Designing a Study You Can Defend
Controls should represent people who could have contributed cases under the study's eligibility and sampling rules. Learn why this principle matters, how eligibility changes over time, and what happens when controls could never experience the outcome being studied.
Read Guide →
Designing a Study You Can Defend
Matching cases and controls can improve efficiency or address particular confounding concerns, but it is not always necessary or beneficial. Learn when matching is justified, how different matching strategies work, and what their implications are for analysis and interpretation.
Read Guide →
Designing a Study You Can Defend
Cases and controls may remember or report previous exposures differently, potentially distorting an observed association. Learn how recall bias develops, which prevention strategies are most defensible, and how to evaluate its possible influence on study findings.
Read Guide →
Designing a Study You Can Defend
A case-control study can investigate several exposures associated with the same outcome, but measuring more variables does not automatically improve the research. Learn how to prioritize exposures, plan adjusted analyses, assess interactions, and distinguish confirmatory findings from exploratory associations.
Read Guide →
Designing a Study You Can Defend
Not every question about change requires longitudinal data, but some cannot be answered credibly without observing participants or units over time. Learn how to recognize questions involving trajectories, temporal ordering, transitions, and within-person change, and decide whether repeated measurements are methodologically necessary.
Read Guide →
Designing a Study You Can Defend
Two time points may be enough to estimate an observed difference, but more are usually needed to investigate trajectories, rates of change, and nonlinear patterns. Learn how to determine the number of measurement occasions from your research question, analytical model, expected change, and practical constraints.
Read Guide →
Designing a Study You Can Defend
Two measurements of the same participants can provide legitimate longitudinal evidence, but they cannot reveal the trajectory between observations. Learn what a two-wave design can establish, why some questions require more measurements, and how to judge whether two time points are sufficient.
Read Guide →
Designing a Study You Can Defend
Not every variable in a longitudinal study needs to be measured at every occasion. Learn how to distinguish stable characteristics from changing constructs, determine when repeated measurement is essential, and design a schedule that preserves analytical validity without unnecessary participant burden.
Read Guide →
Designing a Study You Can Defend
Differences between age groups may reflect aging, historical conditions, or experiences shared by particular cohorts. Learn how age, period, and cohort effects differ, why they cannot be uniquely separated without additional assumptions, and how to design research that avoids misleading generational conclusions.
Read Guide →
Designing a Study You Can Defend
Observing one variable before another strengthens temporal reasoning but does not establish that the first caused the second. Learn what longitudinal data can demonstrate, which alternative explanations remain, and what additional assumptions or designs are needed for causal inference.
Read Guide →
Designing a Study You Can Defend
Longitudinal studies provide valuable information about change and temporal relationships, but repeated measurement can add costs, attrition, and analytical complexity without improving every research question. Learn when longitudinal research is justified, when simpler alternatives may be preferable, and how to evaluate the trade-offs before committing to follow-up.
Read Guide →
Designing a Study You Can Defend
A pretest-posttest design measures an outcome before and after an intervention. Learn when this approach is useful, how to implement it, and why observed improvement does not necessarily establish causation.
Read Guide →
Designing a Study You Can Defend
A significant difference between pretest and posttest scores does not necessarily mean an intervention caused the change. Understand the counterfactual problem, competing explanations, and the design features that support more credible causal conclusions.
Read Guide →
Designing a Study You Can Defend
An outcome can improve or deteriorate even when an intervention has no effect. Understand the alternative explanations for before-and-after changes and how to investigate them before making causal claims.
Read Guide →
Designing a Study You Can Defend
A baseline is more than the first measurement in a study. Learn how to choose its timing, instrument, and measurement conditions so that subsequent changes can be interpreted accurately.
Read Guide →
Designing a Study You Can Defend
One baseline measurement may be sufficient for some studies, but repeated observations can reveal variability, existing trends, and unstable starting conditions. Learn how to determine how many baseline measurements your research actually needs.
Read Guide →
Designing a Study You Can Defend
Measuring outcomes too early may miss an intervention's effects, while measuring too late can introduce competing explanations. Learn how to select post-intervention assessment times that match the outcome, intervention mechanism, and research question.
Read Guide →
Designing a Study You Can Defend
Randomized controlled trials are valuable for estimating intervention effects, but not every research question requires randomization. Understand when an RCT adds little useful evidence, which alternative designs may be appropriate, and how to justify your methodological choice.
Read Guide →
Designing a Study You Can Defend
Randomized controlled trials can provide strong causal evidence, but randomization may be impossible, unethical, or unsuitable for certain interventions. Understand the reasons, distinguish genuine constraints from convenience, and identify defensible alternative designs.
Read Guide →
Designing a Study You Can Defend
Random assignment strengthens causal inference by using chance rather than choice to allocate study conditions. It does not guarantee identical groups, representative samples, perfect implementation, or an unbiased study by itself.
Read Guide →
Designing a Study You Can Defend
A control group can make some research questions answerable, but not every study needs one. Whether you need a control depends on the contrast required by your question and the claim you intend to make.
Read Guide →
Designing a Study You Can Defend
Control group and comparison group are sometimes used interchangeably, but not always. The more important issue is what the group represents, how it was formed, and whether it provides the comparison your research question requires.
Read Guide →
Designing a Study You Can Defend
An appropriate comparison group is not simply a group that looks similar to the intervention group. It must represent the alternative required by the research question and support a credible interpretation of the resulting contrast.
Read Guide →
Designing a Study You Can Defend
Participants can sometimes serve as their own controls, allowing researchers to compare outcomes within the same individuals. Learn when this approach is appropriate, what it controls for, and why time-related changes, carryover effects, and measurement decisions still matter.
Read Guide →
Designing a Study You Can Defend
A second comparison group can answer an additional research question, distinguish intervention-specific effects, or strengthen the interpretation of existing comparisons. Learn when the extra group is justified and when it adds complexity without improving the evidence.
Read Guide →