Correlational Research

1
Correlational Research

1. Spare the rod and spoil the child
2. Idle hands are the devils workplace
3. The early bird catches the worm
4. You cant teach an old dog new tricks
5. Faint heart never won fair maiden

2
Nature of Correlational Design?

• No manipulation
• Describes important human behaviors
• Predicts likely outcomes
• Involves two or more variables (one dependent and
  one or more independent variables)
• Predictor independent variable (nAch)
• Criterion dependent variable (mental health)

3
Correlation Coefficient

• Existence
• Degree
• Direction
• Usually expressed as
• r (simple) or R (multiple) -1.00 to 0 to 1.00
• Eta 0.00 to 1.00 for curvilinear data set
• Use of scatter plots?

4
Warning!

• Relationship does not necessarily indicate
  cause-effect (causal connection)
• (it may suggest cause-effect but does not
  establish one)
• the independent variable DOES PLAY A ROLE in the
  occurrence of the dependent variable (but does
  not necessarily cause it)

5
Levels of correlation coefficient

• 0.00 to 0.20 Negligible
• 0.20 to 0.40 Low
• 0.40 to 0.60 Moderate
• 0.60 to 0.80 Substantial
• 0.80 to 1.00 High to very high
• (Guiford Fruchter, 1981)

6
Interpret the following
7
Prediction

• The more highly related two variables are, the
  more accurate are predictions based on their
  relationships
• Scatter plot ? regression line ? regression
  equation
• Y a bX (simple regression)
• Y a b1X1 b2X2 b3X3 (multiple regression)
• a b are constants
• a intercept
• b slope

8
Standard error of estimate (SE)

• Predictor and criterion dont usually have a
  perfect correlation
• So, an attempt to use X to predict Y is likely to
  result in a certain degree of error
• Y predicted vs. true/actual Y (difference in
  this is known as error score)
• The standard deviation of the error scores across
  all individuals is known as SE
• Note the smaller the SE, the more accurate the
  prediction!

9
The Coefficient of Determination

• Indicates the percentage of the variability
  (variance) among the criterion scores that can be
  attributed to differences in the scores on the
  predictor variable
• Coefficient of Determination r2 x 100
• E.g. r .60 ? r2 x 100 .36 x 100 36
• What does 36 mean?
• Especially useful when there are more than one
  independent variables (predictors)

10
Percent of variance accounted for
11
Steps

• Selecting a problem
• Choosing a sample (n30 at least)
• Selecting or developing instruments (tests,
  questionnaires, observation)
• Determining procedures
• Collecting and analyzing data
• Interpreting results (caution!)