Data problems in Iowa caucus results

It wasn’t just issues with an app. There appears to be many more problems with the Iowa caucus results. The Upshot broke it down with a closer look at the data:

Some of these inconsistencies may prove to be innocuous, and they do not indicate an intentional effort to compromise or rig the result. There is no apparent bias in favor of the leaders Pete Buttigieg or Bernie Sanders, meaning the overall effect on the winner’s margin may be small.

But not all of the errors are minor, and they raise questions about whether the public will ever get a completely precise account of the Iowa results. With Mr. Sanders closing to within 0.1 percentage points with 97 percent of 1,765 precincts reporting, the race could easily grow close enough for even the most minor errors to delay a final projection or raise doubts about a declared winner.

When did voting get so complicated?

Tags: , , ,

Iowa liquor sales data, 3m rows

Iowa released liquor sales data for weekly purchases at the store level.

This dataset contains the spirits purchase information of Iowa Class "E" liquor licensees by product and date of purchase from January 1, 2014 to current. The dataset can be used to analyze total spirits sales in Iowa of individual products at the store level.

There are over three million rows that contain a store name, address, liquor category, liquor vendor, and cost. I imagine this could be a fun spatial time series dataset to play with. Look for seasonal trends, when stores expect to sell more rum or vodka, brand bestsellers, or regional favorites. Even though it's just for Iowa, there's probably a close relationship to national sales.

See some preliminary documentation by Dan Nguyen on how to get started.

Tags: ,