# Business Analyst Interview Questions — Statistics & Probability Source: Codeayan (https://codeayan.com) Canonical: https://codeayan.com/get-hired/business-analyst/statistics Questions: 25 Last updated: 2026-08-24 Licence: free to read. Please cite Codeayan when quoting. --- ## 1. When would you report the median instead of the mean, and what is skew doing to those two numbers? *Easy · Very Common* **Short answer.** Report the median when the distribution is skewed or has extreme values, because the mean is pulled towards the long tail while the median tracks the typical case. In a right-skewed distribution the mean sits above the median. The mean still wins when you need totals, since mean times count gives the sum. Take monthly household incomes in a colony of ten families, in rupees: ``` 22,000 24,000 25,000 27,000 28,000 30,000 31,000 33,000 35,000 4,20,000 ``` The mean is ₹67,500. Nine of the ten families earn less than half that. The median is ₹29,000, which describes almost everyone in the colony. The mean is not wrong; it is answering a different question. It is the total divided by the count, so it tells you what each family would get if the colony’s entire income were shared equally. Nobody asked that. Skew is the diagnostic. In a right-skewed distribution, a long tail on the high side drags the mean above the median. Left-skewed pulls the mean below. When mean and median are close, the distribution is roughly symmetric and the choice barely matters. The mode is the most frequent value, and it earns its place on categorical data where a mean is meaningless, or where you care about the most common outcome rather than a central one. A garment factory reporting the modal shirt size is answering a real production question that a mean size cannot. The decision that goes wrong: a business quoting “average customer spend” from a right-skewed distribution, setting a pricing tier around it, and discovering most customers spend far less. The average was reported honestly and interpreted as typical, which it was not. One thing worth saying in the room: ask what the number will be used for. For forecasting total revenue you need the mean, because totals are what you are adding up. For describing a typical customer you need the median. **Likely follow-ups** - If the mean is the wrong summary, what would you show alongside the median? - A survey has a bimodal response — does the median help there? - When would you deliberately want the mean despite heavy skew? --- ## 2. Why do we square the deviations when computing variance? Why not just take the average distance from the mean? *Easy · Very Common* **Short answer.** Deviations from the mean always sum to zero, so averaging them raw gives nothing. Squaring removes the sign, and unlike absolute values it is mathematically well behaved and weights large deviations more heavily. Standard deviation is the square root, which returns the number to the original units. Monthly rainfall in millimetres at a station, over five months: ``` 80 120 100 60 140 ``` The mean is 100. The deviations are −20, +20, 0, −40, +40, and they sum to exactly zero. That is not a coincidence; deviations from the mean always cancel, by construction. So a plain average of deviations is useless as a spread measure. Two fixes exist. Take absolute values, giving mean absolute deviation of 24. Or square them: ``` (400 + 400 + 0 + 1600 + 1600) / 5 = 800 ``` That 800 is the variance, in squared millimetres, which is not a unit anyone can picture. Take the square root and you get a standard deviation of about 28.3 mm, back in the original units and directly comparable to the mean. So why squaring rather than absolute values? Two reasons worth giving. Squaring penalises large deviations disproportionately, which matches how risk usually behaves: one month 40 mm off matters more than two months 20 mm off. And squared deviations behave well under addition, which is what makes variances of independent quantities add up and makes almost all downstream statistics tractable. Absolute deviation is a perfectly valid measure and is genuinely more robust to outliers; it is simply harder to build theory on. The misconception to correct: standard deviation is not the average distance from the mean. It is the square root of the average squared distance, which is always at least as large, and larger when the deviations are uneven. For comparing variability across different units, use the coefficient of variation, standard deviation divided by mean, which is unitless. **Likely follow-ups** - If squaring exaggerates large deviations, is that a bug or a feature? - Two datasets have the same standard deviation but very different shapes — how? - How would you compare variability between two things measured in different units? --- ## 3. Someone hands you a column and asks which values are outliers. How do you decide? *Medium · Very Common* **Short answer.** There is no universal definition, only rules. The IQR rule flags points beyond 1.5 times the interquartile range below Q1 or above Q3. The z-score rule flags points more than about three standard deviations from the mean, but the mean and standard deviation are themselves distorted by the outliers you are hunting. Start with a question the interviewer will respect: outlier relative to what? A ₹40,000 monthly electricity bill is absurd for a two-bedroom flat and unremarkable for a cold-storage unit. Without a stated population, “outlier” has no meaning. Two rules, and they disagree in a specific way. The IQR rule uses quartiles. With Q1 = ₹1,200 and Q3 = ₹2,400, the IQR is ₹1,200, so the fences sit at 1,200 − 1,800 = −₹600 and 2,400 + 1,800 = ₹4,200. Anything above ₹4,200 is flagged. The z-score rule uses the mean and standard deviation, flagging anything beyond roughly three standard deviations. The problem is circularity: a handful of extreme bills pull the mean up and inflate the standard deviation, which widens the threshold, which lets those very values pass. Quartiles barely move when the tails are extreme, which is why IQR is the more robust choice for detection. IQR rule z-score rule Based on quartiles mean and SD Affected by the outliers barely strongly Assumes symmetry not really roughly The 1.5 multiplier is a convention from symmetric distributions, not a law. On a strongly right-skewed quantity like electricity billing, it flags a large chunk of the legitimate upper tail. On skewed data, work on a log scale, or use raw percentile cut-offs such as the 1st and 99th. What to do with what you find is a separate decision, and interviewers listen for whether you treat it as one. A bill of ₹0 is a meter error and should be corrected or excluded. A bill of ₹38,000 from an industrial connection wrongly sitting in a residential dataset is a population problem, not an outlier problem. A genuinely large bill from a large house is real, and deleting it teaches your analysis that such houses do not exist. If a rule flags 8% of your rows, the rule is wrong for that distribution. **Likely follow-ups** - You flag 8% of your rows as outliers — what does that tell you? - Would you drop them, cap them, or keep them? - How would you find outliers when the distribution is heavily skewed? --- ## 4. Why does the sample variance divide by n minus 1 instead of n? *Medium · Very Common* **Short answer.** Because you estimated the mean from the same sample. The sample mean sits closer to your data points than the true population mean does, so squared deviations from it are systematically too small. Dividing by n−1 rather than n corrects that bias. The n−1 is the degrees of freedom left after estimating the mean. Population parameters describe every unit you care about; sample statistics estimate them from a subset. The correction exists because one of those estimates gets reused. Take four tea estates sampled from a district, with yields in quintals per hectare: ``` 18 22 26 34 ``` The sample mean is 25. Squared deviations from 25: 49, 9, 1, 81, summing to 140. Divide by n = 4 and you get 35. Divide by n − 1 = 3 and you get 46.7. Here is why the second is right. If the true district mean were, say, 27, the squared deviations from 27 would sum to 81 + 25 + 1 + 49 = 156, larger than 140. That is not a coincidence about this example. The sample mean is, by definition, the value that minimises the sum of squared deviations for your particular sample. Any other value, including the true population mean, gives a larger sum. So deviations measured from the sample mean are always at least slightly too small, and dividing by n understates the variance systematically. Degrees of freedom is the other way to say it. Once you know the mean and any three of the four values, the fourth is determined. Only three deviations are free to vary, so you divide by three. The correction matters when n is small and fades fast. At n = 5, the difference is 25%. At n = 500, it is 0.2%, which nobody will notice. One honest caveat interviewers appreciate: dividing by n − 1 gives an unbiased estimate of the variance, but taking the square root of it does not give an unbiased estimate of the standard deviation, because the square root is not a linear operation. Almost everyone uses it anyway, and the residual bias is small. If you genuinely have the whole population, divide by n. **Likely follow-ups** - At n = 5,000 does the correction matter at all? - Does the same correction apply when you take the square root for standard deviation? - If you had data on the entire population, which would you use? --- ## 5. What is the difference between two events being independent and two events being mutually exclusive? *Easy · Very Common* **Short answer.** Mutually exclusive means the two cannot happen together, so their joint probability is zero. Independent means knowing one occurred tells you nothing about the other, so the joint probability equals the product of the individual probabilities. They are not related ideas, and for events with non-zero probability they cannot both be true. These are confused constantly, and the confusion is worth clearing with numbers. A machine on a production line produces a component. Event A is that it fails a dimensional check, with P(A) = 0.05. Event B is that it fails a surface finish check, with P(B) = 0.08. **Mutually exclusive** would mean a component can fail one check or the other but never both. Then P(A and B) = 0. **Independent** means failing the dimensional check tells you nothing about the surface finish. Then P(A and B) = 0.05 × 0.08 = 0.004, which is not zero. Notice they cannot both hold. If the events are mutually exclusive, knowing A occurred tells you B definitely did not, which is a very strong dependence. So for events with non-zero probability, mutual exclusivity implies dependence. That is the point an interviewer is usually driving at. Mutually exclusive is a statement about whether outcomes overlap. Independent is a statement about whether one carries information about the other. The rules follow from that. For mutually exclusive events, P(A or B) = P(A) + P(B). Otherwise you must subtract the overlap: P(A or B) = P(A) + P(B) − P(A and B). Add 0.05 and 0.08 without subtracting when the events can co-occur, and you have double-counted the components that failed both. For independent events, P(A and B) = P(A) × P(B). Multiply probabilities that are not independent and the answer is wrong, usually badly. If a batch runs hot, both checks are more likely to fail together, so the true joint probability might be 0.02 rather than 0.004 — five times what independence predicted. The practical check: compare the observed joint frequency against the product of the marginals. If they differ materially, the events are not independent, whatever the process design claims. **Likely follow-ups** - Can two events with non-zero probability be both independent and mutually exclusive? - If P(A and B) is 0.2, are A and B independent? - How would you actually check independence from real data? --- ## 6. Define conditional probability, and show me where the multiplication rule comes from. *Medium · Very Common* **Short answer.** P(A given B) is the probability of A within the restricted world where B has already occurred, computed as P(A and B) divided by P(B). Rearranging gives the multiplication rule: P(A and B) equals P(B) times P(A given B). Conditioning shrinks the denominator to the cases where B happened. Conditioning means changing what you are dividing by. A bank reviews 10,000 loan applications. 2,000 come from applicants with no formal credit history. Of those 2,000, 300 eventually default. Across all applications, 700 default in total. The unconditional default rate is 700 / 10,000 = 7%. The default rate given no credit history is 300 / 2,000 = 15%. Nothing about the data changed. You restricted attention to a subgroup, so the denominator shrank from 10,000 to 2,000 and only the defaults inside that subgroup count in the numerator. That is all conditioning is. Written out: ``` P(default | no history) = P(default AND no history) / P(no history) = 0.03 / 0.20 = 0.15 ``` Multiply both sides by P(no history) and you have the multiplication rule: ``` P(A AND B) = P(B) × P(A | B) ``` That rule is how you chain events. The probability that a randomly picked application has no credit history *and* defaults is 0.20 × 0.15 = 0.03, matching the 300 out of 10,000 you started with. Two things worth stating in the room. P(A given B) and P(B given A) are different quantities and confusing them is one of the most consequential errors in applied statistics. Here, P(default | no history) is 15%, while P(no history | default) is 300 / 700 = 43%. Both are correct and they answer completely different questions. And when A and B are independent, P(A given B) reduces to P(A), so the multiplication rule collapses to the familiar product. Independence is the special case, not the default assumption. Sampling without replacement is the everyday example where conditioning is unavoidable, since the second draw’s probabilities depend on what the first draw removed. **Likely follow-ups** - Is P(A given B) ever equal to P(B given A)? - How would this change if the two events were independent? - If I sample without replacement, what happens to the second draw? --- ## 7. A screening test is 99% accurate and the disease affects 1 in 1,000 people. Your test comes back positive. How worried should you be? *Hard · Very Common* **Short answer.** Less than most people assume. With a prevalence of 1 in 1,000 and a test that is 99% accurate in both directions, a positive result means roughly a 9% chance of actually having the disease. False positives from the large healthy group vastly outnumber true positives from the tiny infected group. Work it with a concrete population rather than the formula. Imagine 100,000 people screened for a rare condition. Prevalence is 1 in 1,000, so 100 people actually have it and 99,900 do not. The test catches 99% of the sick: **99 true positives**, and 1 missed. The test correctly clears 99% of the healthy, so it wrongly flags 1%: 1% of 99,900 is **999 false positives**. Total positives: 99 + 999 = 1,098. Of those, only 99 are genuinely ill. ``` P(disease | positive) = 99 / 1,098 = 0.090 ``` About 9%. Ninety-one out of every hundred people receiving a positive result do not have the condition. Nothing is wrong with the test. The arithmetic is driven by the base rate: the healthy group is a thousand times larger, so even a small error rate applied to it produces far more false positives than the small sick group produces true positives. Bayes’ theorem is just this bookkeeping written compactly: ``` P(D|+) = P(+|D) × P(D) / P(+) ``` The piece people drop is P(D), the prior. Reading “99% accurate” as “99% chance I have it” is confusing P(positive | disease) with P(disease | positive), and those differ by a factor of ten here. Two things that follow. Screening a general population for a rare condition produces mostly false alarms, which is why confirmatory testing exists and why mass screening programmes are designed carefully rather than run on everyone. And raising the prior changes everything: test only people with symptoms, where prevalence might be 20% instead of 0.1%, and the same test gives a positive predictive value above 96%. The same trap appears in fraud alerts, security scanning and rare-event prediction generally. Any rare-event classifier evaluated on accuracy alone will look excellent and be nearly useless. **Likely follow-ups** - What would raise that 9% figure meaningfully — a better test or a different population? - How does a second independent positive test change the answer? - Where else does this same arithmetic mislead people outside medicine? --- ## 8. What is the empirical rule, and what does it let you say about a normally distributed measurement? *Easy · Very Common* **Short answer.** For a normal distribution, about 68% of values fall within one standard deviation of the mean, about 95% within two, and about 99.7% within three. It lets you convert a mean and standard deviation into a statement about how often values of a given size occur. A cement bagging line fills bags with a mean weight of 50.0 kg and a standard deviation of 0.4 kg, and the process is approximately normal. Range Weights Share of bags ±1 SD 49.6 to 50.4 kg about 68% ±2 SD 49.2 to 50.8 kg about 95% ±3 SD 48.8 to 51.2 kg about 99.7% So a bag under 48.8 kg happens roughly 15 times in 10,000, since 0.3% falls outside three standard deviations and half of that is on the low side. That number turns a vague worry about underfilling into a rate you can put against a regulatory limit. The z-score generalises it. A bag at 50.9 kg is (50.9 − 50.0) / 0.4 = 2.25 standard deviations above the mean, which lets you compare it against any other normally distributed quantity regardless of units. Two corrections worth making unprompted. The normal distribution is a model, not a property real data possesses. Bag weights are approximately normal because many small independent factors add up in the filling mechanism. Quantities that cannot go below zero and have a long upper tail, such as claim sizes or waiting times, are not normal and the empirical rule badly understates their extremes. And “95%” is a rounding of 1.96 standard deviations, not exactly 2. It matters when you are constructing intervals rather than describing data. The failure this prevents: applying three-sigma control limits to a right-skewed quantity. The rule predicts 0.15% of points above the upper limit; on a skewed distribution you might see 3%, and the operators learn to ignore the alarm. Check the shape before applying the rule. A histogram plus a Q-Q plot tells you more than any single test. **Likely follow-ups** - Your data isn't normal — does the empirical rule still roughly hold? - How would you check normality without just eyeballing a histogram? - Where does the normal distribution get assumed without anyone saying so? --- ## 9. State the central limit theorem precisely. What is it a claim about, and what is it not? *Hard · Very Common* **Short answer.** The central limit theorem says the sampling distribution of the sample mean approaches a normal distribution as sample size grows, regardless of the population's shape, provided the population has finite variance. It says nothing about the data itself becoming normal, and nothing about any individual sample. The misstatement is everywhere, so state it carefully: the CLT is about the distribution of a statistic across hypothetical repeated samples, not about the distribution of your observations. Daily order values on a food ordering platform are heavily right-skewed. Most orders are ₹200 to ₹500, a few are ₹4,000, and the histogram has a long tail. That histogram never becomes normal no matter how much data you collect. Collect a crore orders and it is still right-skewed, just better resolved. What does become normal is this: take a sample of 200 orders and compute the mean. Do it again with a different 200, and again, thousands of times. Plot those means. That distribution is approximately normal and centred on the true population mean, with standard deviation: ``` SE = sigma / sqrt(n) ``` If the population standard deviation is ₹600, the standard error at n = 200 is 600 / 14.14 = ₹42.4. So sample means cluster within roughly ±₹85 of the truth about 95% of the time, even though individual orders scatter far more widely. This is why a t-test on skewed data at large n is defensible. The test needs the sampling distribution of the mean to be approximately normal, not the data. Three conditions people skip. Observations must be independent and identically distributed. Clustered data, such as several orders from the same household, breaks it and the effective sample size is smaller than the row count. The population needs finite variance. Genuinely heavy-tailed quantities where variance is undefined do not obey the standard CLT. And “n = 30 is enough” is a rule of thumb, not a theorem. For mild skew, 30 is fine. For the order-value distribution above, you might need several hundred before the sampling distribution looks convincingly normal. The heavier the skew, the larger n has to be. **Likely follow-ups** - Someone says their data became normal because of the CLT — what would you tell them? - With a heavily skewed population, is n = 30 enough? - Does the CLT say anything about the median? --- ## 10. Explain the main sampling methods and when you'd pick one over another. *Medium · Very Common* **Short answer.** Simple random sampling gives every unit an equal chance. Stratified divides the population into groups and samples within each, improving precision when groups differ. Cluster samples whole groups, which is cheaper but less precise. Systematic takes every kth unit, which is easy and fails if the list has a periodic pattern. A state education department wants to estimate average learning outcomes across 12,000 schools. **Simple random.** Draw 400 schools at random from the full list. Unbiased and simple, but the surveyors travel to 400 scattered locations, and by chance you might get very few schools from one region. **Stratified.** Divide schools into strata by type or region, then sample within each proportionally. If outcomes differ substantially between government and private schools, this is materially more precise than simple random at the same sample size, because you have eliminated the between-stratum variation from your sampling error. It also guarantees every stratum appears, which simple random does not. **Cluster.** Pick 40 districts at random and survey every school in them. Far cheaper, since travel collapses. The cost is precision: schools within a district resemble each other, so 400 schools from 40 districts carry less information than 400 scattered schools. The effective sample size can be a fraction of the nominal one, and treating it as 400 independent observations makes your confidence interval too narrow. **Systematic.** Sort the list and take every 30th school. Easy to execute and usually fine. It fails when the ordering has a cycle matching your interval. Sample every 7th day of transactions and you get the same weekday every time, which is a real and well-documented way to produce a badly biased sample from a procedure that looks rigorous. The trade-off to state plainly: stratification buys precision when strata differ; clustering buys cost savings and pays in precision. Choose based on whether budget or accuracy is the binding constraint. What none of these fix is non-response and coverage bias. If the sampling frame omits unregistered schools, or if the schools that decline to participate differ systematically from those that agree, a perfect sampling method still gives a biased estimate. That is the failure worth flagging, because it survives every technical improvement to the sampling design. **Likely follow-ups** - Cluster sampling is cheaper but you said it costs precision — how much, and why? - Where does systematic sampling go badly wrong? - What kind of bias does none of these methods fix? --- ## 11. Give me an example of a dataset that looks representative and isn't. What kinds of bias would you look for? *Medium · Very Common* **Short answer.** Selection bias means the sample was drawn in a way that favours certain units. Survivorship bias means the failures dropped out before you looked. Non-response bias means those who declined differ systematically from those who answered. All three produce confident, precise, wrong estimates. A gym chain surveys members about satisfaction and finds 84% rate it highly. The survey went out by email to members who had visited in the past month. Everyone who stopped coming, which is the group whose dissatisfaction you most needed to measure, is absent from the sample by construction. That is **survivorship bias**: the units that failed are no longer in the population you can see. **Selection bias** is the broader version. A political poll conducted through a smartphone app reaches people with smartphones and the app. A study of startup practices that samples currently operating startups excludes every one that shut down. **Non-response bias** is the one that survives good sampling design. Draw a perfect random sample of 2,000 members, and if only 400 respond, you have a self-selected subsample of 400. People with strong opinions in either direction respond more, and the middle stays quiet. Here is why more data does not help. Increase the survey from 400 to 40,000 responses collected the same way, and the standard error shrinks by a factor of ten while the bias stays exactly where it was. You end up with a tighter confidence interval around a wrong number, which is worse than a wide interval around it, because it invites confidence. That distinction is what interviewers are testing: sampling error shrinks with n, bias does not. What you can do. Compare respondents against the frame on variables you have for everyone: joining date, membership tier, visit frequency. If respondents visit twice as often as the full membership, you know the direction of the problem even if you cannot fix it. Follow up a random subsample of non-responders more aggressively, and compare them against the original responders. Weight by known population characteristics, accepting that weighting corrects only for the variables you measured. And state the limitation in the report. An estimate labelled as covering active members only is useful; the same estimate presented as member satisfaction is misleading. **Likely follow-ups** - You can't fix the bias with more data — why not? - How would you check whether your non-responders differ from your responders? - Which of these three is hardest to detect after the fact? --- ## 12. What is the difference between the standard deviation and the standard error? *Medium · Very Common* **Short answer.** Standard deviation describes how spread out the individual observations are. Standard error describes how much a sample statistic, usually the mean, varies from sample to sample. Standard error equals the standard deviation divided by the square root of the sample size, so it shrinks as you collect more data while the standard deviation does not. They get confused because both are measures of spread, but they describe entirely different things. A logistics firm measures delivery times for 900 parcels. The mean is 38 hours, the standard deviation is 15 hours. **Standard deviation = 15 hours.** Individual parcels vary a lot. Some arrive in 20 hours, some in 60. That number describes the parcels. **Standard error = 15 / sqrt(900) = 0.5 hours.** If you repeated this exercise with a fresh sample of 900 parcels, the sample mean would land somewhere near 38 hours, typically within about an hour. That number describes how precisely you have pinned down the mean. ``` SE = s / sqrt(n) ``` Collect 90,000 parcels instead and the standard error drops to 0.05 hours, while the standard deviation stays at 15. More data makes you more certain about the average; it does not make deliveries more consistent. The square root is the part with practical consequences. To halve the standard error you need four times the data. To get it down by a factor of ten you need a hundred times the data. That is why precision gets expensive quickly, and why a proposal to “just collect more data” needs a number attached. Where the confusion causes real damage: an error bar labelled ambiguously. A chart showing mean delivery time with ±0.5 hour bars, read as if it described parcels, tells a manager that deliveries are extremely consistent. They are not; that bar describes the estimate, not the parcels. Label which one you plotted. Choose by the question. Describing the customer experience needs the standard deviation, or better, percentiles. Stating how confident you are in the reported average needs the standard error. Standard error is not exclusive to means. Any statistic has one, though the formula differs. **Likely follow-ups** - To halve your standard error, how much more data do you need? - Does the standard error keep shrinking forever as n grows? - Which of the two would you show a stakeholder describing their customers? --- ## 13. Your analysis reports a 95% confidence interval of 3.1% to 4.7%. What does that actually mean? *Hard · Very Common* **Short answer.** It means the procedure that produced this interval captures the true parameter 95% of the time across repeated samples. This particular interval either contains the true value or it does not. It says nothing about the probability of individual values inside it, nothing about where future observations fall, and nothing about practical importance. The correct statement is about the method, not about this interval. Imagine drawing 100 independent samples and constructing an interval from each. About 95 of those intervals would contain the true population value. You have one of them, and you do not know whether it is one of the 95 or one of the 5. The parameter is a fixed number; the interval is what varies. Three misreadings, each worth correcting explicitly. **“There’s a 95% probability the true value is between 3.1 and 4.7.”** Under frequentist reasoning the parameter is not random, so it has no probability of being anywhere. It either is or is not in this interval. A Bayesian credible interval does support that phrasing, and it is a different object built on different assumptions. Naming that distinction is what separates a strong answer. **“95% of the data falls in this range.”** No. That would be a prediction interval, and it is far wider. With a defect rate estimate of 3.9% ± 0.8%, individual batches vary far more than that. **“The values in the middle are more likely.”** The interval carries no distribution of belief inside it. Values near the edge are not less plausible in any statement the interval makes. The width itself is the useful information, and it is underused. An interval of 3.1% to 4.7% says the estimate is reasonably tight. An interval of 0.4% to 12.2% around the same point estimate of 3.9% says you know almost nothing, even though both report the same central figure. On the zero question: an interval containing zero means you cannot rule out no effect at this sample size. It does not establish that the effect is zero. An interval running from −0.2% to +9.5% is entirely consistent with a large positive effect and is simply too imprecise to say. Narrower comes from more data, less variability, or a lower confidence level, and the last one is buying width by accepting more misses. **Likely follow-ups** - So can I say there's a 95% chance the true value is between 3.1 and 4.7? - The interval includes zero — does that prove there's no effect? - What would make this interval narrower besides collecting more data? --- ## 14. How would you set up the null and alternative hypotheses for a question like "does the new packaging reduce spoilage"? *Easy · Very Common* **Short answer.** The null is the position of no effect or no difference, and it is what you assume until the data argues otherwise. The alternative is what you are claiming. Set them before seeing the data, because choosing the hypothesis after looking is how a chance pattern becomes a finding. For the packaging question, with spoilage rates: ``` H0: p_new = p_old (no difference in spoilage rate) H1: p_new < p_old (new packaging reduces spoilage) ``` The null is the boring, sceptical position. It exists because you can only ever gather evidence against something, never conclusively for it. The logic is a proof by contradiction with probability attached: assume no difference, ask how surprising the observed data would be under that assumption, and reject only if the answer is “very”. That asymmetry explains a question interviewers like. Why not make “the new packaging works” the null? Because failing to reject a null is not evidence for it. If your claim were the null and the test came back inconclusive, you would be treating “we could not disprove it” as support, which it is not. Putting the sceptical position in the null means the burden of proof sits with the claim. Three things to get right. State them before you see the data. Deciding on a one-tailed test in the direction the data happens to point doubles your effective false positive rate, and it is a well-known and easily detected form of cheating. Choose one-tailed only when a change in the other direction would lead to the same decision as no change. If packaging that increased spoilage would matter, use two-tailed. And a “fail to reject” outcome is not proof of equivalence. It means insufficient evidence at this sample size. If the business genuinely needs to establish that the two are the same, that requires an equivalence test with a stated margin, which is a different design. Ask the interviewer what decision follows from the result. A test with no decision attached usually has the wrong hypothesis. **Likely follow-ups** - Why can't you just make your claim the null hypothesis? - When would you use a one-tailed test rather than two-tailed? - If you fail to reject the null, what have you actually shown? --- ## 15. Define a p-value. Then tell me the ways people get it wrong. *Hard · Very Common* **Short answer.** A p-value is the probability of observing data at least as extreme as yours if the null hypothesis were true. It is not the probability the null is true, not the probability your result was a fluke, and not a measure of effect size. It is a statement about data under an assumption, not about the assumption. The definition has to be said carefully because the wrong version is more common than the right one. A fertiliser trial reports p = 0.03 for a yield difference. What that means: if the fertiliser genuinely had no effect, you would see a difference this large or larger in about 3% of trials of this size, purely through sampling variation. Four misreadings, and each has cost someone an offer. **“There’s a 3% chance the null is true.”** That would require a prior probability of the null, which the p-value never uses. This is the same conditional-probability inversion as reading a positive test result as the probability of disease. **“There’s a 97% chance the effect is real.”** Same error, restated. It also ignores that a real effect could be far smaller than the one you measured. **“p = 0.001 means a bigger effect than p = 0.04.”** No. The p-value depends on effect size and sample size together. A trivial difference across two crore observations produces a tiny p-value. Report the effect size and its interval alongside. **“p = 0.049 is a finding and p = 0.051 is nothing.”** The 0.05 threshold is a convention with no natural basis. Those two results are nearly identical evidence, and treating them as categorically different is what drives selective reporting. Two structural points that matter more than the definition. Multiple testing destroys the interpretation. Fourteen comparisons at the 5% level gives roughly a 51% chance of at least one result below 0.05 with nothing real going on. One hit out of fourteen is what chance produces, and reporting it as a finding without saying how many tests were run is the most common form of misleading analysis in industry. And p = 0.06 does not mean the feature does not work. It means insufficient evidence at this sample size. Ask what the confidence interval looks like: if it runs from −1% to +14%, the honest answer is that the study was underpowered, not that the effect is absent. **Likely follow-ups** - If I showed you p = 0.06, would you tell the team the feature doesn't work? - What does p = 0.001 tell you about the size of the effect? - You ran fourteen comparisons and one came back at 0.03 — what would you say? --- ## 16. Explain Type I and Type II error, and tell me which one you'd rather make in a fraud detection system. *Medium · Very Common* **Short answer.** Type I is a false positive: rejecting a true null, concluding there is an effect when there is none. Type II is a false negative: failing to detect a real effect. Lowering one raises the other at a fixed sample size, so the choice depends on which mistake costs more in your situation. Null is true Null is false **Reject null** Type I error (α) correct **Fail to reject** correct Type II error (β) Power is 1 − β, the probability of detecting a real effect when one exists. For fraud detection, map the errors onto real consequences before answering. A **false positive** blocks a legitimate transaction. The customer at a payment terminal is embarrassed, calls support, and possibly stops using the card. Cost: support handling, some churn, reputational damage. A **false negative** lets a fraudulent transaction through. Cost: the transaction value, plus chargeback handling, plus regulatory exposure if it is systematic. Put numbers on it. If a blocked legitimate transaction costs roughly ₹400 in support and churn, and a missed fraud averages ₹22,000, then one missed fraud costs as much as fifty-five false alarms. That ratio says tolerate false positives and tighten against misses. Now flip the domain. For a screening test that triggers an invasive follow-up procedure, false positives carry real medical risk and the calculation runs the other way. The point interviewers are testing is that neither error is universally worse. Anyone who answers “Type I is more serious” without asking about the application has recited a textbook. The trade-off is real at fixed sample size. Lowering the threshold to catch more fraud necessarily flags more legitimate transactions; there is no threshold setting that reduces both. The only way to improve both simultaneously is a better model or more data, which is a different conversation from where to set the cut-off. One thing worth naming: the base rate matters enormously here. When fraud is 0.1% of transactions, even a small false positive rate applied to the huge legitimate population produces far more false alarms than true catches, which is the same arithmetic that makes rare-disease screening produce mostly false positives. **Likely follow-ups** - You tighten the threshold to cut false positives — what happens to the other error? - How would you decide the relative cost of the two in rupees? - What does power actually mean in this framing? --- ## 17. Your test comes back significant with p = 0.002. The effect is a 0.3% improvement. What do you tell the business? *Medium · Very Common* **Short answer.** That the effect is real and probably too small to act on. Statistical significance says the difference is unlikely to be sampling noise; it says nothing about whether the difference is large enough to matter. With a very large sample, trivially small effects become significant. Significance and importance are independent questions, and conflating them is one of the most common analytical failures in industry. A bank tests a redesigned statement layout on 40 lakh account holders and finds that the digital payment rate rises from 62.0% to 62.3%. With that sample size, p = 0.002. The result is real. Now do the business arithmetic. A 0.3 percentage point lift across 40 lakh accounts is 12,000 additional accounts using digital payments. If each is worth ₹90 a year in reduced processing cost, that is around ₹11 lakh annually. Set against a redesign, a testing cycle and a rollout costing ₹60 lakh, the answer is no. The sample size is what made the p-value small, not the effect. Run the same test on 4,000 accounts and the same 0.3 point difference would be nowhere near significant. Nothing about the underlying effect changed; only your ability to detect it did. Two things to bring into the room. Report the effect size with its confidence interval, always, next to the p-value. An interval of 0.1% to 0.5% tells the business the plausible range of what they are buying. The p-value alone tells them only that it is not zero, which is rarely the question they asked. And decide the threshold that matters before running the test. Ask the business what lift would justify the investment, and use that as the minimum effect of interest. A test designed around a 2% target that returns 0.3% has answered the question clearly, and the answer is no. The reverse case deserves mention too. A large, commercially important effect measured on a small sample can come back non-significant, and dismissing it as “no effect” is equally wrong. The correct reading is that the study was too small to tell. **Likely follow-ups** - How would you decide what size of effect is worth acting on? - Would you have reached this conclusion with a smaller sample? - The confidence interval runs from 0.1% to 0.5% — does that change your answer? --- ## 18. Ice cream sales and drowning deaths move together across the year. Why, and how would you generalise the lesson? *Easy · Very Common* **Short answer.** Temperature drives both. Hot weather increases ice cream sales and also increases swimming, which increases drownings. Neither causes the other, and the correlation is real but produced by a third variable affecting both. Correlation constrains what causal stories are possible; it does not select among them. That example is famous, so use it to set up the general structure rather than dwelling on it. An observed correlation between A and B has four possible explanations, and the data alone cannot distinguish them. **A causes B.** The straightforward reading, and the one people jump to. **B causes A.** Reverse causation, and it is easy to miss. A hospital finds that patients receiving a certain intensive treatment have higher mortality. The treatment is given to the sickest patients; severity causes both the treatment and the outcome. **A third variable causes both.** The ice cream case. In business data this is everywhere: a company observes that customers who use its mobile app spend more. Engaged, higher-income customers both download apps and spend more. The app may add nothing. **Chance.** With enough variables compared, some pairs correlate strongly for no reason at all. Compare 200 metrics pairwise and you have 19,900 correlations; a few will look impressive. The confounder problem is what makes this hard rather than merely a caution. You can control for the confounders you thought of and measured. You cannot control for the ones you did not, and there is no statistical test that tells you an unmeasured confounder exists. That is a genuine limitation of observational data, not a technique problem. What establishes causation: randomised assignment, because randomising breaks the link between the treatment and every confounder, measured or not. Where randomisation is impossible, quasi-experimental designs such as difference-in-differences, instrumental variables or regression discontinuity make explicit assumptions that let you argue for causation, and those assumptions are the weak point to interrogate. The answer interviewers want is not “correlation isn’t causation” recited back. It is the question that follows: what would have to be true for the causal reading to hold, and can you check it? **Likely follow-ups** - You control for the confounder you thought of — how do you know there isn't another? - What if the causation runs the opposite way from what you assumed? - What would actually establish causation here? --- ## 19. Explain skewness and kurtosis to me without using a formula. *Medium · Common* **Short answer.** Skewness measures asymmetry: a long right tail gives positive skew, a long left tail negative. Kurtosis measures how much of the variance comes from rare extreme values rather than moderate ones, so high kurtosis means fat tails and more frequent extremes than a normal distribution would produce. Skewness is about which side the tail is on. Insurance claim amounts are the standard right-skewed case. Most claims are small, a few are enormous, and the tail stretches far to the right. Positive skewness. Age at retirement in a workforce is often left-skewed: most people retire near sixty, a few leave much earlier, so the tail runs left. Negative skewness. The practical consequence: with positive skew, mean above median; with negative skew, mean below. Skewness near zero means roughly symmetric, and that is all it means. It does not mean normal. Kurtosis is about the tails, not about “peakedness”, which is the description most textbooks give and it misleads people. It measures how much of the total variance is produced by rare extreme observations rather than by typical ones. A distribution of daily percentage changes in a commodity price usually has high kurtosis. Most days move slightly, and occasionally a day moves eight percent. The standard deviation calculated from ordinary days badly understates how large the extremes get. Normal distribution has a kurtosis of 3, or an excess kurtosis of 0, depending on which convention your software uses. Check which one, because a reported “kurtosis of 0” means something completely different under each. Where this bites: a risk model assuming normal-shaped tails on a fat-tailed quantity systematically underestimates how often large losses occur. The model looks fine for two years and then a single week produces a move the model called a once-in-a-century event. Both statistics are unstable on small samples. Below a few hundred observations, a skewness of 0.4 tells you very little, and a histogram tells you more than either number. **Likely follow-ups** - Your data has a skewness of 0.05 — does that mean it's normal? - Which of these would worry you more before fitting a linear model? - How would you reduce right skew before analysis? --- ## 20. A student is told she scored in the 90th percentile. What exactly does that mean, and what does it not mean? *Easy · Common* **Short answer.** It means roughly 90% of candidates scored at or below her mark. It says nothing about what she scored out of 100. Percentile is a rank position within a specific group, so the same raw marks give different percentiles in a different cohort or a different year. The confusion is between percentage and percentile, and it costs marks in interviews as often as it does in exam halls. A candidate scoring 62 out of 100 in a competitive exam where most candidates scored between 30 and 55 could easily be in the 90th percentile. Percentage is what she got. Percentile is where she stands. Quartiles are just three named percentiles: Q1 is the 25th, Q2 is the 50th and is the median, Q3 is the 75th. The interquartile range, Q3 minus Q1, spans the middle half of the data and is the standard robust measure of spread. Three things people get wrong. Percentiles are relative to a group. The same 62 marks might be the 90th percentile in one year and the 74th in a year with a stronger cohort. Any percentile figure without a stated reference population is incomplete. Percentiles are not evenly spaced in raw units. Moving from the 50th to the 55th percentile might take 3 extra marks, while moving from the 95th to the 99th takes 18, because scores bunch up in the middle and spread out at the top. So a five-percentile improvement means very different things at different points. And percentiles compress the extremes. Everyone above the 99th percentile is reported identically, whether they are marginally above or far beyond. Where this genuinely matters outside exams: reporting response times. Mean latency hides the experience of the slowest requests entirely. If p50 is 120 ms and p99 is 3 seconds, one in every hundred users waits three seconds, and no average will surface that. Report percentiles for anything where the tail is what the user feels. **Likely follow-ups** - Two students both sit in the 90th percentile — did they score the same marks? - Why does the same raw score give different percentiles in two different years? - How would you report a latency figure, and would you use the mean? --- ## 21. Walk me through how you'd use expected value to decide between two options. *Medium · Common* **Short answer.** Multiply each outcome by its probability and add them up. That gives the long-run average value of a decision if you could repeat it many times. It is the right criterion for repeated decisions with modest stakes, and a poor one for a single decision that could ruin you. A distributor is deciding whether to extend ₹8 lakh of credit to a new retailer. If the retailer pays, the distributor earns ₹1.2 lakh in margin. If the retailer defaults, ₹8 lakh is lost. Based on the credit assessment, the default probability is 8%. ``` EV = 0.92 × 1,20,000 + 0.08 × (−8,00,000) = 1,10,400 − 64,000 = ₹46,400 ``` Positive, so on average this decision makes money. Across a hundred such retailers, the distributor comes out roughly ₹46 lakh ahead even though eight of them will not pay. The break-even is worth computing too, because it is more useful than the point estimate. Set the expression to zero and solve for the default probability: it flips negative once the default rate rises above about 13%. Now you can ask whether 8% is plausibly off by five points, which is a far more answerable question than whether it is exactly right. Three limits to state in an interview. Expected value is a long-run average and there is no long run for a one-off decision. Extending ₹8 lakh once, to a business whose entire cash reserve is ₹8 lakh, is a bet with positive expected value that can end the business. This is why insurance exists and is rational despite having negative expected value for the buyer. The number inherits every assumption in the probabilities. If the 8% came from a model trained on a different retailer segment, the EV is arithmetic performed on a guess. And outcomes are rarely two-valued. Partial recovery through settlement, delayed payment with interest cost, and the value of a retailer who becomes a long-term customer all belong in a realistic version. Ask the interviewer whether this decision repeats. If it does, follow the expected value. If it does not, and the downside is unrecoverable, say so. **Likely follow-ups** - The expected values come out nearly equal — what would you look at next? - When would you deliberately ignore the expected value and pick the lower one? - How much does your answer depend on probabilities you guessed? --- ## 22. When would you model something as binomial and when as Poisson? *Medium · Common* **Short answer.** Binomial counts successes in a fixed number of independent trials, each with the same probability. Poisson counts events occurring in a fixed interval of time or space, where there is no natural number of trials. Poisson also approximates the binomial well when trials are many and the probability is small. The distinguishing question: is there a fixed denominator? **Binomial.** A telecom operator sends 500 service SMS messages; each has a 3% chance of failing to deliver. Fixed n = 500, fixed p = 0.03, and you are counting how many of those 500 fail. Expected failures = 500 × 0.03 = 15. Variance = np(1−p) = 14.55. **Poisson.** A helpdesk receives calls at an average of 12 per hour. There is no “number of trials” here; calls arrive continuously and there is no upper bound on how many could arrive in an hour. Expected = 12, and the defining property is that the variance also equals 12. That equality is the Poisson’s signature and its main diagnostic. If observed call counts have a mean of 12 and a variance of 40, the Poisson assumption is wrong. Usually that means the arrival rate is not constant, calls cluster after an outage, or the intervals are not comparable. The two connect. When n is large and p is small, binomial converges to Poisson with lambda = np. With n = 10,000 and p = 0.0002, computing binomial probabilities involves enormous factorials, while Poisson with lambda = 2 gives essentially the same answers and is far easier to work with. Both assume independence, and that is where real data most often breaks them. For the SMS case, if failures are caused by one network element going down, the messages routed through it fail together. The mean is still 15, and the variance is far larger than 14.55, so any interval or test built on binomial assumptions is too narrow. Manufacturing defects behave the same way when a bad raw material batch causes correlated failures. The practical check is to compare the observed variance against what the model predicts. Overdispersion is common enough that negative binomial is often the honest alternative to Poisson. **Likely follow-ups** - You have 10,000 trials and a probability of 0.0002 — which would you use and why? - What breaks if the events aren't independent? - What does it mean when the observed variance exceeds the Poisson mean? --- ## 23. How is the law of large numbers different from the central limit theorem? *Medium · Common* **Short answer.** The law of large numbers says the sample mean converges to the true mean as sample size grows. The central limit theorem describes the shape of the distribution around that convergence. One tells you where you are heading; the other tells you how far off you are likely to be at a given sample size. They answer different questions and get merged constantly. A quality inspector samples finished ceramic tiles. The true defect rate is 4%. **Law of large numbers.** Inspect 50 tiles and you might see 1 defect, or 5, so an observed rate of 2% or 10%. Inspect 50,000 and the observed rate will sit very close to 4%. The sample mean converges to the population mean as n grows. That is a statement about where the estimate goes. **Central limit theorem.** At n = 1,000, how far off might you be? The standard error of the proportion is: ``` SE = sqrt(0.04 × 0.96 / 1000) = 0.0062 ``` So roughly 95% of samples of 1,000 give an observed rate between about 2.8% and 5.2%. That is a statement about the spread around the true value at a given n, and it is what makes confidence intervals possible. Law of large numbers Central limit theorem Claims the mean converges the shape is normal Tells you where you end up how far off you are Enables trust in large samples intervals and tests The misconception both are used to justify is the gambler’s fallacy. If this month produced far more defects than expected, nothing “corrects” it. Future output is unaffected by past output, assuming the process has not changed. What happens is dilution: the excess becomes a smaller fraction of a growing total, so the running average drifts back toward 4% without any compensating run of good tiles. Say that clearly, because the phrase “it evens out” describes the wrong mechanism and interviewers listen for it. Neither theorem guarantees anything about one particular sample. Both describe behaviour as n grows or across repeated samples. **Likely follow-ups** - A process has produced far more failures than expected this month — will it "even out"? - Which of the two lets you build a confidence interval? - Does the law of large numbers guarantee anything about a single sample? --- ## 24. Most of your business metrics are right-skewed. When would a log transform help, and when would it not? *Medium · Common* **Short answer.** A log transform compresses the long right tail and often makes a multiplicative relationship additive, which helps when a technique assumes roughly symmetric residuals. It does not help when zeros or negatives are present, and it changes what your coefficients mean, so interpret on the log scale deliberately. Right skew is the normal condition for business quantities, not an anomaly. Time on a page, order values, city populations, page views per article, and firm revenues are all bounded below by zero with no upper bound, so the tail can only run one way. Take article page views on a news site over a month: ``` median: 1,400 mean: 9,800 max: 2,40,000 ``` The mean is seven times the median because a handful of viral pieces dominate. Any technique that treats the mean as typical will mislead. Logs help in two specific situations. **When the underlying relationship is multiplicative.** If a headline change increases views by 30% rather than by 400 views, that is multiplicative, and taking logs turns it into an additive effect that linear methods can handle. The coefficient then reads as an approximate percentage change, which is often what the business wanted anyway. **When a method assumes roughly symmetric errors.** Fitting on raw views lets three viral articles dominate the fit entirely. Three things logs do not fix. Zeros. `log(0)` is undefined. Adding a small constant is common and it is a genuine choice, not a neutral one, since the result depends heavily on the constant you pick. For count data with real zeros, a model built for counts is usually more honest than a shifted log. Negative values. Logs are simply unavailable, so profit or temperature change need a different approach. Interpretation. Exponentiating a prediction from a log-scale model gives approximately the geometric mean, not the arithmetic mean, and those differ substantially on skewed data. If someone needs a revenue forecast to add up, that gap matters. If the goal is description rather than modelling, you often do not need a transform at all. Report the median, the quartiles and a few high percentiles, and show the histogram on a log-scaled axis so the shape is visible without altering the numbers. **Likely follow-ups** - You take logs, fit a model, and exponentiate the prediction — what have you actually predicted? - Some values are zero — what now? - If the goal is just describing the data, do you need to transform at all? --- ## 25. Where do the uniform and exponential distributions actually turn up? *Medium · Occasional* **Short answer.** Uniform means every value in a range is equally likely, which fits genuine randomisation and rounding error more than natural measurements. Exponential describes waiting time until the next event when events arrive at a constant average rate, and it is memoryless: how long you have already waited tells you nothing. **Uniform** shows up less often in nature than people expect. Where it genuinely applies: random number generation, allocation of applications to reviewers by a randomising system, and rounding error, where a value rounded to the nearest rupee has an error uniformly spread between −0.5 and +0.5. For a continuous uniform between a and b, the mean is (a+b)/2 and every subinterval of equal width has equal probability. That last property is what makes it a poor model for most measured quantities, which cluster around a typical value rather than spreading flat. **Exponential** describes the gap between events when events arrive at a constant average rate. A power distribution feeder experiences faults at an average of 2 per month. The time until the next fault is exponential with mean 1/2 = 0.5 months. The probability of going more than a month without a fault is: ``` P(T > 1) = e^(-2 × 1) = e^(-2) = 0.135 ``` About a 13.5% chance of a clear month. The memoryless property is the part interviewers probe. If the feeder has already run 20 days without a fault, the distribution of remaining time is exactly the same as it was on day one. No fault is “due”. That is a genuine property of a constant-rate process and it is exactly why exponential is the wrong model for equipment wear, where failure probability rises with age. Weibull handles that case. The link worth stating: if events per interval are Poisson with rate lambda, the gaps between events are exponential with mean 1/lambda. They are two views of the same process. For a bus wait, neither is automatically right. If buses run to a schedule you do not know, arriving at a random time gives a roughly uniform wait within the headway. If they arrive erratically at a constant average rate, exponential fits better. Ask which situation you are in. **Likely follow-ups** - What does memorylessness mean for someone already waiting fifteen minutes? - If arrivals are Poisson, what distribution do the gaps between them follow? - Would you model a bus wait as uniform or exponential, and why? --- More Business Analyst sets: https://codeayan.com/get-hired/business-analyst All interview prep: https://codeayan.com/get-hired