YouTube2Text

Case Interview Fundamentals: Lesson 29 [Hypothesis Testing: the 3-step approach] — Transcript

by CraftingCases · 3,413 words · 555 segments · language en · Watch on YouTube

Full transcript

  1. 0:00Hey guys, how are you doing today? In
  2. 0:02this video, I want to show you our
  3. 0:04three-step approach to hypothesis
  4. 0:06testing, the last building block in this
  5. 0:10course.
  6. 0:11Uh this is by far the hardest one to
  7. 0:13teach, but it's not necessarily the
  8. 0:15hardest one to do.
  9. 0:17And that's because it's where you can
  10. 0:18get the most
  11. 0:20creative. You can
  12. 0:22There are many, many possible answers
  13. 0:24that work, uh not only one or a few that
  14. 0:27are correct.
  15. 0:28Because you can do anything as long as
  16. 0:32you're actually testing the hypothesis
  17. 0:33in a feasible way.
  18. 0:35Our three-step approach it will help you
  19. 0:37to do that, but if you don't follow it,
  20. 0:39it doesn't mean you're going to get it
  21. 0:41wrong.
  22. 0:42So,
  23. 0:43uh
  24. 0:45because of this,
  25. 0:47the best way to learn this is to
  26. 0:48practice. This three-step approach is
  27. 0:50going to help you, but you're only going
  28. 0:54to learn by practicing a lot. And that's
  29. 0:56why you're have you're going to see
  30. 0:58several drills in this chapter, and
  31. 1:00that's also why I'm going to teach you
  32. 1:02by the end of this video how to come up
  33. 1:04with your own hypothesis testing
  34. 1:06follow-up questions to practice, okay?
  35. 1:11So, let's move on.
  36. 1:13The three-step approach involves
  37. 1:16uh step number one, defining your
  38. 1:18hypothesis, step number two,
  39. 1:21breaking it down, and step number three,
  40. 1:24gathering and processing data,
  41. 1:26uh and showing it the way you'd present
  42. 1:29it.
  43. 1:30Uh you're not necessarily going to do
  44. 1:33all the steps here.
  45. 1:34You
  46. 1:36might not need to define your
  47. 1:38hypothesis, or you might choose not to
  48. 1:40define your hypothesis to the
  49. 1:41interviewer. You may not need to break
  50. 1:44the hypothesis down into simpler
  51. 1:46hypotheses because maybe you're able to
  52. 1:49test that hypothesis straight away
  53. 1:52instead of breaking it down into simpler
  54. 1:53hypotheses.
  55. 1:54Maybe you won't need to show the
  56. 1:57interviewer how you would gather and
  57. 1:58process data
  58. 2:00or at least you just want to show them
  59. 2:03some crucial things of how you do that
  60. 2:05and not others.
  61. 2:07As long as you tell them what you need
  62. 2:10to be closer to proving or disproving
  63. 2:11the hypothesis and how you'd get it, you
  64. 2:13are fine. So, this is different from
  65. 2:15estimations, for example.
  66. 2:17In an estimation, if you don't do your
  67. 2:20structuring or if you don't do the
  68. 2:22reality check, the interviewer is
  69. 2:24actually going to take points away from
  70. 2:26you for that. In the hypothesis testing
  71. 2:29questions, as long as you get what you
  72. 2:33tell them how you're going to get what
  73. 2:34you need and what you need,
  74. 2:37um you're fine. You're do you're going
  75. 2:39to do well in this question. Okay?
  76. 2:42Now, I want to show you
  77. 2:44uh a case question to which you can come
  78. 2:47up with your own answer for.
  79. 2:49And then I want to show you the
  80. 2:52follow-up questions, which is the
  81. 2:53hypothesis testing question. And as I
  82. 2:55said, as I've done in other
  83. 2:58uh
  84. 2:59uh step-by-step videos, I'm not going to
  85. 3:01show you the answer for the framework
  86. 3:03question, only for the follow-up
  87. 3:04questions. But, feel free to come up
  88. 3:07with your with your answer for it um cuz
  89. 3:11it might help you with your hypothesis
  90. 3:13testing. You might You might have
  91. 3:15actually covered the hypothesis question
  92. 3:17test hypothesis testing question in it.
  93. 3:19Um and you're still getting more
  94. 3:22practice for framework questions. And
  95. 3:24this is a cool framework question. So,
  96. 3:27DSC is a large drugstore chain in Brazil
  97. 3:30with 10% market share,
  98. 3:321,100 stores,
  99. 3:35and 9 billion Brazilian reais in
  100. 3:37revenues. The company faces challenges
  101. 3:40with losses for as long as they can keep
  102. 3:42track of their data. So, they
  103. 3:46are losing products that were otherwise
  104. 3:48good to be sold.
  105. 3:50Currently, their losses are at 1% of
  106. 3:52revenues, and that's cost of product
  107. 3:55lost divided by total revenues.
  108. 3:57While their main competitors are at
  109. 4:000.5%, and they've hired you to help them
  110. 4:03reverse those losses.
  111. 4:04So, how can DSC reduce their losses?
  112. 4:08And now to the follow-up question,
  113. 4:10the hypothesis testing question.
  114. 4:12You mentioned it's possible that poor
  115. 4:14handling of products in distribution
  116. 4:16centers
  117. 4:17is a possible cause
  118. 4:19uh
  119. 4:20for the losses. How would you test
  120. 4:23whether that is true or not?
  121. 4:25You should know that we have two
  122. 4:27different distribution centers, and each
  123. 4:29one of those serves roughly half of our
  124. 4:31stores.
  125. 4:32All right, so take your time, come up
  126. 4:34with your answer, but definitely come up
  127. 4:36with your answer again after hearing
  128. 4:38each one of the steps, so that you can
  129. 4:40learn each one of them thoroughly, and
  130. 4:42then apply them in your practice from
  131. 4:44then on.
  132. 4:45Now, I'm going to explain step one of
  133. 4:48the three step process, and then I'm
  134. 4:50going to show you how I would do step
  135. 4:52one, and you should probably do step one
  136. 4:56before watching my answer, and then
  137. 4:58watch it and try to understand what's
  138. 4:59different, and only then move on to step
  139. 5:01two. Okay?
  140. 5:03So,
  141. 5:04the way I
  142. 5:05the what you need to know is
  143. 5:08defining your hypothesis means
  144. 5:11defining what is the cause, the effect,
  145. 5:12and the consequence in this hypothesis.
  146. 5:14And all hypotheses must have a cause, an
  147. 5:17effect, and a consequence. I'll show you
  148. 5:19an example of this in a second. And next
  149. 5:22thing you need to know is this step is
  150. 5:23only here to help you. You might decide
  151. 5:26to
  152. 5:28do it and show it to the interviewer,
  153. 5:29you might decide to
  154. 5:31do it and not show it to the
  155. 5:32interviewer, and you might even not
  156. 5:34decide you might even decide not to do
  157. 5:36it at all, and that does not mean you're
  158. 5:38not going to pass this question, you're
  159. 5:40not going to do well in this question.
  160. 5:42But, this step is meant to help you
  161. 5:44clear your thoughts and come up with
  162. 5:46your test, and it really, really does
  163. 5:48help it. Okay? So,
  164. 5:51when you break up your hypothesis like
  165. 5:52this, you have
  166. 5:56much more clarity of what is there, what
  167. 5:58is it you want to test, what is it you
  168. 6:00can test, and what you can't. Okay? Let
  169. 6:02me show you an example so this is going
  170. 6:03to be more clear.
  171. 6:05Uh so, the question is,
  172. 6:08is poor handling of products in
  173. 6:10distribution centers a cause the cause
  174. 6:12for the losses?
  175. 6:14Mm the
  176. 6:16cause, effect, and consequence here are:
  177. 6:19Cause, distribution center employees are
  178. 6:21poorly trained or reckless, and that
  179. 6:24causes them to
  180. 6:27handle products poorly,
  181. 6:29and the consequence of that is that
  182. 6:31distribution centers cause losses of
  183. 6:34otherwise fine products. And
  184. 6:38even if you decide not to show this
  185. 6:39first part to the interviewer, the
  186. 6:40second part of this step is crucial, and
  187. 6:43I would definitely recommend that you
  188. 6:44show it to the interviewer, which is
  189. 6:46stating the sentence of the hypothesis
  190. 6:48you want to test. And the one here would
  191. 6:50be, the excessive product losses faced
  192. 6:53faced by drugstore chain DSC are a
  193. 6:56result of bad handling in either one or
  194. 6:59both of their distribution centers.
  195. 7:02Okay? So,
  196. 7:04let's move on to step two, breaking the
  197. 7:06hypothesis down into simpler hypotheses.
  198. 7:09And the logic behind this step is
  199. 7:11breaking your hypothesis down into
  200. 7:15those other hypotheses that need to be
  201. 7:17true in order for that one to be true.
  202. 7:19Uh you need to do that because
  203. 7:22you need simple hypotheses to test. You
  204. 7:25can't test overly complicated
  205. 7:26hypotheses.
  206. 7:28Uh
  207. 7:29if your hypothesis is simple enough,
  208. 7:30though, when you can test it already,
  209. 7:32then you don't need to do that. How do
  210. 7:33you know if it's simple enough?
  211. 7:35Again,
  212. 7:36it's only simple enough if you can test
  213. 7:38it straight away. If you can't test it
  214. 7:40straight away, then you probably need to
  215. 7:42break it down. Okay? So,
  216. 7:45the first principle that you need to
  217. 7:48know here is that logical breakdowns are
  218. 7:50better. Your breakdown might not be 100%
  219. 7:53MECE, but the closer to that the better.
  220. 7:56And you might not test all of the
  221. 8:00sub-hypotheses, the simpler hypotheses
  222. 8:01that you come up with, but you need to
  223. 8:03at least show your interviewer that you
  224. 8:05have a plan
  225. 8:07of what you're going to do if you need
  226. 8:08even more certainty. Okay? And this
  227. 8:12brings me to the second principle, quick
  228. 8:14tests versus exhaustive tests. So, if
  229. 8:17you need to run quick tests, you can
  230. 8:19always pick the one or two most
  231. 8:20important ones and controversial ones
  232. 8:24and uh in your breakdown. So, that means
  233. 8:28that you'll pick the ones that are going
  234. 8:30to mean the most to the main hypothesis
  235. 8:33and that you know the least the answer
  236. 8:36to.
  237. 8:38So, if it's something quite simple or
  238. 8:39that you have a good idea of what it's
  239. 8:41going to be, then that might not be the
  240. 8:43test that you want to do straight away.
  241. 8:45Uh and if it's something that's not
  242. 8:47going to affect so much the main
  243. 8:49hypothesis that you're trying to test,
  244. 8:51then you might not want to test it
  245. 8:52straight away, too. You might want to
  246. 8:54pick one that you don't know the answer
  247. 8:56to and that it's really going to change
  248. 9:00the chances that the main hypothesis is
  249. 9:02true or false.
  250. 9:04Okay?
  251. 9:05And
  252. 9:06uh this is not the only way to do
  253. 9:08quicker tests. Uh we're going to see
  254. 9:10this in the third step, but it's one of
  255. 9:13the ways and you're probably going to
  256. 9:14use both of them together. Next, uh
  257. 9:17availability of data should be in your
  258. 9:19mind when you're doing this step. You're
  259. 9:21not yet going to tell the interviewer
  260. 9:23how you're going to come up with the
  261. 9:25with the data, how you're going to
  262. 9:26gather that data, and how you're going
  263. 9:27to process it, but you don't want to put
  264. 9:30yourself in a dead end. So, if
  265. 9:34you obviously can't get data for some of
  266. 9:36the most important tests uh in your
  267. 9:39breakdown, then you probably should
  268. 9:42try to come up with a different way to
  269. 9:44break down this problem because
  270. 9:46otherwise you're going to put yourself
  271. 9:47in a dead end and you're not going to be
  272. 9:48able to complete your hypothesis test.
  273. 9:51And now I'm going to show you my answer
  274. 9:53or at least the second step of my
  275. 9:54answer. So make sure you've done it
  276. 9:57yourself too before you watch mine.
  277. 10:00So the hypothesis I want to test here is
  278. 10:03are excess losses caused by poor
  279. 10:05handling in distribution centers?
  280. 10:07And by excess losses, I mean those
  281. 10:11losses that are above 0.5% of revenues,
  282. 10:15which is my benchmark. So there are two
  283. 10:17main hypothesis that will answer this
  284. 10:19question. First one is do these excess
  285. 10:21losses really come from distribution
  286. 10:23centers? And the second one, are they
  287. 10:25really caused by mishandling? If they
  288. 10:27come from distribution centers and are
  289. 10:29caused by mishandling, then this
  290. 10:30hypothesis has to be true.
  291. 10:32And to find out if they come from
  292. 10:34distribution centers,
  293. 10:35I would try to figure out where in the
  294. 10:38product's path from supplier to consumer
  295. 10:41do these excess losses come from? And in
  296. 10:43a second I'm going to talk about how I
  297. 10:45would find that out.
  298. 10:47And to find out if they're really caused
  299. 10:49by mishandling, first I'd need to know
  300. 10:51what kind of loss is behind the excess
  301. 10:54losses. So are they caused by
  302. 10:57expired products? Are those products
  303. 10:59going missing? Or are those products
  304. 11:01damaged? Number two, where in the
  305. 11:04distribution center process do these
  306. 11:06losses come from? So
  307. 11:08uh from supplier to leaving to the
  308. 11:10store, leaving distribution center to
  309. 11:12the store, there are several steps in
  310. 11:14the distribution center process. I want
  311. 11:16to know where it is and then I want to
  312. 11:18investigate that
  313. 11:20process that step of the process to
  314. 11:22figure out what exactly is behind that.
  315. 11:26Okay? So
  316. 11:28now let's move on to step three. I'm
  317. 11:30going to show you how it works and then
  318. 11:32I'm going to show you
  319. 11:33um the last step of my answer.
  320. 11:36The third step is how you would gather
  321. 11:40and process data. That actually means
  322. 11:43what you would be doing in real life.
  323. 11:45This is what we want to tell the
  324. 11:46interviewer. So, what data do you want
  325. 11:48to get? How would you get it? How would
  326. 11:50you turn it into information? What do
  327. 11:51you expect it to look like?
  328. 11:53Again, of course, you might not say all
  329. 11:55of these things to the interviewer, but
  330. 11:58you at least need to show them that if
  331. 12:00you were in a real project, if this was
  332. 12:03was your assignment, you'd be able to
  333. 12:06complete it well and it'd be clear to
  334. 12:09your manager, to the partner, to the
  335. 12:11client uh whether the hypothesis is true
  336. 12:14or not and you'd be based on data to do
  337. 12:16that and of course you can get that data
  338. 12:19and process it yourself.
  339. 12:21And next
  340. 12:23indicative tests versus conclusive
  341. 12:25tests. As I said before,
  342. 12:27uh choosing only a part of the
  343. 12:30hypothesis to test is not the only way
  344. 12:32to do it quicker. You can do indicative
  345. 12:34tests uh which are tests that don't
  346. 12:38necessarily prove your hypothesis to be
  347. 12:41false, but shows that it's much more
  348. 12:43likely for it to be true or false versus
  349. 12:46doing conclusive tests where the data
  350. 12:47data actually prove the hypothesis or
  351. 12:50very very very close to proving the
  352. 12:51hypothesis. I'm going to show you some
  353. 12:53examples of this in my own answer. Okay?
  354. 12:57So,
  355. 12:58take your time and come up with the
  356. 12:59third step of your answer or with the
  357. 13:01full answer all over again if you want
  358. 13:03to and only then
  359. 13:05uh watch my answer which I'm going to
  360. 13:07show you now.
  361. 13:08Uh first thing, well, here is my second
  362. 13:12step
  363. 13:13and I'm going to I want to show you how
  364. 13:15I would test the first part of my little
  365. 13:20framework here. And this is
  366. 13:23I tried to draw a chart like this. All
  367. 13:25the steps from supplier to consumer. So,
  368. 13:28arrival from the supplier, distribution
  369. 13:30center, transport, and store. And I want
  370. 13:32to compare my losses with my benchmark
  371. 13:35in each one of those steps. What I
  372. 13:37expect it to look like is
  373. 13:40I have a lot more losses in the
  374. 13:41distribution center than my benchmark.
  375. 13:44And the losses are similar in all other
  376. 13:47steps.
  377. 13:48If I can't get data from my benchmark,
  378. 13:51which is likely, I would test one
  379. 13:54distribution center against the other.
  380. 13:56Uh so, I'd get my losses as a percentage
  381. 13:59of revenues
  382. 14:00in each one of the steps. And what I
  383. 14:02would expect it to look like is one
  384. 14:04distribution center having much more
  385. 14:06losses than the other. Or else,
  386. 14:09uh both distribution centers with much
  387. 14:11more losses than all other steps.
  388. 14:14So,
  389. 14:15take a look at this.
  390. 14:17These second tests are not conclusive.
  391. 14:21Uh so,
  392. 14:23benchmarking one distribution center
  393. 14:25against the other does not conclude to
  394. 14:27me, even if one of these Actually, even
  395. 14:30if this last one is the case, it doesn't
  396. 14:33prove to me that distribution center is
  397. 14:36the cause. Maybe all of the losses are
  398. 14:38supposed to come from distribution
  399. 14:40centers. And what's weird in the FC is
  400. 14:44that we have losses in arrival from
  401. 14:45supplier, we have losses in transport,
  402. 14:47we have losses in store, or something
  403. 14:49like that. And we wouldn't know because
  404. 14:51we can't compare it to the benchmark.
  405. 14:53Which might even have more losses than
  406. 14:55us in the distribution centers, but not
  407. 14:57so many so much more losses. But no no
  408. 15:00losses at all in the other steps. Okay?
  409. 15:03But this, at least, indicates that we
  410. 15:06have a problem in distribution center,
  411. 15:08indeed.
  412. 15:10And another lesson here is that anytime
  413. 15:13you have the opportunity to draw charts,
  414. 15:15and that would make communication
  415. 15:16simpler,
  416. 15:17take that chance and do it. That's going
  417. 15:19to help you speak the interviewer
  418. 15:21language, language. Communication is
  419. 15:22going to be much more much simpler and
  420. 15:25much much clearer.
  421. 15:26And also, something that I want you to
  422. 15:29notice here is there's additional
  423. 15:31information in this chart that would
  424. 15:33have been hard, or at least not as quick
  425. 15:36to convey if I were speaking it only.
  426. 15:38And this is what are the steps in the
  427. 15:40distribution center uh
  428. 15:43in the process from the consumer to the
  429. 15:44supplier. So,
  430. 15:46whereas before I only said the steps
  431. 15:49from supplier to consumer, now I'm
  432. 15:51showing what would be the steps that I
  433. 15:54would use in my analysis.
  434. 15:56And next, what I expect the data to look
  435. 15:59like. So,
  436. 16:00I drew I clearly drew here
  437. 16:02the my expectations of
  438. 16:05uh
  439. 16:06what the data should look like. If it
  440. 16:08does not look like this, then my
  441. 16:09conclusion would be
  442. 16:11slightly or a lot different.
  443. 16:13Now, let's move on to the next step
  444. 16:15where we're to the next uh sub
  445. 16:17hypothesis, where you're going to show
  446. 16:20where going to we're going to see all of
  447. 16:21this again.
  448. 16:22So, what kind of loss is behind the
  449. 16:24excess losses? How am I going to find
  450. 16:26that out? I want to break down those
  451. 16:28losses between those three types that I
  452. 16:30said before.
  453. 16:31Uh missing products, expired products,
  454. 16:33and damaged products. And I want to
  455. 16:35compare distribution centers if I can't
  456. 16:37compare it to benchmarks. And I believe
  457. 16:40it's highly unlikely that I can't
  458. 16:41compare it to benchmarks. If I can,
  459. 16:43well, that's great. If I can't, I'm
  460. 16:45going to do this indicative test here
  461. 16:47and have an idea of what's going on. And
  462. 16:50what I expect to see is
  463. 16:53either uh
  464. 16:55a lot more damaged products in
  465. 16:57distribution centers
  466. 16:59or in only one of them.
  467. 17:01Okay? And to find out where in the
  468. 17:04distribution center process these losses
  469. 17:07come from, I would divide
  470. 17:09I would do pretty much the same as I did
  471. 17:12in the
  472. 17:14in the first test.
  473. 17:16Draw out the process, but I would find
  474. 17:20the losses by type of loss in each step
  475. 17:23of the process. So,
  476. 17:25what losses are
  477. 17:27expired products, missing products, and
  478. 17:29damaged products in the reception, in
  479. 17:31sorting, in storage, in picking, in
  480. 17:34boxing, and in exit. And if I see
  481. 17:36something like what I'm seeing in this
  482. 17:38chart, which is
  483. 17:40a lot of damaged product losses, which
  484. 17:42are much more than all other losses in
  485. 17:45one or two steps, or even all of them,
  486. 17:48then
  487. 17:49there's probably something going on, and
  488. 17:51I want to go invest and investigate the
  489. 17:54root cause of that to find out that is
  490. 17:57actually it actually is
  491. 17:59uh
  492. 18:00mishandling. And the way I do that is
  493. 18:03probably go to the distribution center
  494. 18:06or to the distribution centers,
  495. 18:08and
  496. 18:09look at the process, see what's going
  497. 18:11on, try to see the damaged products, try
  498. 18:13to see them being damaged. I would talk
  499. 18:16before that probably to the manager and
  500. 18:19to whoever is in charge of that specific
  501. 18:22process, and try to get try to gather
  502. 18:25from them uh
  503. 18:26what they think is causing those losses,
  504. 18:29what is different from the other
  505. 18:31distribution center in there, and then
  506. 18:33I'd get to the root cause of the
  507. 18:35problem.
  508. 18:36And now,
  509. 18:37to tell you, like I said in the
  510. 18:39beginning,
  511. 18:40an interesting way to come up with your
  512. 18:43own hypothesis testing questions, it's
  513. 18:45fairly simple. So, you've seen in
  514. 18:47frameworks before
  515. 18:49that you
  516. 18:51are always going to come up with several
  517. 18:52hypotheses by the end of your framework.
  518. 18:55Uh and
  519. 18:57the interviewer could pick any one of
  520. 18:59those and ask you how you would test
  521. 19:01them.
  522. 19:02Uh and what you can do is
  523. 19:05draw your framework for any case,
  524. 19:08check look at what your hypotheses are,
  525. 19:11and each one of those can work as a
  526. 19:14follow-up question for hypothesis
  527. 19:15testing. Uh pick any one of those or
  528. 19:18pick all of those and test them and that
  529. 19:20is your way of practicing. As I said,
  530. 19:22practicing is the only real way to uh
  531. 19:26learn hypothesis testing. If you do that
  532. 19:29with partners, not necessarily one
  533. 19:30interviewing the other, but both come up
  534. 19:33with your own answers and then you
  535. 19:34discuss both of them, which one is
  536. 19:38more
  537. 19:39which one will give you more certainty
  538. 19:41of your conclusion, which one won't,
  539. 19:43which one's quicker or less costly, and
  540. 19:45which one is less.
  541. 19:47And
  542. 19:48you're going to have you're going to
  543. 19:50uh
  544. 19:52you're going to be able to practice and
  545. 19:54learn a lot with that. Okay? So, I hope
  546. 19:56this tool helped you. I hope this theory
  547. 19:59tool helped you and I'll see you in the
  548. 20:01next videos. You're going to have
  549. 20:03several
  550. 20:04different hypothesis testing videos to
  551. 20:06help you here and I honestly hope you
  552. 20:09can learn this and apply this and
  553. 20:10improve your case performance and
  554. 20:12hopefully get your offer. See you in the
  555. 20:14next one.

About this transcript

This page contains the full transcript of Case Interview Fundamentals: Lesson 29 [Hypothesis Testing: the 3-step approach] by CraftingCases, generated from the public captions YouTube serves with the video. The transcript has 3,413 words across 555 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to choose a caption track available on the video.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, caption language selection, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.