First, a recruiter called me and asked about my background and asked me some easy stats questions.
Then I got an email to schedule a zoom interview with a biostatistician. In that interview, they spent the first 10-15 minutes talking about the company and role, then asked me 30+ minutes of statistics questions.
Interview questions [9]
Question 1
(First interview) You have mean, median, standard deviation, correlation. (a) If the scale was off by 5 units, which statistics would you have to recompute? (b) if you wanted to change from standard to metric, which statistics would you have to recompute?
(First interview) Assuming the population is normally distributed, answer each statement true or false.
(a) When viewed as a random variable, the sample mean follows a normal distribution.
(b) The population mean is always equal to the population median.
(c) The sample mean is always equal to the sample median.
(d) The MLE of the population standard deviation is unbiased.
(First interview) Answer the statements true/false.
(a) The law of large numbers says that as the sample size increases the distribution of the sample mean will become closer to the normal distribution.
(b) When performing simple univariate linear regression, the slope of the regression line after regressing X on Y is always the same as the slope of the regression line after regression Y on X.
(c) If a father whose height is far above average has a son whose height is barely above average, this can be explained by the regression effect.
(d) The distribution of the sample standard deviation follows the t-distribution with the number of degrees of freedom equal to the sample size minus one.
(Second interview) Tell me in your own words, pretending that your audience is a non-statistician, what is simple linear regression (or simple univariate regression) and what might you use it for?
(a) What are the assumptions or conditions that need to be met to use this model?
(b) What is the mathematical model that we are trying to fit?
(c) Let’s say you have a bunch of data, like (x, y) pairs. Mathematically, how do you actually fit the model? How do you transform those x’s and y’s into a slope and intercept?
(d) Let’s say that you have some data, you check all the assumptions and conditions, and everything looks good except there doesn’t seem to be normality. Can you still use the model? Or can you use parts of it and not others?
(Second interview) What is a p-value?
(a) Mathematically, how do you compute it?
(b) Some journals have banned reporting p-values in articles they’ve published. Can you describe what a reasonable person’s objections might be to using p-values?
(c) If you were going to present results without a p-value, what other statistical tool would you use to communicate the same thing?
(Second interview) You have a trial for a drug to lower blood pressure. The treatment group gets the new drug, control gets the old drug. Conduct experiment under optimal conditions, p-value is really really tiny. Can we conclude the old drug is more effective than the old drug?
A recruiter called me for a phone interview where I answered basic questions about my background and projects, and then did a short quiz. The rest of the process is exactly as how the other posts describe, including the recruiter not responding to emails and ghosting. Don't waste your time.
First introducing the company, and giving you some time to ask questions. Then presents technique statistical questions. Statistical questions are simple but very theoretical, which they asked about the definition about statistical terms.
Interview questions [1]
Question 1
What is p-value?
Why single linear regression assumption has normality assumption?
