This data as json
| day | description | everyone | facilitators | facilitators_twitter | id | length | notepad | room | time | timeblock | title | transcription |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Thursday | Sometimes journalists tell stories with regression: we create models to show trends, we use a simulation to fill in missing data, we categorize data, and we explain patterns. As our reporting methods shift from finding examples to crunching counts and percentages to running statistical models, we are still beholden to the journalistic principles of accuracy and fairness. Additionally, we are responsible for using statistical techniques appropriately and describing them accurately. When used carelessly, models can frame a comparison incorrectly or overstate the importance of a trend. Let’s talk about what basic assumptions and considerations journalists should make when using statistical methods, and what kinds of red flags statisticians look for in bad model selection or diagnosis. How should a journalist be thinking about these questions as opposed to a social scientist or researcher? What are some basic techniques all journalists should know when running data through a regression model? We’ll also discuss how newsroom constraints and culture affect our ability to do this kind of work well. If possible, please bring a laptop with you that’s set up with python, jupyter, numpy and pandas. We’ll have some jupyter notebooks to play with. We’ll also have R code you can use instead if you like. If that all sounds foreign to you, no worries! You can look on with someone. | Sam Petulla, Hannah Fresques | spetulla, HannahFresques | stats-newsroom | 75 minutes | y | Thomas Swain | 2:30-3:45pm | thursday-pm-1 | Regression in the newsroom: When to use it and thinking about best practices |