Justin Bélair on LinkedIn: QUANTFISH: On-demand statistics courses for researchers (2024)

What do we mean when we say "marginal", "conditional" or "joint" distributions?A reminder : simplifying a bit, a random variable's distribution is both the values it can take and the probability that it takes on these values.1️⃣ In the case of the throw of a die, the distribution is 1/6 probability of landing on 1, 1/6 probability of landing on 2, etc. It's a simple example where all probabilities are equal, but it could be a lot more complicated. This would be the "marginal distribution of X".2️⃣ Now, suppose we throw two dice and note the result. X = "result of the first die throw" and Y = "result of the second die throw", assuming these variables are independent, i.e. the result of a die throw has 0 influence on the other one.Here, having two variables we can speak of the "joint distribution of X and Y". To describe the distribution, we would say: X = 1, Y = 1 with probability 1/36, X = 1 and Y = 2, with probability 1/36, etc. By considering both variables simultaneously, we create a new probability distribution that we call the "joint distribution of X and Y".3️⃣ We could also speak of the "conditional distribution of Y when X = 1", for example. Here, both dice are independent, so the fact that X = 1 does not change the distribution of Y. But it is easy to think of examples where this matters. Say, X is the age of a population and Y is the mortality, we could speak of the distribution of mortality conditional on X being greater than 80 years old. This distribution would be different than the marginal distribution of Y, mortality in general, when we don't refer to any specific values of age.💡 Advanced tip : When we build a regression model of Y as a function of X1, X2, X3, for example, we are modelling the expected value, i.e. the mean, of Y conditional on X1, X2, X3. Basically, we are assuming there is a relationship between Y and the distribution of the regression variables, and we are looking at the conditional distributions for different values of the Xs.

96

17 Comments

How to become skilled in a technical field like statistics?1️⃣ Learn how to manipulate data. It all starts there. Pick one software tool and go! I'm biased towards #R but many great careers are built on different skill sets!2️⃣ Work on a project - academic, professional, weekend side-project, whatever! Just make sure you're actually interested in the data and the results that could emerge - if you're truly curious about the project, you'll learn so much.What are some of your strategies to become technically proficient ?#biostatistics #statistics #data #science

34

13 Comments

This textbook has served as a reference for a long-time. It takes what the authors call a "theoretical statistics" approach, which means that they build the ideas behind statistical inference from probability theory.A math background, especially calculus and a bit of matrix algebra is useful.The book is self-contained, but it could help to have a solid background in probability theory!#biostatistics #statistics #data #science #book #reading

550

87 Comments

Thanks for the incredible support - I'm approaching 10k followers and would love to give back 😃 During my last live event, I loved answering questions!I get a lot of them and I cannot always answer in depth, so I'm dedicating a full hour to answering all your questions, live!👉 Click "Attend" to get a reminder and drop by on May 30th 12PM EDT!See you all 😎#statistics #biostatistics #data #science #live #AMA

12

I found this gem visiting my dad in Washington DCA beautiful book useful for anyone that uses data visualization - this covers precisely everybody, since we all read charts, graphics, and diverse data displays in our daily lives!The whole book is structured around discussing examples and distilling from them the common principles that underlie elegant and useful data visualizations.Do you have any other good resource recommendations for upgrading my #dataviz skills?#biostatistics #statistics #data #sciencePS. I plan on writing about stuff from this book in the next few weeks, stay tuned!PPS. Yes, we lived on a boat for the week :)

36

9 Comments

#LinkedIn tip!Use Boolean (i.e. logical) expressions to enhance searches!I use this all the time to find interesting people to add to my network.Do you have any neat Linkedin search tips?

30

4 Comments

What kind of work does a biostatistics consultant do?Robert Rachford and I discuss technical aspects of a real-life research project I've worked on.#biostatistics #statistics #data #science

28

2 Comments

Biostatistics related fields ➡ Bioinformatics ➡ Biotechnology ➡ Clinical Research ➡ Public Health ➡ Epidemiology ➡ Pharmacoepidemiology ➡ Pharmacoeconomics ➡ and more...Did I miss any? 🤔 If you work in one of those fields, please connect with me, I’d love to have you in my network.Please consider sharing this post for maximum outreach, thanks 😊#biostatistics #statistics #data #science

68

26 Comments

The classic textbook on Generalized Linear Models (GLM), a very widespread tool in biostatistics.There are of course more recent textbooks which will be more hands-on and up-to-date, but this book is a masterpiece.The style is concise and instructive, with many examples sprinkled throughout!I loved the first few chapters. They give historical context on how two important researchers in GLM viewed their technique in through the statistical lens of that period.#statistics #biostatistics #data #science #books #reading

412

34 Comments