Glossary of Terms

Review Types

Systematic Review

A systematic review uses a detailed, comprehensive search strategy, often with the help of a professional librarian, to identify all relevant studies on a particular research question. Each study is critically screened against specific inclusion criteria. The quality of each study is appraised, and key information is extracted in a standardized way. Systematic reviews provide a qualitative synthesis of the findings across studies, usually in summary tables, to draw overall conclusions about the state of evidence on a topic. Because the methods are transparent and replicable, systematic reviews aim to minimize bias and present a comprehensive summary of the available research. Some systematic reviews also include a meta-analysis, which statistically combines results across studies (see definition below).

Meta-analysis

Meta-analysis is a statistical technique often used within a systematic review. In addition to summarizing findings qualitatively, meta-analysis combines numerical effect sizes from multiple studies to produce an overall estimate of the association between two variables. By pooling data across studies, meta-analysis increases statistical power and precision. It can also examine why results differ between studies. For example, by testing whether associations vary by population characteristics, study design, or measurement approach.

Umbrella Review

An umbrella review is a systematic review of systematic reviews. Using structured and transparent methods similar to a systematic review, umbrella reviews identify and synthesize findings from multiple systematic reviews on a broad topic. Because they summarize evidence across many review-level studies, umbrella reviews provide a high-level overview of what is known, where findings are consistent, and where uncertainty remains. Some umbrella reviews also include a quantitative synthesis of meta-analyses; these are sometimes referred to as meta-umbrella reviews.

Scoping Review

A scoping review uses systematic search methods to map the existing literature on a topic. Scoping reviews are commonly conducted when a research area is emerging, complex, or has not been comprehensively reviewed. Compared with systematic reviews, scoping reviews usually have broader research questions and more flexible inclusion criteria. Their goal is to describe the range, characteristics, and gaps in the available research to help guide future research rather than to critically appraise study quality or produce a pooled effect estimate.

Narrative Review

A narrative review provides a qualitative overview of the research on a topic using a less structured and comprehensive search strategy than a systematic review. While narrative reviews often describe the general literature they considered, they often do not report detailed search strategies, explicit inclusion and exclusion criteria, or standardized data extraction methods. Narrative reviews are useful for summarizing broad themes, providing context, and discussing theoretical or clinical implications. However, because their methods are less systematic and less transparent, they are more vulnerable to selective coverage of the literature.

For more information on review types, the reader is encouraged to download this free e-book published by Covidence.

Study Types

Longitudinal Study

A longitudinal study collects data from the same participants at multiple time points, allowing researchers to examine how outcomes change over time. In ACEs research, longitudinal designs may be used to track developmental trajectories following childhood adversity or to evaluate longer-term effects of interventions. Because the same individuals are followed over time, longitudinal studies can better account for differences between participants and help clarify the timing and direction of associations. However, they are often time-intensive, costly to conduct, and vulnerable to participant dropout (attrition), which can affect results.

Cross-sectional Study

A cross-sectional study collects data across participants at a single point in time. These studies examine whether differences in exposure (such as ACEs) are associated with differences in outcome measured at the same time. For example, researchers might compare mental health outcomes between individuals with higher versus lower ACE exposure. Cross-sectional studies are generally faster and less costly to conduct than longitudinal studies. However, because exposure and outcome are measured simultaneously, they cannot establish timing or direction of associations. As a result, cross-sectional designs are limited in their ability to support causal conclusions.

Statistical Concepts

Odds Ratio (OR)

The odds ratio (OR) compares the odds of an outcome in one group to the odds in another group. In the ACEs catalogue, an OR often compares the odds of an outcome among people exposed to ACEs with the odds among people not exposed to ACEs.

An OR of 1 means that the odds of the outcome are the same in both groups. An OR > 1 means that the odds of the outcome are higher among people exposed to ACEs. An OR < 1 means that the odds of the outcome are lower among people exposed to ACEs. For example, an OR of 1.5 means that the odds of the outcome are 50% higher in those exposed to ACEs. An OR of 0.7 means that the odds of the outcome are 30% lower in those exposed to ACEs.

Correlation Coefficient (r)

The correlation coefficient, r (also known as Pearson’s r), measures the strength and direction of the relationship between two variables. Two variables are related when changes in one variable are associated with changes in the other. The value of r ranges between –1 and +1. When r is positive, both variables increase or decrease together. When r is negative, as one variable increases, the other variable decreases. Values of r closer to 1 (either positive or negative) signify a stronger relationship, while values close to zero suggest a weaker relationship. An r of 0 means that there is no relationship between the two variables – there is no pattern between the values of the two. Benchmarks for the value of r are often used to qualify the strength of the relationship:

Hedge’s g

Hedge’s g represents the difference between two group means (averages). It expresses how far apart the groups are in standard deviation units while accounting for the variability within each group.

Hedges’s g is conceptually very similar to Cohen’s d but includes a correction factor that reduces bias, particularly in studies with small samples. Rule-of-thumb benchmarks are often used to interpret magnitude:

Cohen’s d

Cohen’s d represents the standardized difference between two group means (averages). It expresses how far apart the groups are in standard deviation units while accounting for the variability within each group.

Cohen’s d is conceptually very similar to Hedge’s g but does not include a small-sample correction factor, which means it can be slightly biased in studies with smaller sample sizes. Rule-of-thumb benchmarks are often used to interpret magnitude:

Confidence Interval (CI)

A confidence interval (CI) provides a range of values that are plausible for the true effect in the population, based on the study data and statistical model. The most commonly reported interval is the 95% confidence interval.

A 95% CI means that if the same study were repeated many times, and a 95% confidence interval were calculated each time, about 95% of those intervals would contain the true population value.

Confidence intervals convey the precision of an estimate. Narrower intervals indicate more precise estimates, whereas wider intervals indicate greater uncertainty.

Confidence intervals also help assess statistical significance. If the interval includes the value that represents “no effect,” the result is typically not statistically significant at the 0.05 level. For example:

For mean differences (e.g. Cohen’s d, Hedges’s g) and correlation coefficients (e.g., Pearson’s r) the “no effect” value is 0 rather than 1.

p-value

A p-value is the probability of observing results as extreme as (or more extreme than) those found in a study, assuming there is truly no association or difference in the population.

A commonly used threshold for statistical significance is p < 0.05.