What researchers mean by...

For a dozen years, from 2005 to 2017, the Institute published a regular column called "What researchers mean by..." in its newsletter At Work. The column was designed to help readers better understand what researchers do and the language they use when reporting their findings. More than 35 common research terms used in the health and social sciences were covered in the column, each explained in simple language using everyday examples.
Thumbnail

Absolute and relative risk

Absolute risk is the number of people experiencing an event in relation to the population at large. Relative risk is a comparison between two groups of people or in the same group of people over time. Knowing which type of risk is being reported is important in understanding the magnitude of the risk.
Published:
Thumbnail

Bias (part 1)

Bias refers to flaws in the design, conduct and analysis of research that can (usually unintentionally) creep into a study and skew the findings. Five types of bias are reviewed here: selection, attrition, measurement, analysis and publication bias.
Published:
Thumbnail

Bias (part 2)

Bias refers to flaws in the design, conduct and analysis of research that can (usually unintentionally) creep into a study and skew the findings. Two types of bias related to the collection of data are reviewed here: recall and surveillance bias.
Published:
Thumbnail

Blinding

Blinding is a practice whereby study participants are prevented from knowing certain information that may somehow influence them and, in turn, affect the study’s results.
Published:
Thumbnail

Bootstrapping

Bootstrapping is a statistical technique for determining how confident we can be in the findings of a study
Published:
Thumbnail

Case control study

Case control studies start with an outcome (such as a disease) and work backwards to find exposures that may be linked to it.
Published:
Thumbnail

Cohort study

A cohort study follows a group of people over time to understand the relationship between some attribute shared by the group of people at the beginning of the study and the eventual outcome.
Published:
Thumbnail

Confidence intervals

A confidence interval is the range of values above and below a finding in which the actual value is likely to fall. It represents the accuracy or precision of an estimate.
Published:
Thumbnail

Confounding variables

A confounding variable is an unforeseen or unaccounted-for factor that may call into question the finding of a relationship between two other factors or variables. In other words, it “confounds” the relationship by being the “something else” that may explain the relationship.
Published:
Thumbnail

Cross-sectional vs. longitudinal studies

Cross-sectional studies make comparisons at a single point in time, whereas longitudinal studies make comparisons over time. The research question will determine which approach is best.
Published:
Thumbnail

Data linkage

Many organizations collect data (sometimes called secondary or administrative data) to do their business. This data, when linked to another source of data, can become more fruitful in answering questions, and potentially generate new knowledge.
Published:
Thumbnail

Difference in differences

A method called "difference in differences" helps identify the effect of an intervention when intervention and control groups have meaningful differences.
Published:
Thumbnail

DOI

A DOI (or digital object identifier) is a permanent name given to studies, publications and other Internet resources to ensure a permanent link to an electronic article even when its’ URL has changed.
Published:
Thumbnail

Epidemiology

The cornerstone of public health, epidemiology investigates which groups in a population are affected by disease, and why.
Published:
Thumbnail

Factor analysis

Factor analysis is a technique that helps researchers study a concept that cannot easily be measured by looking for patterns in the movement among certain measured variables.
Published:
Thumbnail

Generalizability

Generalizability refers to the degree to which the results of a study can be applied to a larger population, or the degree to which time- and place-specific findings, taken together, can result in a universal theory.
Published:
Thumbnail

Grey literature

Documents and other information that haven’t gone through peer review before being published are referred to as “grey literature.” Magazine articles and conference proceedings, for example, fall in this category.
Published:
Thumbnail

Grounded theory

If you’re a grounded theorist, you engage a ‘zig-zag’ approach to research—jumping from the field to the drawing table, then back again—in an ever-changing process of fine-tuning your findings. Grounded theory is all about having an open mind and seeing where the data take you.
Published:
Thumbnail

Internal validity

Internal validity ensures a study’s findings are the result of the intervention being studied and not due to chance or some other factor. In that sense, internal validity indicates how well a study was designed and carried out to prevent systematic errors or bias.
Published:
Thumbnail

Mean, median and mode

Related to numbers-based findings, ‘mean’ is the average, ‘median’ is the number that separates the higher half from the lower half, and ‘mode’ is the value that occurs most often.
Published:
Thumbnail

Meta-analysis

A type of systematic review, meta-analysis integrates or adds the findings from many studies to create one large overview. By combining results, it reduces the time and energy spent looking at the difference pieces of research.
Published:
Thumbnail

Missing data

Research data may have holes for a number of reasons — from questions left blank on a survey to people dropping out of a study. Sometimes the missing information matters; sometimes it doesn’t.
Published:
Thumbnail

Multiple regression

Multiple regression is a popular technique in statistics used to measure the relationship between many variables and an outcome.
Published:
Thumbnail

Observational vs. experimental studies

Observational studies observe the effect of an intervention without trying to change who is or isn't exposed to it, while experimental studies introduce an intervention and study its effects. The type of study conducted depends on the question to be answered.
Published:
Thumbnail

Path analysis

In path analysis, researchers use models to map out relationships between many variables and test them for strength.
Published:
Thumbnail

Peer review

Peer review is a quality control process in which researchers submit their work to other experts—their peers—for evaluation.
Published:
Thumbnail

Primary data and secondary data

Primary data and secondary data are two types of data, each with pros and cons, each requiring different kinds of skills, resources.
Published:
Thumbnail

Primary, secondary and tertiary prevention

Primary, secondary and tertiary prevention are three terms that map out the range of interventions available to health experts.
Published:
Thumbnail

Probability

Probability provides information about the likelihood of something happening. In public health research, it looks at the likelihood of a health effect due to exposures to risk factors.
Published:
Thumbnail

Psychometrics

Research on psychometrics examines the properties of a measure to ensure it’s accurate, consistent and sensitive to change.
Published:
Thumbnail

Qualitative research

Qualitative research aims to make sense of human experience, beliefs and actions. As such, it provides a rich source of information on social systems and processes.
Published:
Thumbnail

Quality of life

Subjective but measurable, quality of life as an outcome measure provides vital clues about the success of an intervention, which is often missing from a clinical point of view.
Published:
Thumbnail

Randomized controlled trial

One of the most powerful research tools, the randomized controlled trial is considered by some to be the “gold standard” for generating reliable evidence.
Published:
Thumbnail

Regression to the mean

Regression to the mean is a statistical occurrence that may result in distorted or misleading findings if not taken into account.
Published:
Thumbnail

Sample size and power

Sample size refers to the number of participants or observations in a study. Power refers to the probability of finding a significant relationship. Often researchers begin a study by asking what sample size is necessary to produce a desirable power.
Published:
Thumbnail

Sampling

Sampling is the process of identifying the representative part of a larger whole that will allow findings from the sample to be applied to the whole. It is one of the most challenging aspects of study design.
Published:
Thumbnail

Selection bias

Selection bias is a common type of error where the decision about who to include in a study can throw findings into doubt.
Published:
Thumbnail

Simple regression

Simple regression helps researchers understand the relationship between two items, which can then be used to make predictions.
Published:
Thumbnail

Statistical significance

A statistically significant finding means that the differences observed in a study are likely real and not simply due to chance.
Published:
Thumbnail

Statistically adjusted

When determining the relationship between two factors, scientists need to take into account other factors that may affect that relationship. When they do, they statistically adjust their findings to reflect the impact of these other factors.
Published:
Thumbnail

Subgroup analysis

Subgroup analysis is a tool for exploring differences in how people respond to a health intervention, but it must be used with care.
Published:
Thumbnail

Survival analysis

Survival analysis techniques allow researchers to study lengths of time, often to predict when a given event or end point will occur.
Published:
Thumbnail

Systematic review

A systematic review helps users of evidence keep up to date on a body of research by synthesizing the findings of higher quality studies on a given topic.
Published:
Thumbnail

Validity and reliability

Validity and reliability are concepts that capture the measurement properties of a survey, questionnaire or another type of measure.
Published: