Sampling
Population-->
Sampling frame-->
Sample
Probability
sampling
Simple random
Stratified random
Cluster
Nonprobability
sampling
Convenience
Quota
Purposive
Types of
surveys
Cross-sectional
Longitudinal
Trend
Panel
Cross-sequential
Advantages
Large samples
Non-observable behavior
Efficient
Disadvantages
Not causal
Reactivity
Response styles (social desirability, response
acquiescence, response deviance)
Volunteer problem
Correlational
study
Correlation coefficient
r
From -1 to +1
Magnitude vs. sign vs. significance
Positive/negative correlations
Examples
High
Correlations
IQ test reliability
.93
IQ and first grade grades .85-.90
traffic fatalities and indices of progress in 3rd world
countries -.72
Examples
Moderate
Correlations
IQ in high school and college grades .50-.55
Physical similarity of spouses .40
IQ of fraternal twins .54
Aversive maternal and child behaviors .55
Examples
Low
correlations
Reading achievement and TV viewing -.05
GRE and grad school grades .17
Examples--table
Third
variable problem
The more ice cream being bought in a region, the
greater the number of drownings.
The more active children are in Boy/Girl Scouts, the
less likely they are to commit a street crime.
People who eat breakfast tend to live longer than
people who skip breakfast.
Third
variable problem
The more time fathers spend with their children, the
less likely they are to sexually abuse them.
The larger a womans feet, the less trouble she has in
childbirth.
Scatterplots
aka scatter diagrams
Three
things that can bias correlations
Truncated range
Outliers
Small sample size
Causal
relationships
Cross-lagged correlations
Path analysis
Structural equation modelling
Correlations:
4 sets of 3
Three parts to a correlation: sign, magnitude,
significance
Show rs: raw data, correlation coefficient,
scatterplots
3rd variable problem
Possible biases: truncated range, outliers, small
sample size