Story 1 

Ruhi: “I met Chinar on a dating platform. We started dating each other. We shared the same emotions for the animals which developed a special bond between us. After almost 3 months, we decided to marry each other. It is going amazingly well thereafter.” 

Story 2 

Rishav: “I met Anamika at one of my cousin’s marriages. We exchanged our numbers and even after we returned to our respective homes, we used to talk over chat till late night. We met a few times and started loving each other. It went on for almost 3 years, till one day, I discovered her to chat with many more guys and talking with the same intimacy as she does with me… I was shocked. I, though, had such hunches before also, but without clear proof, which is why I was taking long before I propose to her.” 

Story 3

Misha: “I met Manav at a conference. We discussed multiple staffs. He showed interest in my research topic. We started meeting frequently after that. One day he proposed to me. I was not sure if I should say Yes, though I wanted to. One day, we met at a restaurant and by mistake the waiter dropped some food on my dress while serving. He behaved extremely rude to the waiter in front of everyone. I am scared to marry such a man. My family, though, telling me to ignore one such situation, and give him more chances.” 

Qs: How long is good enough to know a person and decide for marriage?  

 Well, I am not giving any matrimonial gyan here 😊. But the question is, can statistics help in answering one such life’s critical question? 

At a very high level, our life is the biggest example of what we study in Statistics classes. Have you learnt about Sampling Theory and Sample size? See that here also, when you are dating someone, what you are collecting are a few samples from that person’s life. You cannot date all the suitable bachelors of the town before you decide, nor you can date a person life long before you decide. So, we try to collect as many samples as possible through the dating process and decide ‘is this the right choice for me’.   

Now statistics class teaches us about what should be the sample size which is good enough to decide for the population! I know many would answer ‘30’ or ‘100’ etc. But what determines the sample size? It is the sample variance.  

What is Sample variance? – Every time you pick up a sample and calculate the sample mean, how much does it vary from that of the previous samples. Now in this case, every time you meet the person, and you learn new things about his/her nature, behavior in different situations, likes-dislikes – how much does it vary from time to time? Ideally, we as humans look for consistency in behavior as a parameter of trust. The higher the sample variance, the lower the trust. The lower the sample variance, higher is the trust. And therefore, you may say, that if in most situations the person has behaved in a similar fashion, chances are high that after marriage the person would probably be the same and suitable, therefore.  

Therefore, you have a hypothesis: The person will not change after marriage, that is, his/her behavior during dating is going to be same/ similar after marriage too.  

We statisticians start with a Null Hypothesis: That the behaviors will be different. (Why we start with the opposite than what we are trying to prove? Let’s save that for another blog) 

Now notice 2 words above – ‘most situations’ and ‘probably’. Can we quantify them? Well, YES! You remember studying p-value and confidence intervals? What do they say?  

They teach us that life does not always go the same. There are situations when the sample may behave differently. However, lower than % of cases such erratic cases happen, easier it is for us to feel confident about the population using a smaller sample. So, it is like, during your dating sessions, less the number of hick-ups, even a shorter dating period is good enough to decide. Therefore, p-values <0.001 means that if the same experiment is repeated 1000 times, less than 1 such situation the experiment may fail to give back the same result. So, in this example, out of 1000 situations, if the person has behaved differently (i.e. deviates from the sample mean) only once, we can reject the Null Hypothesis, that the person will behave differently. (Why we only reject Null hypothesis and cannot accept alternative hypothesis? Let’s park that for another blog) 

Otherwise, Given the current sample, it is difficult to determine. 

So, what do we suggest to Rishav and Misha? 

Clearly Rishav was taking a long time and collecting more samples about Anamika before he is confident enough to reject the Null hypothesis. Unfortunately, he could not. 

While, in the case of Misha, she may ignore, if p<0.001 or else, give more chances and observe carefully before deciding.  

Are the concepts clear now? Soon we are having practice sessions where you will learn to apply these statistical concepts in solving industry use cases. Stay tuned for that! 

Any further doubt/ comments/ suggestions – please write to mitra@setuschool.com