Research Process & Design I

Why do Research?

Research Assumptions

Inferential Testing → Types of Sampling Error

Types of Error

Internal validity

Deals with how well a study is designed and conducted

External validity

Deals with how easily the results of a study can be extrapolated in a real-world setting

We try to balance…

internal & external validity

Steps of Clinical Research

Determine the research question → FINER CRITERIA

Examples of Research Questions → highlighted questions are good questions

good bc they can be tested on

How can we make the following research questions better? How would we design a study to answer these questions?

Just make them testable

Perform background research on the question

Types of Clinical Study Designs

Who are good study patients?

Why is Sampling Important

Target Population

Inclusion Criteria

Exclusion Criteria

Probability (Random) Samples vs Non-probability Samples

Simple Random Sample (SRS)

– Sampling units selected at random

– Use a randomly generated number table

Systematic Sample

– Every kth sampling unit is selected

– Shouldn’t be used for events that occur in a cyclic pattern

• I.e., month of the year when viruses occur

Stratified Sample

– Subgroups are formed based on certain characteristics of interest

• Age

• Disease(s)

• Site

– Patients are randomly selected from these subgroups, this ensures representation from the subgroups.

Convenience Samples

• Select subjects who are convenient to researcher

• Cannot ensure subjects are representative of the target population

• Can be useful for exploratory research, but not recommended other types of research

Judgment or Purposive Samples

How are patients “selected” to participate in medical research?

Why are random samples seldom used?

Random Assignments

Block Randomization

Different types of randomization

Double Blind Study

Therapeutic misconception

Most of us believe there is always something better

Sample Size

How to calculate sample size

Non-response error

Variables

Nominal /Categorical Variables

Ordinal Data

Continuous Variables

Examples of Variables

Age → continuous

Race/ethnicity → categorical

Race → categorical

Cocaine history → categorical

Rules for Transforming Variables

Independent Variables

Dependent Variable

Definition: Confounding Variables

Confounding Variables

Objective vs. Subjective measurements

Is objective data better than subjective data?

Surrogate Measurements

Clinical Endpoints vs. Surrogate

Accuracy & Precision

Incidence & Prevalence

Sensitivity & Specificity

practice Q

practice Q

Calculating Sensitivity & Specificity

fill in a b c d

dont have to know how to calculate

In a given population of 1,000 people, 100 patients are HIV positive. A rapid HIV test detects HIV in 90 patients with HIV and in 5 patients without HIV.