What is the research question? Is there Independent Variables (IV) and Dependent Variables (DV)?
E.g. Statistics anxiety (DV) is a result of a weak maths self-concept (IV), little previous experience of maths (IV), gender (IV), and age (IV).
There are two relationships between variables. Dependence and Interdependence
Dependence relationships are as illustrated above. It is typical when experiments are designed to test predictions.
Interdependent relationships on the other hand contain no IVs or DVs. Variables correlate with one another, forming an inter-reliable relationship.
Examples:
1) Correlation (r). Coffee intake vs. Being an undergraduate
2) Factor analysis.
Step 2
What are the concepts and what variables are associated with them?
E.g. Statistical anxiety is measured by the statistical anxiety rating scale 1 (low) - 100 (high)
- Math self-concept scale, 1 (low) - 70 (high)
- No. of maths or statistics courses completed (O<)
- Age (in years) and Gender
Step 3
What type of variables are they?
Quantitative vs. Qualitative
- Statistical anxiety (Quant.)
- Maths self-concept (Quant.)
- No. of courses (Quant.)
- Gender (Qual.)
Step 4
What is the sample size (n)?
Sample size effects statistical power. Statistical power is the ability of a test to find an effect when it exists. However, the danger of having too much power, is finding an effect when it doesn't exist.
When there's too much..
Type I Error : Detecting a r'ship when one does not exist (risk rises with larger n)
When there's not enough..
Type II Error : Not detecting significant r'ship when it exists (risk rises with smaller n)
Power can range from 0 (no power) to 1 (full power). The recommended power value is .80 (Cohen, 1988)
Power is affected by:
1) Sample size - Power increases with n
2) Effect size - The strength of the effect ( .20 = small, .50 = medium, .80 = large)
3) Alpha - Power increases with alpha (e.g. p < .05 has more power than p < .001)
Step 5
Choose your analysis
1) Multiple regression (dependence r'ships)
2) Between groups ANOVAs (dependence r'ships)
3) Repeated measures ANOVAs (dependence r'ships)
4) Analysis of Covariance (dependence r'ships)
5) MANOVA (dependence r'ships)
6) Exploratory Factor Analysis (interdependence r'ships)
7) Structural Equation Modelling (dependence r'ships)
8) Path Analysis (interdependence r'ships)
All content shown here are as produced by Christine Critchley of Swinburne University of Tech, 2006.
