Data collected for the study for the first time are called primary data. This means the researchers generate or gather new information specifically to answer their current research question, using methods like surveys, experiments, interviews, or direct observations. Collecting data firsthand gives the investigators control over what is measured, how it is measured, and when it is measured, which helps ensure the data are directly relevant and of high quality for the study. In contrast, secondary data already exist from previous studies, reports, or records and were not collected with the current question in mind. Relying on these data can introduce limitations related to how they were collected, their age, or whether they fit the current definitions and units of analysis. Tertiary and quaternary are not standard terms for the data collected in a study; they typically refer to methods of sourcing or compiling information beyond the primary data, rather than data gathered anew.

Within PICOTS, the comparator is the alternate or control condition used to provide a baseline for evaluating the intervention’s effect. It acts as the counterfactual—the group or condition you compare the intervention against to determine whether observed outcomes are due to the intervention rather than other factors. This could be a placebo, a standard of care, or no treatment, depending on what makes sense for the question and ethical considerations. The comparator enables calculation of relative measures of effect, such as how much better or worse the intervention performs compared to the comparator. It is distinct from the population being studied, the intervention itself, the outcomes looked at, the timing, and the setting in which the study occurs. Without a meaningful comparator, it's hard to attribute differences in outcomes to the intervention rather than to random variation or external factors.

Strong data privacy in research hinges on a layered approach that reduces risk at several points. First, remove direct identifiers, so obvious personal details like names or exact IDs aren’t attached to the data. Then, minimize re-identifiability by evaluating what indirect or quasi-identifiers could combine with other data to reidentify someone. Governance comes next: a data-use agreement lays out who can access the data, for what purposes, and how results can be shared. Limiting access means only authorized individuals can work with the data, supported by authentication, role-based controls, and audit logs. Finally, store and transmit the data securely with encryption so that even if data are intercepted or accessed without authorization, they remain unreadable. This combination reduces privacy risk, supports ethical research, and aligns with regulatory expectations. Shaping the practice this way is preferable because relying on encryption alone while keeping a direct-identifier file for quick reference still creates a vulnerable linkage map if the key or file is compromised. Sharing raw data without agreements bypasses essential governance and oversight. Relying on de-identified data without any governance ignores ongoing risk management and accountability, since de-identification can fail or be reverse-engineered and access controls are still needed to monitor and enforce permissible uses.

Power is the probability that a study will detect a true effect if one exists. It reflects how likely the study is to uncover a real difference or association when there actually is one, and it equals 1 minus the probability of a Type II error (failing to detect an effect that is present). When planning a trial, researchers often target a power of 80% or 90%, meaning there’s an 80% or 90% chance of finding a true effect of the specified size given the chosen significance level and sample size. Achieving higher power makes it less likely that a real effect will be missed, but it typically requires a larger sample size, lower variability, or a larger true effect. This concept is distinct from controlling false positives (the Type I error rate) and from the actual sample size or trial duration, which are related planning aspects but different quantities. In practice, you set the desired power (commonly 0.8 or 0.9) when calculating the required sample size to balance resources with the ability to detect meaningful effects.