CORRELATION & REGRESSION
1) Bivariate frequency distribution involves two set of variable and if there be m rows and n
columns, then there will be (mn) cells, 2 marginal distribution, (m+n) conditional at most.
2) Correlation is concerned with the measurement of the ‘strength of association’ between
variables.
3) Regression eqn is a linear relationship between two variables which is utilised in predicting
the dependent variable for a known value of the independent one.
4) Coefficient of correlation is a unit free measure and it always lie in the closed interval [-1,+1]
for a perfect correlation r = ±1
5) In educational and psychological measurements Coefficient of correlation is used in
problems of reliability and validity of tests.
6) The regression line of y on x is derived by the minimisation of vertical distances in the scatter
diagram.
7) The regression line of x on y is derived by the minimisation of horizontal distances in the
scatter diagram.
8) (6) and (7) is derived by the method of least square.
9) The difference between the observed value and the estimated value by regression equation
is called error or residue.
10) If R = 1, there is a perfect agreement between the variables and if R = 1, there is a perfect
disagreement.
11) Pearson’s product – moment correlation coefficient is utilised to calculate correlation for
variables and Spearman’s Rank correlation for attributes.
12) For the relation y= a+bx, correlation coefficient r = 1 if b>0 and r = -1 if b<0.
13) High values of one set are associated with the high values of the other and low values of one
set are associated with the low values of the other if r is positive.
14) High values of one set of variable are associated with the low values of the other and low
values of one set are associated with the high values of the other if r = -1.
15) Correlation may be linear or non-linear but we study about only the linear type of
association between the two variables, what we get from product moment correlation.
16) While calculating rank correlation we use to find only the extent of association between the
attributes and as such it is not necessary to be linear.
17) The sum of difference of rank (d) in Rank correlation coefficient is always zero.
18) If two variables are independent, their correlation coefficient is zero, but the converse is not
true.
19) Number of points on either side of the regression line should be approximately equal and
equidistant also from the line such that the positive and negative deviations nullify each
other leading to total sum of deviation be zero.
20) Two regression lines coincide with each other if r = ±1.
21) -1 ≤ r ≤ +1, -1 ≤ R ≤ +1.
22) Coefficient of alienation = √1-r2
23) Coefficient of determination is r2 and Coefficient of non-determination is 1-r2
CAMistri
Copyright© 2013 CAMistri. All rights reserved http://www.camistri.com/
