Question 1
Difficulty: medium
How do you approach designing an epidemiologic study to investigate a suspected disease outbreak?
Sample answer
I start by clarifying the outbreak question as quickly and precisely as possible: what disease we’re dealing with, who is affected, where cases are clustered, and when the increase began. From there, I define a clear case definition so we can separate confirmed, probable, and possible cases consistently. I’d then choose a study design based on the situation, often a case-control study if the outbreak is ongoing and we need speed, or a cohort study if the exposed population is well defined. I focus heavily on data quality, including exposure histories, symptom onset dates, and potential confounders. At the same time, I look for operational value, not just statistical significance, because outbreak response needs actionable answers. I also make sure findings are communicated in a way that supports control measures, whether that means infection control, environmental remediation, or targeted public health messaging.
Question 2
Difficulty: easy
Tell me about a time you had to work with incomplete or messy data. How did you handle it?
Sample answer
In epidemiology, messy data is more the rule than the exception, so I’m careful not to let imperfect data stop the analysis. In one project, case reporting from multiple sites had missing onset dates, inconsistent exposure fields, and duplicate records. I first documented the gaps so I could be transparent about limitations. Then I cleaned the dataset by standardizing variables, flagging duplicates, and creating logic rules for impossible values. For missing onset dates, I used a hierarchy of sources, such as lab dates, admission records, and interview notes, to reconstruct timelines where possible. I also ran sensitivity checks to see whether the results changed under different assumptions. What mattered most was keeping the analysis honest and decision-useful. By the end, I was able to provide a clear interpretation with caveats, which helped leadership make timely decisions without overclaiming certainty.
Question 3
Difficulty: medium
How do you decide which statistical methods are appropriate for an epidemiologic analysis?
Sample answer
I begin with the research question, because the method should follow the question rather than the other way around. If I’m estimating prevalence or incidence, I’ll think about measures of frequency and the right denominators. If I’m evaluating an association between exposure and outcome, I’ll consider the study design, the type of outcome, and whether I need to control for confounding or effect modification. For example, logistic regression can be useful for binary outcomes, but I’d also think about whether the outcome is common enough that odds ratios may overstate the association. I’m comfortable with stratified analyses, Poisson or negative binomial models for count data, and time-to-event methods when timing matters. I also check assumptions carefully and keep the audience in mind, because the best method is the one that is valid, interpretable, and aligned with the public health decision being made.
Question 4
Difficulty: easy
Describe a situation where you had to communicate a complex public health finding to a nontechnical audience.
Sample answer
I’ve found that the biggest challenge in public health communication is often not the science itself, but making the results meaningful without losing accuracy. In one instance, I had to explain a rising trend in a communicable disease to local stakeholders who were concerned about school closures and public messaging. Instead of leading with statistical terms, I framed the issue around what the trend meant for risk and prevention. I used simple visuals, avoided jargon, and explained the difference between correlation and causation in plain language. I also made sure to address uncertainty directly, because people tend to trust information more when you’re honest about what is known and what is still under investigation. That approach helped the group focus on practical steps, like targeted outreach and symptom monitoring, rather than getting stuck on technical details that weren’t essential to decision-making.
Question 5
Difficulty: hard
How do you control for confounding in an epidemiologic study?
Sample answer
My first step is always to think carefully about the causal structure of the question, because controlling confounding blindly can introduce other problems. I identify variables that are associated with both the exposure and the outcome but are not on the causal pathway. Then I decide whether to address them at the design stage, through matching or restriction, or at the analysis stage using stratification or multivariable models. I’m also careful to distinguish confounding from effect modification, since the latter is not something to simply adjust away. In practice, I rely on subject-matter knowledge, not just statistical tests, because some important confounders may not be obvious from the data alone. I also check whether adjustment changes the estimate in a meaningful way, and I make sure the final model remains interpretable. My goal is to produce results that are both scientifically defensible and useful for public health action.
Question 6
Difficulty: medium
What would you do if surveillance data suggested an increase in cases, but the signal might be due to reporting changes rather than a real outbreak?
Sample answer
I’d treat it as a potential signal that needs verification rather than assuming it is a true outbreak. First, I would check whether anything changed in the surveillance system itself: reporting deadlines, laboratory methods, case definitions, data entry processes, or staffing. I’d compare the current pattern with historical trends and look for consistency across data sources, such as emergency department visits, lab results, and hospital admissions. If the increase is only showing up in one source, that raises the possibility of an artifact. I’d also contact local reporters or facilities to confirm whether the apparent rise reflects real illness or simply faster reporting. At the same time, I would not dismiss the signal too quickly, because system changes can sometimes reveal real increases sooner. The best approach is to investigate both possibilities at once so the response is balanced, evidence-based, and timely.
Question 7
Difficulty: easy
Give an example of how you would prioritize tasks during a fast-moving public health investigation.
Sample answer
During a fast-moving investigation, I prioritize by asking what will most improve decision-making in the next few hours or days. I typically separate tasks into immediate, short-term, and follow-up priorities. Immediate tasks include confirming the case definition, identifying the exposure window, and making sure the right people are aware of the potential risk. Short-term tasks involve line listing, descriptive analysis, and identifying common exposures or geographic clusters. Follow-up tasks include deeper analytic work, drafting a report, and documenting limitations. I also think about who can do what best, because delegation matters in high-pressure settings. If laboratory confirmation is pending, I’d keep moving on epidemiologic interviews and environmental assessment so we don’t lose time. I’m very deliberate about communication as well, since a lot of confusion in an investigation comes from unclear updates rather than lack of data. My goal is to keep the response focused, efficient, and adaptable as new information comes in.
Question 8
Difficulty: medium
How do you evaluate the quality of a public health dataset before using it for analysis?
Sample answer
I evaluate data quality through a combination of structure, completeness, consistency, and plausibility checks. First, I review the data dictionary and source documentation so I understand how variables were collected and coded. Then I look for missingness patterns, duplicate records, impossible values, and inconsistent formatting across fields. I also compare key variables against what I’d expect clinically or operationally—for example, onset dates that occur after hospitalization dates, or age and diagnosis combinations that don’t make sense. If the dataset includes multiple sites, I check whether reporting practices differ in ways that could bias results. I also want to know how the data were captured, because a well-designed registry is very different from an ad hoc spreadsheet. I don’t assume perfect data, but I do expect enough reliability to support the intended decision. If not, I either clean it carefully or recommend caution in interpreting the findings.
Question 9
Difficulty: hard
Tell me about a time when your analysis changed a public health recommendation or operational decision.
Sample answer
In one project, initial analysis suggested that a risk factor was tied broadly to a disease increase, but after stratifying by location and timing, I found the association was really concentrated in a smaller subset of facilities. That changed the response significantly. Instead of recommending a broad intervention across all sites, I advised a targeted approach focused on the highest-risk settings. I made sure to explain the reasoning clearly, including why the earlier pattern looked more general than it really was. The change mattered because it saved resources and reduced unnecessary disruption for facilities that were not actually driving the problem. What I took from that experience is that good epidemiology should sharpen decisions, not just produce numbers. I’m always looking for the most precise interpretation possible so public health action can be effective, proportionate, and credible to the people affected.
Question 10
Difficulty: easy
Why are you interested in working as an epidemiologist, and what strengths would you bring to the role?
Sample answer
I’m drawn to epidemiology because it sits at the intersection of data, people, and real-world impact. I like work that goes beyond describing a problem and helps explain why it is happening and what can be done about it. The part of the role I find most rewarding is translating evidence into action, especially when the stakes involve prevention, equity, and community health. My strengths include strong analytical thinking, careful attention to data quality, and the ability to stay calm when information is incomplete. I’m also comfortable collaborating with clinicians, laboratorians, environmental health staff, and policymakers, which is essential in this field. I don’t see epidemiology as purely technical; it also requires judgment, communication, and humility about uncertainty. I bring a practical mindset and a genuine commitment to making data useful, whether the goal is outbreak response, chronic disease surveillance, or evaluating a public health program.
