Covid 19 is messing up everything (well my time series continue to work, I have no idea how). And our political process is collapsing. Essentially the distribution is changing.

Can’t argue there … even abstractions like public opinions are changing dramatically, as your

*time series* continues to work.

*Gallup* recently polled

Americans on their viewpoints on a wide range of industries in 2020 (9/8), tracking the percentages of

*favorable*,

*neutral* or

*unfavorable* opinions. Gallup uses the metric of relative favorability (% very/somewhat positive) of twenty disparate industry, and compares 2020 to 2019 for the industries that are the most impacted by the COVID crisis:

Healthcare and farming had the biggest boost in favorability, due to their ongoing essentialness.

The discretionary industries of travel and sports fared the worst in their drop in public opinion. Travel probably sagged in support since it’s basically irrelevant until further notice. Perhaps sports industry favorability is eroding due to the

*kabuki-theater-with-no-audience* nature of televised games and the social-justice campaigns of the various US professional leagues. The article notes that the biggest drops in sports viewer-favorability -- find detailed charts with demographic breakdown on previous link -- were Republicans (-46% drop, year-minus-year) and the mid-range age group (35 to 54, -44%).

*Does trolling -- with no one noticing -- *

*still constitute 'trolling'..???*
The industry-favorability data developed and presented by Gallup is done on a percentage basis, and then those percentages are used in subsequent calculations, like with the above year-to-year changes by demographic. In an

earlier *TalkStats* post (which actually

spurred this thread), the following was proposed, in reference to percentage characterization used in calculations:

THEORY: *Linear characterization of nonlinear functionality convolutes calculations, at best, and, at worst, corrupts capability and understanding, greatly increasing the probability of error and misprojection when utilizing that linear data in subsequent operations.*

Instead of just subtracting same-basis numbers, like Gallup does, to show the linear magnitude of change, let’s look at their industry-favorability study from a

*nonlinear ratio* standpoint for the five industries highlighted for their changes 2019 to 2020. The relevant metric here will be ratio of the

*positive*- to the

*negative*-impression percentages. (The

*neutral*-impression percentages are assumed to be constant year-to-year, allowing for the calculation of the negative-impression percentages in 2019 for the industries.)

The chart below includes the favorability ratio for years 2020 and 2019, as well as the relative change in that metric from 2019 to 2020, for each industry. While the

*Positive/Negative*-ratio data allows the same basic observations as the Gallup tables (above, and

at link) of which industries are viewed favorably, this study also facilitates a better understanding of the system dynamics:

Embracing the nonlinear nature of phenomena -- in this case, public opinion -- and so describing them with rational numbers (i.e., ratios) allows for optimal analyses. It's safe to say that much of the Coronavirus analyses out there have been

*irrational* and

*sub-optimal*.