Module 1: Research Design

Prospective Studies

Identifies individuals and collects information as events unfold (for ex., clinical trials).

Researchers follow a group of similar people (a cohort) over a long period. The participants do not receive an active intervention; researchers simply track their habits, environments, and health outcomes.

Retrospective Studies

Collects data after events have taken place (for ex., surveys, medical reviews, outbreak investigations).

Researchers look backward at data, events, or outcomes that have already occurred.

Simple Random Sampling

Every member of a population has an exact, equal chance of being chosen for the sample. (Think of the baseball example)

Stratified Sampling

Reseacher divides a total population into smaller, non-overlapping and homogeneous subgroups called strata. They are chosen so that similar cases are grouped together.

Treatment Group

Set of participants that receive the experimental manipulation.

Researchers apply the independent variable (treatment) to this group to measure its effects on the outcome.

Control Group

Set of participants that are handled identically to the treatment group, but do not receive the test intervention. They may instead receive no treatment or a placebo or the standard treatment.

Blocking

Ensures an equal representation of patients in each treatment group from both the low-risk and high-risk categories.

Blind Study

When researchers keep the patients uninformed about their treatment.

Placebo

A fake treatment; an effective placebo is the key to make a study truly blind.

Placebo Effect

Often times, a placebo results in a slight but real improvement in patients.

Operationalization

When we take things that are not directly measurable and try to measure it (thinking about why we major in a subject).

Classical Test Theory (CTT)

Helps us better understand how imperfect measurement relates to true, underlying constructs. Built on a simple, fundamental linear equation: X = T + E

CTT Equation

Observed Score (X) = True Score (T) + Error Score (E)

Observed Score (X)

The actual score a person receives on a test / what our measurement gives us (weighing scale result).

True Score (T)

Represents the true level of the trait or knowledge / hypothetical “true” underlying score.

Error Score (E)

Random, unsystematic influence that causes the observed score to deviate from the true score. Includes external factors (noise, room temp.) or internal factors (fatigue, illness, anxiety).

Random Error

Happens by change, not because there was a problem with the test or tester.

Reduced by doing more tests and can never be truly eliminated.

Systematic Error

Caused by things that can be remedied or mistakes that cause an error to be repeated. Caused by human error more frequently than random error.

Reliability

How precisely you measure something.

How consistent your measurement is. If a scale is calibrated incorrectly, it is still reliable if the results are the same each time.

Test-Retest Reliability

If you repeat the measurement at a later date, do you get the same measure?

Inter-Rater Reliability

If two people measure something, will they get the same answer?

Parallel Forms of Reliability

Does your measurement correspond to other measurements of the same construct?

Internal Reliability

If your measurement is formed from smaller, similar measures, do those individual parts produce the same result?

What do X and Y represent?

X is usually the independent variable, while Y is the dependent variable.

Independent Variables

Affects dependent variables.

Experimental Research

Contains a variable that is being manipulated, and a second or other variables react accordingly.

Non-Experimental Research

Cannot definitely conclude causation (e.g., surveys).

Quasi-Experimental Research

Unable to randomly assign people to conditions or control independent variables.

Internal Validity

Interested in whether you can draw causal conclusions from the relationship between two variables.

External Validity

Concerned with how generalizable your findings are. Can you apply the results of you findings outside of your specific experimental conditions?

Construct Validity

Gets to whether or not you’re measuring what you’re trying to measure.

Face Validity

Extent to which the measurement looks like its measuring what it’s supposed to.

Ecological Validity

Concerned with how similar your experiment approximates the real world in whcih the topic of study may be found.