- 101
- 19 471
Didier Ruedin
Switzerland
Registrace 21. 07. 2019
Quantitative analysis in the social sciences -- by intuition. I have been teaching quantitative methods in the social sciences at various levels since 2014. Here I focus on basic concepts of quantitative analysis and how to use them in the open-source package R. This is part of my effort to help others excel in the social sciences. Substantially, I research various aspects of human migration.
Banner image by Andreas Perret.
Banner image by Andreas Perret.
Why use R in the social sciences?
In this video, we explore why R is a good choice for quantitative analysis in the social sciences. It's free and has no limits built in. As you learn new methods, you can continue your journey with R (no ceiling). Packages can add new functionality, though they're not always easy.
00:00 introduction
00:33 no ceiling
00:48 packages
www.r-project.org/
posit.co/products/open-source/rstudio/
00:00 introduction
00:33 no ceiling
00:48 packages
www.r-project.org/
posit.co/products/open-source/rstudio/
zhlédnutí: 4
Video
How to avoid having too many digits in your results
zhlédnutí 5Před 19 hodinami
In this video, we learn how to avoid using too many digits when presenting results in the social sciences. You should look at what’s practically relevant: the kind of difference you would care about. We can have too many digits, and this distracts us from seeing differences where they exist. 00:00 Introduction 00:28 Example
Introduction : Les nouveaux clivages politiques
zhlédnutí 12Před 14 dny
Introduction à "Les nouveaux clivages politiques : migration, environnement, cosmopolitisme" (description du cours; www.unine.ch/) Image de fond de Pexels.
Introduction : analyse quantitative en sciences sociales
zhlédnutí 33Před 14 dny
Introduction à "analyse quantitative en sciences sociales" (description du cours; www.unine.ch/) Image de fond de Pexels.
How to test if two variables are associated
zhlédnutí 14Před 14 dny
In this video, we explore how we check if two variables are associated. What we actually do depends on the kind of variables we have. For two categorical variables, we create a cross table. Often we standardize values using percentages to better compare. Bar charts make things more accessible. For a categorical and a continuous variable, we calculate the mean or median per group. Boxplots or hi...
Quit R in the terminal
zhlédnutí 2,1KPřed 21 dnem
Just a very brief reminder to use q() to quit R in the terminal.
Yes, a linear regression model can fit a curve
zhlédnutí 203Před 28 dny
In this video, we clear up the belief that a linear regression model necessarily reflects a linear relationship between the outcome and the predictor variable. A linear relationship means that we can summarize an association with a line. If we add a non-linear term (e.g. the age squared) to a regression model, mathematically it’s still a linear regression model, but the relationship can now be ...
How to understand quantiles to summarize continuous variables
zhlédnutí 27Před měsícem
In this video, we look at quantiles as a way to summarize a continuous variable like age. Compared to the mean or median, quantiles tell us more about the distribution of values, but we may want to look at the histogram or kernel density plots anyway. 00:00 Introduction 01:12 Example
How datasets are typically structured in the social sciences
zhlédnutí 16Před měsícem
We look at how datasets are typically structured in the social sciences: a table, where each row is a person, and each column is a variable that characterizes each person. Data are abstractions where only some characteristics of each person are kept because this gives us access to quantitative models. 00:00 Introduction 00:35 Abstraction Additional images from Pexels (free licence)
Summarizing variables: finding the typical value
zhlédnutí 19Před měsícem
In this video, we examine summarize variables by identifying the typical value. Depending on the type of variable, we use different ways of summarizing what's typical: the mode for categorical variables, the mean or median for continuous variables. 00:00 Introduction 00:28 Categorical 01:07 Continuous Additional images from Pexels (free licence)
4 ways to create non-linear terms in Rstudio
zhlédnutí 45Před měsícem
In this video, we look at four ways to create non-linear terms in Rstudio. We can create a new variable with the quadratic term, either by multiplying the variable with itself (x*x), using the power notation (x^2), or on the fly using I(). We'll add the quadratic term as an extra term in the regression model. An alternative is recoding the continuous variable into a categorical one, and replaci...
Import from Excel into Rstudio
zhlédnutí 60Před 2 měsíci
This video shows how easy it is to get data from Excel into Rstudio. We can export from Excel (or any other spreadsheet) to CVS, and import this in Rstudio. We can also import Excel documents directly from Rstudio (or R more generally), using the readxl package. Often it's as easy as selecting the spreadsheet and clicking OK, but the package offers more flexibility, such as picking a specific s...
What is the researcher degree of freedom?
zhlédnutí 29Před 2 měsíci
In this video, we look at what that expression "researcher degree of freedom" means. The term highlights how in quantitative research, we (have to) make a lot of choices when doing the analysis. These choices can affect the results we get. I use an example to outline some of the choices in a simple regression model, and emphasize that we should try to empirically the effect of our choices on th...
How to read kernel densities
zhlédnutí 35Před 2 měsíci
Kernel density plots like histograms are ways of summarizing a continuous variable. The height of the bars or the height of the curve tells us how likely certain values or scenarios are.The bandwidth is a parameter to change how smooth the curve is, but it's still a description of the same data. 0:00 Introduction 0:22 Example
No numbers/coefficients (NA) for some of my variables in a regression model?
zhlédnutí 16Před 2 měsíci
Why are there sometimes no numbers or coefficients for some of the variables in a regression model? Why does R show NA instead of a coefficient? Normally, this happens when there are categorical variables that are perfectly correlated. In the example presented, the variables "type" and "character" measure exactly the same thing, so R will automatically remove one. 0:00 Introduction 0:26 Example
How to interpret coefficients of binary variables
zhlédnutí 56Před 3 měsíci
How to interpret coefficients of binary variables
Replace a missing value category with NA in R
zhlédnutí 34Před 3 měsíci
Replace a missing value category with NA in R
Multiple regression vs. many correlations?
zhlédnutí 26Před 3 měsíci
Multiple regression vs. many correlations?
Understanding the anatomy of functions in R
zhlédnutí 31Před 3 měsíci
Understanding the anatomy of functions in R
Equal sign vs. backward arrow in R: which one to use?
zhlédnutí 54Před 4 měsíci
Equal sign vs. backward arrow in R: which one to use?
Set the reference category in a regression model in R
zhlédnutí 105Před 4 měsíci
Set the reference category in a regression model in R
How to get graphs out of R using Rstudio and code
zhlédnutí 62Před 4 měsíci
How to get graphs out of R using Rstudio and code
How to get data into R and learn about the working directory
zhlédnutí 155Před 4 měsíci
How to get data into R and learn about the working directory
Why is there only a single coefficient for a categorical variable?
zhlédnutí 40Před 5 měsíci
Why is there only a single coefficient for a categorical variable?
Barplot vs. histogram: what's the difference?
zhlédnutí 49Před 5 měsíci
Barplot vs. histogram: what's the difference?
Let’s stop talking about the second generation and migration background
zhlédnutí 78Před 5 měsíci
Let’s stop talking about the second generation and migration background
Introducing the migration-mobility nexus
zhlédnutí 30Před 5 měsíci
Introducing the migration-mobility nexus
How to structure an academic paper (in the social sciences)
zhlédnutí 49Před 6 měsíci
How to structure an academic paper (in the social sciences)
Ok
Thank you!! 🥲🥲 I'm writing a paper and it's driving me crazy
Thanks for checking in! I'm glad if it helped a little bit. The proliferation of terminology in quantitative analysis still baffles me sometimes.
Thank you, Didier. I was not clear, but now I think I am clear about Table 2 fallacy
Thanks for checking in! I'm glad if it helped a little bit.
Do you think Zotero's "Insert Note" option in Word can be useful for inserting the text of a quote?
Thanks for checking in! The note feature definitely has potential for inserting quotes. What seems to work best is quotes from text you have highlighted in the PDF: (1) Highlight text you want to quote in the PDF, and (2) choose "Add Note from Annotations". If you then (3) add this note in Word, all the text you highlighted appears, but the cool thing is that the reference _and_ page number are also added! Just delete the other quotes/parts of the notes. Please let us know if you have a better workflow.
thank you for the explanation :)
You're welcome.
If race is a social construct, then we need to abolish the entire science of zoology.
Last time I checked, zoologists used objective criteria to identify different species, so there's no need. The idea of *human* races has no biological foundation (askabiologist.asu.edu/questions/human-races).
@@druedin LMAO. Dude, that is completely absurd. Okay, so, Neanderthals didn't exist? There's no difference between a Homo erectus and an Australopithecus afarensis? C'mon! Listen to yourself. You're a clown.
There are biological subspecies. Europeans have up to 6% Neanderthal DNA. Asians have Neander-Denisovan admixture. Melanesians have Denisovan + a touch of Homo erectus. Africans have up to 19% ghost archaic, very like Australopithecine or Homo naledi.
Thanks for checking in. I'm not sure where you want to go with your comment. In the video I'm referring to the notion of *human* "races", apologies if that wasn't clear enough. There is no biological basis for distinguishing different human "races" (see for example here: askabiologist.asu.edu/questions/human-races), which is why we refer to race as a social construct. You might want to look at what zoologists have to say about this, namely: "The concept of race is the result of racism, not its prerequisite." (Jena Declaration, www.uni-jena.de/en/190910-jenaererklaerung-en). As a social scientist, I note that distinctions such as those you mention do not map well onto "racial" differences as they are used and experienced today, so they must be socially constructed.
Your explanations are so concise and easy to follow😀. Thank you for making this video!
Thanks for your kind words!
I am a heavy pipe user, but as you say, I often mix the styles :)
Thanks for checking in. I think R users really need to understand both approaches (onions, pipes) if they want to benefit from the many tutorials and examples out there.
We have cultural behavior differences in the USA.
Would that be self-stereotypes and fall under the assumptions catagory?
Difficulties arise because of different definitions. Good that you provided a link to the studies, however a few concrete examples would have be more clear.
Yes, the different and overlapping definitions definitely don't help. If you don't mind examples from the US, here's a good summary video: czcams.com/video/lQ_8eOaiz8o/video.html
@@druedin thanks!
Did you mean systemic when you said systematic?
Thanks for checking in. I was referring to "systematic" as in a pattern that we can find in a consistent way. The consequences of these differences are then called "structural" or "systemic" -- some make a clear distinction between the two, others treat them as synonyms.
Great video. Thank you for this easy and quick explanation. Cheers.
Thanks. I'm glad it was useful!