Hypothesis, when to state

Question

This is my last question for today, promise :)

Actually Victor Cheng has already answered this in one of his threads on this website, but I cant say I still fully grasp, so if anyone here does, that would be excellent of course.

When stating a hypothesis, I noticed that my case partner on this website usually states hers very early on, right after she has set up her structure. Often with no information to support it. (e.g. my question: "should client X enter the market?", she: "My hypothesis is that client X Should enter the market, I would like to look at these areas to see whether this is the case". or my question: "what is the reason behind declining profitability?", she: "My hypothesis is that the reason lies on the revenue side".

But in sciences you are taught that the Hypothesis must be based on something, info that points in a direction. Isn't it then odd/invalid to throw out a hypothesis like that with no information yet? At the same time, Victor Cheng says (if I remember correctly) that one shouldve stated a hypothesis a few minutes into the case already. Which is rather contradictory then.

In a parallell manner, I am also wondering if stating a hypothesis in an Interviewer-led/Mckinsey style case should be done (one girl I interviewed with here told me I shouldn't do it), and if yes, if there is anything different in terms of approach/process.

Many thank you!

Answer 1

This is a great question, that applicants struggle with quite often. I personally believe that you can't formulate a valuable hypothesis early on; however, some recruiters really look for this at the beginning of the case so you should give them what they want. Even if the hypothesis is just "yeah I think we can do what you are trying to do", say it. At worst, no harm done; at best, you made your recruiter happy.

Hope that helps?

Thx,

G.

ex-BCG Dallas

Answer 2

I would, personally, discount anyone who starts stating a hypothesis right after drawing out a structure. You know very little about the case, you have asked no clarifying questions, you understand just the top surface of the case yet, so you are basically stating a hypothesis that is grounded in guess-work. Here's the way I typically reach a hypothesis.

1. Once you are handed the case, go through it (some interviewers read it out, then hand data, some others just want you to read it yourself -- MBBs typically "give" the case).

2. Ask a few questions about the case. For example, if it's a falling revenue case, then at least try to narrow it down to (a) is it industry wide / economic downturn (b) are we in a seasonal industry. You can deep dive into whether we are the only one suffering, or whether only the smaller players are suffering due to an amazon-like behemoth eating their lunches later in the case.

3. Narrow down your "deep dive" areas to 1-2 places. For a revenue-loss case, you might want to focus on product quality or change in demographics or new entrants. For a profitability loss case, you might narrow it down to rising costs, inefficiencies in capital allocation (e.g., bad forex hedging, or so on), or pricing pressures.

4. Only then state a hypothesis. Something like "My hypothesis is that this is either an issue of pricing or increasing costs. I'd like to further narrow it down and to do so I'd like to look at the financials around revenues and costs."

5. Once you get the data, have reviewed, then fine-tune the hypothesis and you could say something like, "having gone through, I'd like to hypothesis that this is most likely a cost issue; I'd like to understand whether this is only us, or everyone..." etc etc..

Hemant

Answer 3

Thank you, it does help actually. So the point taken is that probably it mustn't be done so formally. Instead say something like: "I'll assume that we can enter the market, but in order to determine for sure, I'd like to look at variable X, Y Z." Or maybe, even more simple say: "in order to determine whether we should enter this market id like to look at variables X Y Z" [guess problem with the latter is that you haven't stated a hypothesis really..] :) I might be overthinking this. Even thinking why we don't have null hypothesis and alternative hypothesis at this point. Jk. #physicsenvy