lesson 6 - correlational research

how to tell of 2 quantitative variables are correlated

• column depiction

• scatter plots

• correlation coefficient (r)

how to tell of one categoriical and one quantitative variable are correlated

compare means across groups

how to tell if two categorical variables are correlated

compare frequencies/proportions across groups

10 out of 30 males = 33.3%

20 out of 60 females = 33.3%

→ variables are not associated

how to read column depiction

organize one column from lowest → highest

→ do we see a tendency for the scores in the other column to move systematically in comparison??

interpreting scatterplots

• how close together are the dots?

  • do they move in a specific direction/slant??

perfect positive correlation

• every time the x value increases by a unit, the y value increases by a fixed amount

  • on a scatterplot → all points are in a straight upward slope

perfect negative correlation

• as the x value decreases by a unit, the y value decreases bu a fixed amount

  • on a scatterplot → straight line sloping downward

how does a correlation allow us to make a prediction

step 1: finding the regression line

• find the line of best fit in a scatterplot

step 2: using the regression line to make predictions

• as long as the regression is sloped, you can make a prediction of someones y-score if you know their x-score

→ visually:

• find score on x axis

• move straight up until yu hit the regression line

• move across to y-axis

• y-value reached = predicted y-score

6 questions for evaluating statistical validity in correlational studies

1. how strong is the relationship?

2. whats the confidence interval?

3. has it been replicated?

4. could outliers be affecting results?

5. is there restriction of range?

   1. is the association curvilinear?

3 criteria to support a causal claim

1. covariance

2. temporal precedence

   1. internal validity

when is a third variable not a problem?

when the original relationship still exists within each level of the third variable

when does a third variable become a problem?

when the original relationship is completely explained by the third variable

a moderator

a third variable that modifies the relationship between the original two variables