Occupational Therapy Research Methods & Evidence-Based Practice

Vocabulary practice flashcards covering introductory occupational therapy research methods, levels of evidence, experimental designs, data collection, psychometric properties, and measurement scales.

Evidence-Based Practice (EBP)

The integration of the best available research evidence with clinical expertise and client values or preferences to make clinical decisions.

EBP uses evidence

Outcomes Research

Research that specifically examines the results of OT services and interventions.

Needs Assessment Research

Determines what specific services are needed within a population or community.

Participatory Research

Researchers work directly with the people or community being studied, rather than simply studying them.

Peer Review

An evaluation process where at least two other experts in the field evaluate research before publication to determine if it is scientifically sound, appropriate, credible, and worth publishing.

Quantitative Research

Formal, objective, and systemic research that uses numbers and statistics to answer a research question.

Descriptive Research

Uses numbers or statistics to summarize and describe what currently exists without establishing cause and effect.

Exploratory Research

Research that explores a topic that is not well understood to identify patterns or possible relationships without proving causation.

Critical Appraisal

The process of evaluating research to decide how trustworthy, strong, and useful it is.

Independent Variable

The variable that the researcher directly manipulates or changes.

Dependent Variable

The variable the researcher measures to see if it changes as a result of the manipulation.

Random Selection

A sampling process that determines who gets into a study, where everyone in the target population has an equal chance of being selected.

Random Assignment

A process that determines which group a participant enters, giving every participant an equal chance of being assigned to experimental or control groups.

Randomized Controlled Trial (RCT)

research design that randomly assigns participants to groups, includes an intervention, uses a control condition, and compares numerical outcomes between groups.

Level 1 Evidence

Systemic reviews that combine and summarize evidence from multiple relevant studies, often using meta-analyses, to answer a specific research question.

Level 2 Evidence

Randomized Trials that involve randomly assigned groups and compare outcomes but may not include a control group or blinding.

Level 3 Evidence

Non-randomized trials or cohort studies that compare outcomes between groups but do not involve random assignment.

Level 4 Evidence

Case-control studies or cross-sectional studies that analyze outcomes based on already existing groups without manipulation of variables.

Level 5 Evidence

Mechanism-Base Reasoning or Case Studies that provide insights from individual cases or expert opinions but lack broader empirical support.

Double Blinding

Neither the participant nor the researcher knows which condition or group the participant is in until the study is completed.

GRADE

A system used to rate the quality of evidence across four levels: High, Moderate, Low, and Very Low.

Descriptive Vividness

how well the researcher describes what was observed or experienced.

Methodological Congruence

assesses whether the research methods properly fit the research question.

Codebook

A document that explains how collected data are coded, which can include variable names, representations, and coding instructions.

Data Cleaning

The process performed before data analysis to check for missing data, incorrect entries, and impossible values.

Raw Data

Information as originally collected before any errors are corrected or variables are organized.

Cleaned Data

Data that has been processed to correct legitimate mistakes, address missing information, and organize variables for analysis.

Univariate Research

A descriptive design that examines one variable at a time to describe data and identify patterns using frequency, central tendency, and dispersion.

Incidence

The occurrence of new cases of a condition developing within a population over a specific timeframe.

Prevalence

How common a condition is within a population at a particular time, accounting for all existing cases.

Bivariate Research

Research design that examines the relationship between two variables simultaneously.

Correlation

Two variables change in relation to each other

Causation

One variable actually causes a change in another variable

Feasibility Study

A study that evaluates whether a planned intervention or study can actually be done by focusing on the possibility of executing the main study.

Pilot Study

A small-scale version of a planned main study that tests whether all parts of the study work together smoothly.

Fidelity

How closely and consistently an intervention is delivered as it was originally intended in the study protocol.

RE-AIM Framework

A framework evaluating real-world intervention implementation across Reach, Effectiveness, Adoption, Implementation, and Maintenance.

Efficacy

A measure of whether an intervention works under tightly controlled, ideal conditions.

Effectiveness

A measure of whether an intervention works under actual real-world clinical conditions.

Responsiveness/Sensitivity

the ability of a measure to detect meaningful change over time

Ceiling Effect

Occurs when scores are already so high that there is little room for improvement

Floor Effect

Occurs when scores are already so low that there is little room to detect further decline

Minimal Clinically Important Difference (MCID)

The smallest amount of change in an outcome measure that is considered clinically meaningful to the patient.

Individual RCTs

Individual people are randomly assigned

Cluster RCTs

Entire groups are randomly assigned

Crossover Design

An RCT design where participants receive all interventions in the study, but the sequence or order of administration is randomized.

Waitlist Control

An RCT design where participants are randomly assigned to either immediate intervention or delayed intervention.

Stepped-Wedge Design

An RCT design in which different clusters begin receiving an intervention at staggered, sequential time points.

Dropouts

A participant who leaves a study

Can threaten validity because the people who leave may be different from the people that remain

Intention-to-Treat (ITT)

participants are analyzed according to the group they were originally assigned to, even if they drop out or switch treatments.

Per-Protocol Analysis

An analytical approach that excludes participants who drop out or fail to follow the study protocol strictly.

Imputation

A statistical method used to estimate missing values in a dataset using available participant information.

Internal Validity

The extent to which one can be confident that the intervention or independent variable directly caused the observed outcome.

Selection Bias

Systematic error occurring when there is a structural difference between the target population and the actual sample studied.

Information Bias

Systematic inaccuracy occurring in the measurement or collection of study variables.

Confounding Variable

An outside variable that affects the outcome and interferes with determining if the independent variable caused the observed change.

Type I Error

A false positive error where a researcher concludes that a relationship exists when it actually does not.

Type II Error

A false negative error where a researcher concludes that no relationship exists when one actually does.

External Validity

The degree to which research results can be generalized outside the study to real-world settings and populations.

Construct

An abstract idea or concept that cannot be directly measured until it is explicitly defined.

Operationalizing

The process of turning an abstract construct into a concrete, measurable procedure or assessment that yields a score.

Nominal Scale

Names/Categories

A scale of measurement that classifies data into distinct categories without any inherent order or ranking.

Ordinal Scale

Order/Ranking

A scale of measurement where categories have a defined rank or order, but the distance between ranks is not equal.

Interval Scale

Equal Intervals

A measurement scale where numbers have equal distances between them, but there is no true absolute zero.

Ratio Scale

Equal Intervals + True Zero

A measurement scale featuring equal intervals between values and an absolute true zero point.

Classical Test Theory (CTT)

A theory represented by the formula $\text{Observed score} = \text{True score} + \text{Measurement error}$.

Standardization

Create a specific protocol for how the assessment should be administered

Everyone should administer in the same way

Reliability

The consistency of a measurement

Test-Retest Reliability

A reliability metric evaluated by administering the same assessment twice to the same subjects at different times to see if scores remain consistent.

Split-Half Reliability

A reliability metric obtained by dividing an assessment into two halves and comparing scores from each half.

Parallel Reliability

Create two versions if an assessment, give both forms at the same time, and compare the results

Internal Consistency

The degree to which questions or items on the same assessment measure the same general concept, often measured using Cronbach's alpha.

Validity

Is the interpretation/measurement actually appropriate for what is trying to be measured?

Concurrent Validity

The degree to which results of a new assessment agree with an established gold-standard assessment administered at the exact same time.

Predictive Validity

The extent to which an assessment score can accurately predict a future outcome.

Sensitivity

The ability of a measure or diagnostic tool to correctly detect when a target problem or condition IS present.

Specificity

The ability of a measure or diagnostic tool to correctly identify when a target problem or condition is NOT present.

Passive Observation

An observational approach where the researcher observes participants without directly interacting with them.

Participant Observation

An observational approach where the researcher actively joins and participates in activities with the subjects.

Natural Environment

research takes place where the participant normally lives/functions

Semi structured setting

some aspects of the environment are controlled

Standardized/laboratory setting

highest level of control over confounding factors

Structured Interview

An interview style that uses strict pre-established questions and standardization rules to minimize subjectivity.

Semi Structured Interview

Has predetermined open-ended questions, the interviewer has the flexibility to ask follow-ups, and probe for more information

Unstructured Interview

Very flexible, the interviewer relies heavily on background knowledge and onging interaction

Verbal Data

What the participant directly tells you

Behavioral Data

What the participant does

Proxy Data

Information from someone who knows the participant