PSY230H5 Lecture Notes - Lecture 2: Internal Consistency, Explained Variation, Effect Size
32 views2 pages
Lecture 2
Conclusion
• correlations reflect the direction and the strength of the relation between
two variables
• correlations vary from -1 to +1
• correlations depend on the sample (E.g higher in a sample of actresses
than normal women)
• a measure of the strength of a correlation is the amount of explained
variance (�2), which ranges from 0 to 1 (%)
Effect size
• there are multiple ways to quantify the size of an effect
• the binomial effect size display
W- W+ W- W+
H- 50 50 H- 75 25
H+ 50 50 H+ 25 75
r = .00 r = .50
• positive correlation in height
• take the correlation coefficient .50 and split it in half and it becomes .25,
bc you get an extra .25, you now get a .75
• cohen proposed the following verbal labels for different correlations:
• r r2 verbal label
.10 1% small
.30 9% moderate
.50 25% large
• even if a correlation is small, it still matters as it means many other
factors are important
Reliability
• the correlation between two assessments of the same construct with the
same method
1. asking about the same construct with many similar questions or
include many similar items on a multiple choice test (internal
consistency)
2. repeating the same measure twice some time apart (Retest-reliability)
Example: the satisfaction with life scale (SWLS)
1. in most ways my life is close to ideal
2. I am satisfied with my life
• 1= strongly disagree and 5= strongly agree
Internal consistency
• internal consistency can be measured by assigning half of the items of a
scale to one variable, and the other half to another variable, and then
compute the correlation between the two variables (split-half reliability)
Reliability of the SWLS
find more resources at oneclass.com
find more resources at oneclass.com