Page 129 - Quantitative Data Analysis

Quantitative Data Analysis
Simply Explained Using SPSS

Factors that affect Correlation

There are mainly four factors that affect correlation
1. Linearity

Nonlinear relationships will have negative effect on a measure
designed to find a linear relationship. Scatter plot helps to
identify the linearity of variables.
2. Heterogeneous Subsample
Samples of two or more groups may vary. Simpson’s
paradox provides information that shows how the pattern
of data is changes within sample group and as the
aggregated data. It is highly likely that certain variable have
different variance for two groups. For example, male and
female have different variance in statistics anxiety scale, or
Japan and USA students have different variance in math
scores. Following graph is adopted from class power points
of Clark (2014).

In such heterogeneous
subsample case, it is
better to calculate
correlation coefficient of
each group separately
(Clark,2014).

The Theory and Applications of Statistical Inferences 113