
([s2v1]-[mean1])**2
([s3v1]-[mean1])**2
([s4v1]-[mean1])**2
([s5v1]-[mean1])**2
([s6v1]-[mean1])**2
[dev2s1v1]+[dev2s2v1]+[dev2s3v1]+[dev2s4v1]+[dev2s5v1]+[dev2s6v1]
[SumOfSquares1]/5
sqrt([Variance1])
([s1v2]+[s2v2]+[s3v2]+[s4v2]+[s5v2]+[s6v2])/6
([s1v2]-[mean2])**2
([s2v2]-[mean2])**2
([s3v2]-[mean2])**2
([s4v2]-[mean2])**2
([s5v2]-[mean2])**2
([s6v2]-[mean2])**2
[dev2s1v2]+[dev2s2v2]+[dev2s3v2]+[dev2s4v2]+[dev2s5v2]+[dev2s6v2]
[SumOfSquares2]/5
sqrt([Variance2])
([s1v1]-[mean1])/[SD1]
([s2v1]-[mean1])/[SD1]
([s3v1]-[mean1])/[SD1]
([s4v1]-[mean1])/[SD1]
([s5v1]-[mean1])/[SD1]
([s6v1]-[mean1])/[SD1]
([s1v2]-[mean2])/[SD2]
([s2v2]-[mean2])/[SD2]
([s3v2]-[mean2])/[SD2]
([s4v2]-[mean2])/[SD2]
([s5v2]-[mean2])/[SD2]
([s6v2]-[mean2])/[SD2]
([Z11]*[Z21])+([Z12]*[Z22])+([Z13]*[Z23])+([Z14]*[Z24])+([Z15]*[Z25])+([Z16]*[Z26])
[SZP]/5
[Corr]*([SD1]/[SD2])
[mean2]-([mean1]*[Slope])

m1: .00
[mean1]
s1: .00
[SD1]
m2: .00
[mean2]
s2: .00
[SD2]
SZP: .00
[SZP]
r: .00
[Corr]

#### Regression Playground

##### Constant
slope: .00
[Slope]
constant: .00
[Constant]

Once you have entered the raw scores for 2 variables above then you can now play around with predicted Y scores. To do that enter a score for your X variable below. It need not be an actual score (although it should be a "possible" score given the variable you are measuring). You can change scores and see the resulting Predicted Y score.

.00
[Constant]+([Slope]*[actual])