Blog 60: Random Sampling by Xavier Lo, FIA, FRM, MBA

Back to technical this week – lets talk about Random Sampling [隨機抽樣]. Yes I know that most of you readers are actuaries, or at least have an understanding of basic statistical techniques. Although Random Sampling seems quite simple at first glance, there are quite a number of things you need to consider.

To start with the very basics, if we would like to calculate a statistic [統計數字] about a group of people (say the average height of people in Hong Kong), we could literally ask everyone to tell us how tall they are. Alternatively, a simpler way would be to ask a random small group of people and assume that they are representative [有代表性] of the whole population. This is called Simple Random Sampling [簡單隨機抽樣]. But what if my sample only had infants [嬰兒]? You could then improve this by making your sample somehow match the population’s characteristics, such as if half the population are adults, then you make sure that half your sample as adults. This is called Stratified Random Sampling [分層隨機抽樣]. Then comes the question of: how about gender? Race?

The issue with Stratified Random Sampling is that you can never really break down your sample to have the same characteristics [特徵] as the population [人口], or else you’d just be sampling the population. You just need to be aware of what you’ve missed out. This is easy to do with something concrete like height. But how about when we look at the insurance industry and we had to calculate things like “satisfaction levels” [滿意度] or “customer fairness” [客戶公平對待]. How should we split our sample? How do we make sure that we aren’t just looking at the group of people who are generally very happy?

Sampling is a very good and practical way of simplifying calculations, but as with all things in life, just be very aware of the limitations and how it affects your decision making. Its how you use the outcome of a statistic which is more important than how you calculate the statistic itself!

Qualified fellow actuary (in UK and Hong Kong), Financial Risk Manager, and MBA graduate (listed on the Dean's List) with a passion for insurance, data science, and analytics. Experienced in a broad range of insurance roles (pricing, capital modelling, reserving, ERM), along with a touch of knowledge in banking. Member of the General Insurance Committee (2021), Actuarial Innovation Committee (2019 - 2021) in ASHK.

    3 Comments

  1. October 24, 2021
    Reply

    This is interesting! I forgot when I read it last! thanks for sharing 🙂

  2. October 24, 2021
    Reply

    This is interesting. Thank you for sharing it.
    But one thing I would like to have a discussion on how to select the correct explain for knowing the customer satisfaction level.

    • Xavier Lo, FIA, FRM, MBA
      October 27, 2021
      Reply

      Thanks Shefali for asking! Do you have any ideas on how you’d start to capture this information? Would like to know your thoughts first before I answer!

Leave A Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.