Thursday, 10 March 2016, 10:15 a.m.
Test scores. Weather. Census. Elections. There are some data stories that we're always going to have to tackle. In this session, listen to three veterans talk about ways they took the "same old, same old" and turned it on its head. You'll come away with fresh ideas and techniques for digging into familiar stories. This session is good for anyone.
- Chase Davis, New York Times
- Ryan McNeill, Reuters
- John Perry, Atlanta Journal-Constitution
John Perry
Andrew Lloyd Webber's occupation is:
A) Playwright B) Painter C) Athlete D) Sculptor
Real answer is composer/producer. Question-maker thought that would be too hard an answer, so playwright it was.
- Do teachers procter tests for own classes
- Are there monitors? How many?
- Do tests sit around at schools before/after?
- Does stats analysis for cheating exist?
Can be more subtle than changing students answer. Can leave answers on wall (in posters). One teacher walked around with red/green M&Ms to leave on desk signalling answers.
Students don't tend to change answers very often, in particular wrong answers to right answers, can be indicator of cheating.
- Who develops questions?
- Are they reviewed by teachers and topic experts?
- Testing the tests
- P-value (percent right and wrong)
- Discrimination (are kids who score higher more likely to get this question right)
- Differential (are different groups of kids more or less likely to get this question right — could be cultural bias)
- Reports of scoring error
Dept. of Education should be performing these tests. You can ask for them.
In one state, 127 kids flunked a graduation-required state math test because of an incorrectly coded right answer. 7 of them dropped out and didn't graduate.
- Assumes test is valid measure of student learning.
- Who set up evaluation system? (Georgia hired someone who'd written a paper against the method he then had to use)
- Correlation between teacher scores and race or poverty
- Misallocation of teachers? (Do under-performing schools get under-performing teachers?)
- Punishing teachers for teaching poor minority kids
Chase Davis
Not going to talk about cool stuff NYT is doing, going to talk about boring stuff: Elections results themselves.
"Every couple years, we're going to have to tell people who won elections."
Basic info people want to get. There's only so much you can do to change that. Don't want to distract from key point: Who's really winning at any given time.
Evolution of stuff we've done around election results: You don't see a lot of change except basic design.
That's a lie
There's been huge change, that most people will never see. Mostly to keep us sane in the office.
Innovation in terms of process and culture. Just as good ways to move the ball forward.
More than 50 nights throughout primary season that need to be staffed. Burnout city.
Built Elex.
Elex is pronounced El-EX.
Opens up the "elections stack" to everyone who needs access; shifts burden from one to many.
Decided to staff some primaries from home. Convinced themselves that tools worked as they should, didn't need to be in the office all the time.
Cultural shift. Why Chase Davis can be at NICAR this week (two debates and a primary).
"Almost anybody can go in and push the buttons to run an election night if they need to."
Innovation that you can do around things like this is just as valuable around the process as around the output.
Incremental improvement is also valuable.
Ryan McNeill
The problem: What's happening now?
NOAA Tides and Currents — long-used for navigation, but there are other uses.
Many stations have far more than water levels.
SF tide gauge has hourly readings since the end of the 19th century.
Q How to sell/pitch processes?
"Just did it". Build confidence early. Stuff that you need to do that's long been known but unsaid/undone.