Quantitative Research Designs

Quantitative Research Designs

1. Descriptive Studies

   • Often the start of the research process
   • Someone notices something unique or unusual
   • Use of case reports to describe a novel phenomena
   • Can't establish causality
   • Describes trends and generates hypotheses

2. Observational Studies

   • Cross-sectional
     • Descriptive - how many do A, how many do B
     • Univariate or multivariate
     • Measuring at one point in time
     • Fast, cheap, associations can be studied
     • No causality, can't study rare outcomes, methodological issues
   • Case control
     • Retrospective (backwards from outcome to predictor)
     • Sample chosen on basis of outcomes (cases) and comparison group (controls)
     • Determines strength of association between each predictor variable and presence or absence of disease
     • Able to study rare outcomes/long latent period, cheap, efficient, establishes association.
     • Still can't establish causality, methodological issues
   • Cohort
     • Cohort/follow up/longitudinal
     • Subjects followed forward in time to determine if new outcomes occur
     • No new subjects after initial recruitment
     • Rates of outcome among control and sample are compared
     • Collect same data repeatedly over time
     • Some evidence of causality, directly measure incidence of disease
     • Expensive, inefficient for rare outcomes, long follow up or large population, loss of subjects.

3. Experimental Studies

   • Clinical Trials
     • Pre-post
       • SUbjects act as own controls
       • Group A at T1 ->intervention -> Group A at T2
       • Compare pre-post data to establish response to intervention
     • Quasi-experimental
       • Different subject groups over differnt time periods
       • Group A at T1 and T2, Group b at T3 and T4
       • Compare group A with group B (must be similar groups)
       • Establish outcome from intervention
   • Laboratory
     • Controlled Clinical Trial (CCT)
       • Normal or no treatment for half
       • New treatment for other half
     • Random CCT
       • Researcher can't control who is in what group
       • Able to demonstrate causality
       • Only for mature research questions (must know variables and intervention)