PSC 41 Chapter Notes - Chapter 4: Latin Square, Control Variable, Demand Characteristics
Chapter 8
• Bivariate correlation- association involving exactly 2 variables
• When both variables in an association are measured on quantitative scales a scatterplot is usually
the best way to represent the data
• An association with a categorical variable is best depicted with a bar graph
o Examine the difference between the group averages to see if there's an association
o Little difference = weak association between variables
• When both variables are measured the study is correlational and can support an association claim
• Use a t-test when at least 1 of the variables is categorical in an association claim to test if the
difference between means (averages) is statistically significant
• Interrogating Association Claims
o Construct Validity
▪ How well was each variable measured?
o Statistical Validity
▪ Effect size- magnitude of a relationship between 2 or more variables
• Larger effect sizes give more accurate predictions (errors of prediction get larger
when associations are weaker)
• Larger effect sizes are usually more important
▪ Statistical significance-conclusion that a result from a sample is so extreme that the
sample is unlikely to have come from a population in which there's no association or
no difference
• p = probability sample's association came from a population in which the
association = 0
• If less than 0.05 significant
• The stronger a correlation the more likely it is to be statistically significant
• Statistical significance depends on effect and sample size
▪ Outlier- an extreme score that stands out far away from the pack
• Outliers matter the most when a sample is small
▪ Restriction of range- situation where there isn't a full range of possible scores on 1 of
the variables in the association so the relationship from the sample underestimates
the true correlation
• Example: looking at correlation between SAT scores and college grades; college
only accepts students who have SAT score of 1800 or above so they
underestimate the true correlation
▪ Curvilinear association- parabola relationship (increase in variable leads to increase
then decrease in another)
• Correlation isn't causation
o 3 criteria for causation
▪ Covariance of cause and effect (must be correlation between independent and
dependent variables)
▪ Temporal precedence (independent variable must precede dependent variable)
• Directionality problem
▪ Internal validity (no plausible alternative explanations for relationship)
• Third-variable problem
20
PSC 41 Full Course Notes
Verified Note
20 documents
