Establishing Construct Validity using a Correlation Matrix with Survey Data

Correlation Matrix –Construct-Related Evidence of Validity

•

•

1

Correlation Matrix –Construct-Related Evidence of Validity

2

Correlation Matrix –Construct-Related Evidence of Validity

3

Correlation Matrix –Construct-Related Evidence of Validity

4

Correlation Matrix –Construct-Related Evidence of Validity

5

These items target

“Reading Efficacy or

Confidence”

Correlation Matrix –Construct-Related Evidence of Validity

6

These items target “Math

Efficacy or Confidence”

Correlation Matrix –Construct-Related Evidence of Validity

7

Correlation Matrix –Construct-Related Evidence of Validity

8

Correlation Matrix –Construct-Related Evidence of Validity

9

Here we have High Inter-Item Correlation for these three items targeting“Reading Efficacy” (.70 or above is considered high)

Correlation Matrix –Construct-Related Evidence of Validity

10

Here we have High Inter-item Correlation for these three items targeting “Math Efficacy”

Correlation Matrix –Construct-Related Evidence of Validity

11

This is evidence of “Convergent Validity” for each of these constructs because the items within each scale are highly correlated

Correlation Matrix –Construct-Related Evidence of Validity

12

Low INTER-Scale Correlation between Scale 1 items and Scale 2 items.

Correlation Matrix –Construct-Related Evidence of Validity

13

This is an example of “Divergent Validity” because the scales diverge from one

another.

Correlation Matrix –Construct-Related Evidence of Validity

14

Meaning, items targeting “Reading Confidence” do not correlate well with

items targeting “Math Confidence”

Correlation Matrix –Construct-Related Evidence of Validity

15

Meaning, items targeting “Reading Confidence” do not correlate well with

items targeting “Math Confidence”

This is what you would expect, since

the items target different traits or

constructs.