Experimental Research Quantitative Research Aim To determine how one thing determines another in your chosen population. Designs 1. Descriptive (subjects usually measured once) a. Establishes only relationships between variables b. sample >100-100000 participants to estimate relationship of variables 2. Experimental (subjects measured before and after a treatment) a. Establishes causality b. An experiment, usually a crossover, may only need tens of subjects Testing Hypotheses Hypothesis= a statement of expected results, deduced from theory or findings from other studies, with logical reasoning. Null hypothesis= statement of no difference or relationship, tested using statistics. Directional= one tailed, i.e., something will increase something Non-directional= two tailed, i.e., something will affect something Variables Independant: · What you change · The intervention/control/group Dependent: · What you measure · The outcome o Discrete- specific points on a scale that cannot be subdivided any further ie 1-10 scale o Continuous- theoretically can take any value between two points on a continuum Descriptive Designs Observational study-> Experimental Designs Cohort studies Case-control studies Cross-sectional studies Case series Intervention studies-> Pre-post single group Pre-post crossover Pre-post parallel groups
Post-only cross-over/parallel Pre-post Single Group Test a group of participants before and after a treatment/intervention. Any change after treatment should be considered co-incidental. No control groups. Within participant design. Pre-post Parallel Groups Test a group of participants before and after treatment. Also test another group at same timepoints, who didn't receive the treatment. Most common type of controlled trial. Between participant design. Pre-post Cross-over All participants complete conditions. Order varies between participants. Repeated designs measure. Design Pre-post single group - Relatively easy Pre-post parallel group Pre-post crossover Advantages - Control condition - Less time consuming for participants and researchers than crossover - Smaller samples than parallel - Can estimate individual responses (serve as their own control) - Easy and quick - Difficult to infer causation Post-only designs Analysis Disadvantages - Difficult to infer causation - Cannot assess individual responses - Groups of humans unlikely to represent exact controls - large sample size required - Lots of trials required so time consuming Data analysis methods must reflect the research design/question. Such as number of tests, type of data, whether comparing within or between participants. Choose test based on question not the other way round. Important Considerations · Potential confounding/extraneous variables · Blinding · Order/learning effects - counterbalancing · Sample size and group allocation