Some ideas on good report

What is a good data analysis presentation

Three types of reserach questions

Patterns of association – focus on pattern discovery
Causal question – identify the effect of an intervention
- Experiment
- Observational data
Build a predictive model – focus on selection, understand how model works

Let us focus on association

Think questions like

Are people working more hours, make more money/hour?
How much less does a car worth when having 1000km more in odometer?
What is the relationship between hotel prices and location / amenities?

Causal would be

In an intervention, we make some people work more, will they have a high w/h?

Prediction would be

Here is wage data, let us build a model for hourly wage

What to include in a report

Consider a few-page report on an association with some possibility for causal analysis (later)

Key parts of the report

Introduction (why interesting)
Research question
Data description
- source
- variables
Methods
- what method you use (e.g. cross section OLS)
- regression you estimate precisely
Results
- Include exhibits (think about graphs, tables)
- Decide on main result and focus on explaining it (e.g. \(x\) coefficient)
  - Hint: think for the reader – what would help them understand?
Conclusion
- Short: 1-2 para
- Discussion to summarize result (in plain English)
- Place result in context
  - is it causal
  - external validity
- (no need for future research)

---
title: "Some ideas on good report"
---

What is a good data analysis presentation

## Three types of reserach questions

* Patterns of association -- focus on pattern discovery
* Causal question -- identify the effect of an intervention
    * Experiment
    * Observational data
* Build a predictive model -- focus on selection, understand how model works

## Let us focus on association 

Think questions like

* Are people working more hours, make more money/hour?
* How much less does a car worth when having 1000km more in odometer?
* What is the relationship between hotel prices and location / amenities? 

Causal would be

* In an intervention, we make some people work more, will they have a high w/h?

Prediction would be

* Here is wage data, let us build a model for hourly wage

## What to include in a report

Consider a few-page report on an association with some possibility for causal analysis (later)

**Key parts of the report**

* Introduction (why interesting)
* Research question 
* Data description
    * source
    * variables
 * Methods
    * what method you use (e.g. cross section OLS)
    * regression you estimate precisely
 * Results
    * Include exhibits (think about graphs, tables)
    * Decide on main result and focus on explaining it (e.g. $x$ coefficient)
      * Hint: think for the reader -- what would help them understand?
  
* Conclusion
  * Short: 1-2 para
  * Discussion to summarize result (in plain English)
  * Place result in context
    * is it causal
    * external validity
  * (no need for future research)