diff --git a/01_designing-research.Rmd b/01_designing-research.Rmd index fc04421..e080328 100644 --- a/01_designing-research.Rmd +++ b/01_designing-research.Rmd @@ -7,55 +7,53 @@ - Explain why research needs a design - Explain importance of a good research question +## Main Goal {-} -## Research Questions {-} +**How can you do research in such a way that, when you’re done, you have an answer to your research question?** + +## I Have a Question -- Section 1.1, p.3-5 - A question that you plan to answer - Contributes to improved understanding "how the world works" - 1. Well-Defined 2. Answerable 3. Understandable -## Empirical Research {-} -- Section 1.2 p.4-5 +## Empirical Research + - Any research that uses structured observations from the real world to attempt to answer questions. -- Quantitative empirical research uses quantitative measurements (numbers, usually). +- Quantitative **empirical research** uses quantitative measurements (numbers, usually). - Observations rather than reason -- More data sets, fewer interviews. +- *More data sets, fewer interviews* - Measurements hard to to collect precisely or interpret accurately - Numbers we have don’t actually answer the research question we have, what can we do? -- Goal is to collect the right data to answer the question and design reasearch to do this +- Goal is to collect the right data to answer the question and design research to do this -## Why Research needs a design {-} +## Why Research Needs a Design -- Section 1.3 p.5-7 , -- Capable of answering the question it's trying to answer -- Needed to evaluate first pass analysis and what to do instead +- **Capable of answering** the question it's trying to answer +- *Needed to evaluate first pass analysis and what to do instead* - Nutrition Research as example of lack of design 1. Not having research capable to answer question 2. Not actually answering the question - Leads to inconsistent results which people cannot rely on nor use -- Could lead to too difficult a question to answer, but its still a takeway +- Could lead to too difficult a question to answer, but its still a takeaway -## Book Goals {-} +## In This Book To Improve Confidence in crafting a research project, what kind of data needed, and calculations to perform on data --First Half: -1. Ways to build an answerable research question -2. Think of quantitative empirical research to perform to answer question +- First Half: + 1. Ways to build an **answerable research question** + 2. Think of quantitative *empirical research* to perform to answer question --Second Half: -1. "Toolbox" using observational data -2. Tools are widely applicable to answer many questions +- Second Half: + 1. "Toolbox" using *observational data* + 2. **Tools** are widely applicable to answer many questions -## Discussion/Practicals {-} -Questions, Discussions, or Examples to fill in during Book Club Meeting ## Meeting Videos {-} diff --git a/02_research-questions.Rmd b/02_research-questions.Rmd index c354f83..f00e5e6 100644 --- a/02_research-questions.Rmd +++ b/02_research-questions.Rmd @@ -6,73 +6,63 @@ - Compare Data Mining and Research Questions - Determine if a question is good -## What is a Research Question? {-} +## What is a Research Question? -- Section 2.1 p. 9-12, Section 2.3 p. 15-16 -- When answered, will improve understanding of how the world works. -- There exists data when found results in believable answer -- Can be Answerable with evidence: Avoid ambiguous questions such as "best" -- It should inform theory: answer something broader than itself +- **Research Question** + - *can be answered* + - *improve understanding* of how the world works -- Theory - - Tells us why - - May be True or False - but explains why we might see outcomes - -- Good Question takes us from Theory to Hypothesis - - When answered improves ability to explain why - - If this is how the world works, what should I expect to observe? - - Gives us something new about why - -- Theory -> Research Question or Research Question -> Theory - -- Right data for right questions - -- Two Checks for Condition for Research Question: - 1. Could we answer the question? - 2. Does the question tell us about how the world works? +- from **Theory** to **Hypothesis** + +
+Theory +An explanation of **why** the things we observe happen, or otherwise *generalizes* what we observe in one situation to another situation. +
+ +
+Hypothesis +A *specific* statement about what you will observe in the data. +
-- Checks if Research Question Informs Theory: - 1. Would unexpected result change your understanding of the world? - 2. If unexpected result doesn't change understanding, then bad question - 3. If answered, hard to explain away if inconvenient - -## Data Mining vs. Research Q's {-} -- Section 2.2 p.13 - 16 -- Data Mining is good at finding patterns and making predictions under stability -- Not good at improving understanding nor improve theory main reason are: - 1. Answers what's in the data , not explaining why. Correlation != Causation - 2. Does not deal with abstraction, can see observations but not at developing theory - 3. Results in false positives - observations found in sample but not outside of it. Random relationships eventually occur when testing everything +- A good test for whether a research question informs theory is to imagine that you find an *unexpected result*, and then *wonder* whether it would make you change your understanding of the world + +## Why Start with a Question? + +- **data mining**: starting with the data and seeking *patterns* + - less useful for improving theory + - could create false positives + - doesn't generalize to abstraction + +- precautions + - try replication + - research design + +## Where Do Research Questions Come From? + +- theory --> hypothesis --> question +- data --> opportunity --> question + +## How Do You Know if You’ve Got a Good One? + +1. **Potential Results**: If you cant say something interesting from results, Question and Theory not closely linked +2. **Feasibility**: Possible vs. realistically obtainable data +3. **Scale**: Consider time and resource constraints +4. **Design**: Finding a reasonable research design that can answer it +5. **Simplicity** + - *Don't combine* multiple determinants into the question to answer + - [Multiple models](https://christophmolnar.com/books/modeling-mindsets) could lead to the same result -- Can lead to Research Questions - - Come to data without a theory, noticed interesting data patterns - - Confirm it holds up in other data aka replication of data patterns - - -## Considerations for a good Research Q {-} -- Section 2.3 p. 15-16, Section 2.4 p. 16-18 -- Sources of Questions: - - Curiosity - - Theory - - If this is what I expect the world to work, what would I expect to see in the world? - - Opportunity - - What questions would this data allow me to answer? - -- Research Questions tells us why hypothesis to test - -1. Potential Results - - If you cant say something interesting from results, Question and Theory not closely linked -2. Feasibility - - Possible vs. Realistically Obtainable Data -3. Scale - - Consider time, resource constraints -4. Design - - Finding a reasonable research design that can answer it -5. Simplicity - - Don't combine multiple determinants into the question to answer - -## Discussion/Practicals {-} -Questions, Discussions, or Examples to fill in during Book Club Meeting +## Actionable {-} + +"To transform vague stakeholder requests into actionable causal projects, work through these four questions together: + +1. What is the *endgame* of the project? +2. What is the **core question** or hypothesis we are addressing? +3. What *decisions* or actions will be taken based on our insights? +4. How will we **measure** project success on our end?" +---Robson Tigre, [Everyday Causal Inference](https://www.everydaycausal.com/planning-your-analysis.html#four-essential-clarifying-questions) + + ## Meeting Videos {-} diff --git a/index.Rmd b/index.Rmd index a70f0cd..9420db6 100644 --- a/index.Rmd +++ b/index.Rmd @@ -13,7 +13,7 @@ description: "This website is the product of the Data Science Learning Community # Introduction {-} -This is a companion for the book[_The Effect: An Introduction to Research Design and Causality_](https://theeffectbook.net/index.html) by Nick Huntington-Klein (copyright January 5, 2022). +This is a companion for the book[_The Effect: An Introduction to Research Design and Causality (Second Edition)_](https://theeffectbook.net/index.html) by Nick Huntington-Klein (published July 11, 2025). - These are **shared slides!** - Developed by the [Data Science Learning Community](https://dslc.io). @@ -39,29 +39,50 @@ This is a companion for the book[_The Effect: An Introduction to Research Design - We'll discuss even if presenter unavailable - Meetings = **1 hour.** - **Goal:** 1 chapter/week + + + -## Introductions {-} +## Icebreaker {-} -If you feel comfortable sharing, unmute or raise your hand! +If you feel comfortable sharing, in the Zoom chat: -- **Who** are you? -- **Where** are you joining from? -- **Previous clubs?** (DSLC or other) -- **How long** have you been using R? -- **What** are you most looking forward to learning? +- *Who* are you? +- *Where* are you joining from? +- *Previous clubs?* (DSLC or other) +- *How long* have you been using R and/or Python? +- *What* are you most looking forward to learning? ## git and GitHub {-} -- **Git/GitHub** (version control) = **valuable skill.** - - Discussed more in **Chapter 21.** -- If you present, best to **integrate your materials** into this bookdown version. +- **Git/GitHub** (version control) = *valuable skill.* +- If you present, best to *integrate your materials* into this bookdown version. - You'll need a **GitHub account** to submit modifications. -- **Full instructions** in [GitHub repository](https://github.com/r4ds/bookclub-rpkgs) ("repo"). +- Full instructions in [GitHub repository](https://github.com/r4ds/bookclub-rpkgs) ("repo"). + +## Book Design {-} + +- Part 1: The Design of Research + - emphasis on **research design** +- Part 2: Toolbox + - **causal diagram** framework +- Audiences: data scientists, business analysts, researchers, non-researchers + +## Levels of Understanding {-} + +1. Attend book club meetings +2. Watch the author's [video series](https://nickchk.com/theeffectvideos.html) +3. Present a few meetings +4. Work through the [reading homework](https://github.com/NickCH-K/TheEffectAssignments) +5. Work through the [coding homework](https://github.com/NickCH-K/TheEffectAssignments) + +