Research Studies: Subjects and Variables

Study Subjects

• Studies involve selecting a population for research
• Factors considered when choosing a study population are:
  • Inclusion and exclusion criteria
  • Generalizability
  • Cost
  • Recruitment sampling design

Variables

• Variables are key components in the anatomy of research, and include the following:
  • Predictor variables indicate exposures, interventions, or independent factors
  • Outcome variables indicate the results being measured
  • Confounding variables refer to outside influences skewing the results if uncontrolled

Example of Predictors

• A clinical trial evaluates the effectiveness of CardioMed on patients with high cholesterol in reducing the risk of heart attacks

Examples of Variable Types

• Predictor Variables- factors that may influence or predict the outcome of the study
• Exposures- specific risk factors or conditions that study participants may have
  • An example of exposures is high cholesterol levels
• Intervention- the specific treatment or action taken to affect the outcome of the study, like administering CardioMed vs. a placebo
• Independent Variables- variables manipulated to observe the effect on the study's outcome like dietary habits (mediterranean vs. western diet)
• The effect of the intervention is measured by outcome variables
  • Examples of measuring the effect of interventions include:
    • The incidence of heart attacks
    • Changes in cholesterol levels
• Confounding Variables- external factors influencing the outcome, skewing results if not controlled like family history of heart disease

Statistical Issues

• Hypothesis-a supposition or proposed explanation made on the basis of limited evidence as a starting point for further investigation.

• Hypothesis: 50- to 69-year-old women with CHD who take fish oil supplements are expected to have a lower risk of recurrent myocardial infarction than those who do not take such supplements.

• Statistical Significance

  • It is important in analytical studies
  • Purely descriptive studies do not involve significance tests

• Sample size is also a statistical issue

Goal of Clinical Research

• The goal of clinical research is to draw inferences from a study
• There are two types of inferences:
  • Inference #1-internal validity and drawing correct conclusions about the study
  • Inference #2-external validity and applying conclusions to a broader population

Designing the Study

• The choice of a sample of subjects represents the population
• Variables are chosen to represent the phenomena of interest

Implementing the Study

• The actual study should correlate with the study plan

Types of Errors in Research

• In research, mistakes in studies are called errors
• Random error-mistakes due to chance with no predictable pattern caused by variability in sample selection
• An example of a random error is the variation in caffeine consumption rates in a sample 38%, 42% instead of the true 40%
• Increased sample sizes and better measurement of instruments can minimize random errors.
• Systematic error-biases distorting results in one direction due to systematic biases in data collection
• Misclassification of dementia diagnosis influenced by health plan policies is an example of systematic error
• This can be reduced by identifying biases and gathering additional information
• Selection bias-error affecting the generalizability of results due to non-representative sample
  • Researchers can avoid selection bias by ensuring diverse sample selection and consider multiple settings
• Measurement error-errors affecting the accuracy of data collected because of poorly designed instruments
  • Accurate data can be ensured through validated instruments and pilot test questionnaires

Key Takeaways on Errors

• Random errors are minimized by increasing the sample size
• Systematic errors are minimized by addressing biases directly
• Reduce selection and measurement errors with careful study design and data collection methods

Understanding Errors in Research

• Studies will have errors
• Determine if the errors are significant enough to render the study conclusions unreliable
• Error minimization is similar to starting with an ideal scenario, and making adjustments to balance internal/external validity and feasibility
• Considerations:
  • What is essential for the study?
  • What can be realistically achieved?

Evaluating the Study Plan

• After creating the plan, perform a Post-Formulation Review:
  • Assess if it addresses the research question adequately
  • Determine if the implementation maintains acceptable error levels
• Critical step: this evaluation can be overlooked potentially leading to unviable studies

Research and Adaptability

• Good scientists identify and discard flawed ideas
• Emphasis on willingness to refine and better research