How to test a hypothesis?

hypothesis
New answer on Feb 27, 2020
4 Answers
747 Views
Anonymous A asked on Feb 20, 2020

I would like to know "how to test a hypothesis". I have an overview of the structure (to verify hypothesis X, I need to proove that A, B, C factors are true). However, in the case of "I hypothesize that this profitability drop is due to revenue drop, due to client preference change", how do I test this specific hypothesis?

Could I say something like, I need to drill down the root cause first. So after confirming that this is a revenue problem, I would look into price and quantity, and the relevant factors linked. If the other factors are not true, then it must be due to customer demand change. In another way, I can say that the company should benchmark what others are doing (if others are making changes to fulfill client demand, we are not, then this must be the root cause).

Appreciate any thoughts. Thanks!

Overview of answers

Upvotes
  • Upvotes
  • Date ascending
  • Date descending
Luca
Expert
Content Creator
replied on Feb 27, 2020
BCG |NASA |20+ interviews with 100% success rate| 120+ students coached |GMAT expert 780/800 score

Hello,

Don't lose your time trying to understand how to formally test an hypothesis. It's not mandatory to say "to verify hypothesis X, I need to proove that A, B, C factors are true".
More over, you can never say that "since the other factors are not true, it has to be in the other way". You have always to be fact based and to prove your recommendations.
It's enough to have an hypothesis in your head and to drill down the different parts of your frameworks until you don't find the real root cause.

Best,
Luca

Was this answer helpful?
Ian
Expert
Content Creator
replied on Feb 23, 2020
BCG | 100% personal interview success rate (8/8) and 95% candidate success rate | Personalized interview prep

I think you need a specific example here not general rules. So, here goes.

First, unless you've been given some clear evidence, don't say "I hypothesize this is due to a R drop". What if the interviewer goes "why?"

Now, if our prompt is about a retailer in a middle America mall, ok, that hypothesis could be right (i.e. online threat, economic stagnating region, etc.). But, lacking that, your hypothesis should actually be more about "what do I need to find an answer".

So, you could say something like "to figure out what's happening here, I need to identify 1) If this is an "us" or the general world problem". If it's an "us" problem, what are we doing wrong? If it's a world problem, what is it specifically, and what can we do to fix it?

2) If it's an "us" problem, is this due to falling revenue, increased costs, or both? To figure this out I need to look at our company's financials from the past few years to spot trends.

3) If it's a world problem, we can see this by looking at market and competitor data.

Do you see where I'm going with this? You hypothesis is less on staking your claim on a random statement and more on stating what you need to know to figure it out. By the time you've found the data to say "I hypothesize that the profitability drop is due to revenue and client preference change" you're already at the end of the case giving your recommendation!

Was this answer helpful?
Clara
Expert
Content Creator
replied on Feb 20, 2020
McKinsey | Awarded professor at Master in Management @ IE | MBA at MIT |+180 students coached | Integrated FIT Guide aut

Hello!

Something that will help you is to envision the issue tree as a "way" to find the solution, that is at the end of this path. As the tree gets more and more ramificated, and before jumping into one of the branches, you need to totally discard the other one.

Hence, first, you will find yourself in the 1st crossroads > revenue or cost? Once the cost one is discarded, you continue to the revenue branch of this 1st crossroad, and forget about the other one. This way, you keep advancing, discarding.

Hope it helps!

Cheers,

Clara

Was this answer helpful?
Vlad
Expert
Content Creator
replied on Feb 20, 2020
McKinsey / Accenture Alum / Got all BIG3 offers / Harvard Business School

Hi,

First of all - using the hypothesis is not mandatory. I would say - use the hypothesis if you are really good at solving the cases. If not - use the basic approach.

There are two ways to use the hypothesis:

First - presenting a structure using the hypothesis. For example, if you are having a PE (private equity) case, you should do the following:

1) Make classic structure (market, company, competitors, feasibility of exit)

2) Make subpoints (e.g. in market: size, growth rates, profitability, segmentation, etc)

3) Present your 1st level Hypothesis:

  • - "In order to understand whether we should invest in Company A, I would like to check that the Market is Attractive, the Company is Attractive, the competition is favorable and we have good opportunities for of exit"

4) Present the main 2nd level Hypothesis:

  • "In the market, I would like to make sure that the market is big enough and growing;
  • In the company I would like to find additional opportunities for growth;
  • In competition I would like to check that the market is fragmented enough;
  • Finally, I would like to check if we have potential buyers and can achieve desired exit multiples"

Another way to use hypothesis is using the hypothesis to prioritize your analysis:

1) Make a structure: "Problem in sales may be related to Sales Motivation, Sales Strategy, Sales Coverage, and Sales Process:

2) Prioritize a part of the structure based on your knowledge / common sense / available data: "Taking into account that motivation is the core problem of the sales organization, I would like to prioritize this part of the analysis"

Good luck!

Was this answer helpful?