HIGH YIELD RATING:
10 – 2×2 Table
9 – Bias & Study Design (3-Randomization, 2 – Sampling & Selection Bias, 1 – Blinding & Placebo)
9 – Sensitivity & Specificity
9 – Type of Study Design (6-Cohort, 4-Case-Control, 4-RCT)
9 – Prevalence & Incidence
6 – p-Value and Statistical Significance
6 – Positive & Negative Predictive Value
6 – Relative Risk and Odds Ratio
4 – Confidence Interval
4 – Central Tendency (2-Mean, median and Mode, 1-Robustness, 1-Skew)
4 – Standard Deviation
3 – Statistical Testing
2 – # Needed to Treat/Harm
2 – Case Fatality Rate
2 – Power
1 – Absolute Risk Reduction and Attributable Risk
NO YIELD (HYR of 0):
Correlation coefficient & regression
Precision and accuracy
Different levels of prevention
Phases of Clinical trials
Range and interquartile range
Pre Test and Post Test probability
Intention to treat vs. per protocol
Blocking (type of randomization)
Hawthorne Effect and Observer expectancy bias
Lead time and length time bias
Funnel plot & publication bias
Surrogate, Clinical and Composite Outcomes
For a lot of biostats or epidemiology questions you don’t actually have to read the question. They often start with a long paragraph about the study’s design or the disease being studied. However, most of the time they are asking you to regurgitate the definition of a term (which does not change based on that particular study/disease described) or they are asking you to do a calculation (in which case you just need to plug the given numbers into a formula regardless of study design). Even if you have never heard of the disease described in the study, don’t get thrown off. Most of the biostats question could literally be in another language and it wouldn’t change things at all. Sometimes they even use generic terms like “disease X” for a biostats problem.
So don’t get confused by whatever disease or study they are discussing in the problem. In fact, if you immediately notice that a question is a biostats question you may want to skip to the end of the question and not even read the first few sentences. Whenever I see a two by two table or a list of percentages as answer choices I don’t even read the question. It is not a huge time saver, but every little bit counts for an 8 hour exam.
If you are going to save a few seconds by skipping over the paragraph long question, I suggest you spend that extra time on double checking your work and avoiding dumb math error for any questions that require calculations. This is a VERY long test! Your brain will be fried by the end of the exam so making really dumb math error is much more likely. I suggest writing out the formula for each question and double checking your math to avoid dumb mistakes. You will be provided a small erasable board during the exam where you can take these notes. I also use the calculator built into the testing program for even basic calculations. I would hate to get a question wrong because I wasn’t focused and messed up simple addition in my head.
Thankfully you do have some room for error here. The answers for biostatistics or epidemiology questions with calculations are usually pretty far apart. They won’t give you multiple answers that are really close. You will never see a list of answers that looks like this; 43.5%, 43.9%, 44.1%, 44.6%, 45.1%. They will likely give you answers that look something like this; 5%, 23%, 43%, 51%, 95%. The answers aren’t close enough that a small rounding error will mess you up. Forget about fretting over those significant figures like in undergrad!
If your final answer isn’t anywhere near one of the answers given, it is probably a sign that you need to double check your work. That being said, just because your final answer matches a choice given doesn’t mean you can automatically be confident it is correct. Sometimes they will choose their answers specifically to trip you up. For example, if the question is asking for sensitivity and you accidentally calculate specificity, the answer you got may be listed. They know it possible for people to accidentally confuse the formula for similar items like specificity and sensitivity, so they want to make sure their questions differentiate between the two concepts to make sure you really know the difference.
Sometimes they make you calculate the final answer yourself. Other times they list a formula that is one or two steps before the final answer. For example, you might see something like “79%” as an answer or something like “110/140.” I think when they give you this formula, they are trying to be nice and save you one step of calculation. However, this can sometimes backfire since there are numerous different ways to find any given answer. So you may have gone about the question in a slightly different way and gotten an answer that isn’t listed. For example, you might have done the problem one way and gotten down to “157/200,” but that answer doesn’t match anything listed in the question. Don’t freak out if this happen! Just do the math and find the final answer for your answer and all of the answers given (there is usually 4-5 so it won’t take long). Then you will be able to figure out that your answer “157/200” actually matches “110/140,” because they both come out to about 78.5%. You just did your math a bit differently than the person who wrote the answers for the question.
Other Videos in This Section:
1. Standard Deviation, Mean, Median & Mode
2. 2×2 Table FP, FN, TP & TN
3. Sensitivity, Specificity & Screening Tests
4. Negative & Positive Predictive Value
5. Incidence, Prevalence and Case-Fatality Rate
6. Bias & Validity
7. Confounding, Randomization & Blinding
8. Types of Study Design
9. Odds Ratio and Relative Risk
10. Number Needed to Treat & Absolute Risk Reduction
11. p-Value, Statistical Significance and Types of Error
12. Confidence Interval Interpretation