Body

Glossary

 

Term Definition
Block exchangeable

a correlation decay structure in which CAC and IAC are constant.

CAC

cluster autocorrelation, defined as the correlation between the population means from the same group or cluster at two adjacent time periods, also called the over-time correlation at the group-level.

Churn rate

one minus the proportion of individuals in a group who appear in each of a pair of periods. Also called the attrition rate.

Closed cohort design

the same members or participants are observed at each measurement occasion.

Cluster autocorrelation

the correlation between the population means from the same group or cluster at two adjacent time periods, also called the over-time correlation at the group-level.

Cohort design

design in which some or all members are observed at multiple time periods. Includes closed cohort and open cohort designs.

Common interventionist or facilitator

the physical or virtual interventionist that delivers at least a part of the intervention to more than one member or participant.

Condition

the study conditions reflect the levels of the intervention, usually intervention vs. control in a study with two study conditions. Study condition is equivalent to study arm.

Cross-over design

design in which all members receive the intervention or control conditions, possibly transitioning back and forth multiple times. May be individually randomized or group-randomized. 

Cross-sectional design

design in which different members or participants are observed at each measurement occasion.

the hypothesized intervention effect, in units corresponding to the scale of the variance.

df

degrees of freedom.

Discrete-time decay structure

a correlation decay structure analogous to an autoregressive model of order 1 - AR(1) - correlation matrix and allows the CAC and IAC to decay as a function of the distance between observations.

Effectiveness-implementation design

design intended to assess the impact of an intervention on clinical effectiveness and implementation. 

Encouragement design

design that randomizes participants to encouragement or no encouragement to receive the intervention arms. 

Exact test

also called a randomization test, a hypothesis test that doesn’t rely distributional assumptions and instead on only the randomization scheme at hand. 

Group

the units of randomization, often called groups or clusters, nested within the study conditions or arms.

Group-period

also called cluster-periods, a single time period within a sequence of a SWGRT.

IAC

individual autocorrelation, defined as the correlation on the outcome variable for the same individual at two adjacent time periods, also called the over-time correlation at the member-level.

ICC

intraclass correlation coefficient or intracluster correlation coefficient.

Individual autocorrelation

the correlation on the outcome variable for the same individual at two adjacent time periods, also called the over-time correlation at the member-level.

Intraclass correlation

the expected average correlation for the outcome among members of the same group or cluster at the time of the analysis, usually at the posttest. It is also defined as the proportion of variance in the outcome attributable to group membership at the time of the analysis, usually at the posttest.

IRGT

individually randomized group treatment trial, also called a partially clustered design, where individuals are randomized to study conditions but receive at least some of their intervention in groups or from a common interventionist or facilitator.

Member

the units of observation, also called participants, nested within groups.

MOST

multiphase optimization strategy, an approach for the preparation, optimization, and evaluation of an intervention.

Nested factor

a factor is nested if it appears within only one level of a factor higher in the hierarchy. Subjects are nested within groups or clusters if they appear in only one group or cluster. Groups are nested within study conditions if they appear in only one study condition.

Net difference

the difference between two study conditions over two points in time.

Noninferiority trial

trials that are designed to determine whether a new intervention is not appreciably worse than an active control or standard treatment. 

Open cohort design

design in which the members or participants within a group are observed at one or more measurement occasions, but not necessarily all. 

Paired availability design

design in which there are independent pairs of intervention and control groups, and within each pair the intervention is the availability of treatment as opposed to receiving it.

Parallel GRT

parallel group-randomized trial, also called a parallel cluster-randomized trial, where the units of assignment are groups or clusters, and the units of observation are members of those groups or clusters.

Power

the probability of correctly rejecting a false null hypothesis.

RCT

randomized controlled trial, or randomized clinical trial, where the units of assignment and observation are the study participants, and where participants have no interaction with each other or with a common interventionist or facilitator post-randomization.

Regression discontinuity design

also referred to as a cutoff design, a quasi-experimental design in which assignment of members or groups to the intervention is determined using a quantitative score and a cut-point.

Risk prediction models

models estimating the probability or risk of a participant experiencing a health-related outcome or event. 

Sequence

also called step, the series of time periods in which a group or cluster begins in the control condition then transitions to the intervention condition.

Simple difference

the difference between the two study conditions at one point in time.

SMART

sequential multiple assessment randomized trial, a design allowing investigators to identify combinations and/or sequences of intervention components that are optimally effective. Used for the optimization phase of MOST. 

Surrogate endpoint

an indirect measure of a clinical outcome, e.g. measuring blood pressure to indicate stroke prevention.

Strata

the stratification levels, e.g., high vs. low.

SWGRT

stepped wedge group-randomized trial, also called a stepped wedge cluster-randomized trial. A GRT in which groups begin the study in the control condition, are randomly assigned to sequences, and cross-over to the intervention condition at pre-determined time points in a sequential, staggered fashion until all groups or clusters receive the intervention.

Time

the measurement occasions or periods, usually pretest and posttest for GRTs and IRGT trials but there are often three or more time periods for SWGRTs.

Type I error rate

the probability of making a type I error, which is to incorrectly reject the null hypothesis of no intervention effect.

Type II error rate

the probability of failing to reject a false null hypothesis.

Variance

the expected variance of the outcome at the time of the analysis, in either standardized units (=1) or in the natural units of the outcome (e.g., lbs. or mm Hg).

Last updated on