4 Study Design
4.1 Definitions
Dependent Variable : The variable measured e.g. the yield of tomatoes.
Independent variable : The variable that is changed by the researcher e.g. the amount of fertiliser given to tomato plants.
Experimental Study : The independent variable is changed systematically and measurements of the dependent variable are taken. All other variables are controlled (held the same).
Observational Study : There is no control over variables. Observations are recorded and then analysed for patterns.
Causal Relationship (Causation) : One event leads directly to another. Only experimental studies can be used to determine the existence of causal relationships.
4.2 Study Design Stages
Statistics is a key element at all stages of a scientific study. ?tbl-study_stages gives an overview of the typical stages in a scientific study with matching statistical considerations.
| Stage | Statistical Considerations |
|---|---|
| Research Question | Statistical techniques for answering the question |
| Literature Review | Evaluating statistical techniques from previous studies |
| Develop Hypotheses | Relevant statistical models and hypothesis tests |
| Sampling Design | Sample size; effects and estimating effect sizes; power of statistical tests |
| Collect Data - Experiment; observation; secondary sources | Data collected in a format required for the statistical analysis |
| Analyse Data | Data exploration; hypothesis testing |
| Report | Conventions for presenting graphical and tabular information; correct interpretation and presentation of statistical results |
4.3 Examples
Observational Study : Medical researchers accessed the records for all patients who had been treated for eczema at hospitals in a region over the last five years. For each patient they recorded variables such as age, biological sex, treatment type, and treatment success. They analysed this data set to find the relationships between the variables.
Experimental Study : Based on the observational study above, another group of medical researchers developed a new medication for eczema. To test their medication, they selected at random 90 patients who had eczema. They then randomly divided the patients into 3 groups of 30. The first group was given no medication. The second group was given 10mL of the medication once daily for 7 days. The third group was given 10mL of the medication twice daily for 7 days. The recovery of each patient was then recorded and the best performing treatment was selected for further trials.