How do we express state revenue and production rankings?

[shawnbot]

While building out state prototypes (#1355) and our new data "infrastructure" (#1353) I noticed one issue with state rankings: they might be a little complicated to describe if they're derived from state and offshore region data. For instance, in 2013 California was the "top" state producer of Fuel Oil at a whopping $68, but the national total in that year was actually negative $73k, with Western Gulf of Mexico contributing the overwhelming majority.

So there are a couple of questions here:

1. Do we include offshore regions in our rankings? (This is the important one.)
2. How do we express contributions to a negative revenue total?

For 2. I was thinking that we would derive a percentage by taking the absolute values of the regional and national totals, and dividing those. So abs(68) / abs(-73725) * 100 yields .09%. This accounts for both the edge cases of states with negative revenue in a positive national total and vice-versa.

[RyanSibley]

Link to a notes doc after reviewing the California page. I don't actually get to your question about negative numbers until #6:

https://docs.google.com/document/d/1Yju7CYTc90SFg6KaE0vJIgJ-84qWJIZsXRVnellXRW0/edit

[meiqimichelle]

This is still underway as part of the state profile updates.

[shawnbot]

My thoughts on this have evolved. From what I've read around the internets, it sounds like the best practice for negative contributions to a positive total (or vice-versa) are not typically expressed as a numeric percentage, because they don't really make sense.

I can dig into my history a bit to find some references, but I think the thing to do here is to update our state rankings and percentages to be null (undefined) if the numerator and denominator in the percentage fraction for a given state have different signs (±).

[shawnbot]

While investigating #1397, I queried the db for state revenue rankings with a % > 100 and found some other offenders (note: these values are truncated for legibility, so the ones with % = 100 are actually between 100 and 101):

year	state	product	$ revenue	$ total	%	#
2004	ID	Phosphate Raw Ore	3451909	3451909	100	1
2004	WY	Helium	-21423	-21365	100	2
2004	WY	Soda Ash-Granular	-583138	-449974	129	3
2005	CA	Soda Ash-Granular	-726947	-719067	101	2
2006	MN	Copper	28720	18072	158	1
2006	WY	Inlet Scrubber	637064	553912	115	1
2006	WY	Sodium	73976	63918	115	1
2007	NM	Muriate Of Potash-Standard	1452920	1451918	100	1
2008	MN	Hardrock	24299	17618	137	1
2009	CO	Sodium	-142029	64530	220	5
2009	WY	Inlet Scrubber	-1087305	-849504	127	2
2009	WY	Sodium	156577	64530	242	1
2010	VA	Limestone	117397	117197	100	1
2011	CA	Geothermal - Direct Utilization, Millions of BTUs	48086	46295	103	1
2011	ND	Nitrogen	1098	861	127	1
2012	WY	Unprocessed (Wet) Gas	-132500520	99040241	133	23

There are some values that have different signs than the annual total (for instance, in the last row). Then there are some others that have both negative numbers, but the state number is larger (proportionally) than the annual total.

This makes me think that we just shouldn't show percentages for either negative state or totals. But what about when the state total is greater than the nationwide total (and, presumably, there were other states with negative revenues to bring down the nationwide total)?

Thoughts, @RyanSibley, @coreycaitlin, @meiqimichelle, @mentastc?

Note: these numbers are from the new revenue data that Nathan sent us, so they're provisional; but there will certainly be instances of each of these in the final data.

[shawnbot]

Here are the edge cases we need to decide how to represent, if the percentages are even worth listing:

1. Positive state revenue that is greater than the total
2. Positive state revenue when the total is negative
3. Negative state revenue when the total is positive
4. Negative state revenue that is less than the negative total

[shawnbot]

Nathan notes that we should not say "percent of US total revenue" for a given commodity, because we're only working with federal revenue, which accounts for about 25% of all revenue generated by the extractives industry (including private).

[meiqimichelle]

Is this issue mostly resolved now that we have Nathan's categories, or is this something different? And/or, it seems like we're moving away from showing percentages along with many of our revenue/production values, right?

[coreycaitlin]

Nathan's commodities groups solve most of the weird-data issues here. Also, I think we're moving away from showing year-over-year percentages, and we may not wind up showing percentages at all. If we do keep some form of percentages (i.e. "In 2013, federal revenue from coal extraction in Utah accounted for 8 percent of all federal revenue from coal extraction on federal lands"...which is a mouthful), I think there will at least be fewer edge cases.

[coreycaitlin]

Closing this out in favor of #1465, since we'll mostly only be using percentages or rankings in intro sections.