Unexpectedly Intriguing!
06 September 2016

We have a special example of junk science to present today, because this particular example represents the first time we've had to address a repeat offender and to revisit a category from our checklist of how to detect junk science as part of this series.

Here is the category from our checklist that we'll be specifically discussing in today's example of junk science.

How to Distinguish "Good" Science from "Junk" or "Pseudo" Science
Aspect Science Pseudoscience Comments
Inconsistencies Observations or data that are not consistent with current scientific understanding generate intense interest for additional study among scientists. Original observations and data are made accessible to all interested parties to support this effort. Observations of data that are not consistent with established beliefs tend to be ignored or actively suppressed. Original observations and data are often difficult to obtain from pseudoscience practitioners, and is often just anecdotal. Providing access to all available data allows others to independently reproduce and confirm findings. Failing to make all collected data and analysis available for independent review undermines the validity of any claimed finding. Here's a recent example of the misuse of statistics where contradictory data that would have avoided a pseudoscientific conclusion was improperly screened out, which was found after all the data was made available for independent review.

As we revisit the category of Inconsistencies, we're fortunate to have access to all of the source data that the originator of today's example of junk science used in their analysis, which is what allows us to expose the methods by which they either ignored relevant data or suppressed it in order to sustain their pseudoscientific claims.

But better than that, today's example of junk science will take on some really comic proportions. Let's get started, shall we?




Since we're dealing with a repeat offender, let's start by reviewing their now classic example of junk science that fails the scientific requirement of falsifiability, where the offender ruled out drought as a factor that contributed to the impairment of the state of Kansas' economic performance in the period since 2010. The way they achieved that result was to used "baked data" that just so happened to omit any contribution to the state's economy from its agricultural industry, which just happens to be the industry that would be most negatively affected by severe drought conditions. The following chart they produced then shows their desired result of no apparent correlation between the incidence of severe drought conditions, as indicated by the Palmer Drought Severity Index, and the Federal Reserve Bank of Philadelphia's state coincident index indicating Kansas' economic performance.

Philly Fed GSP and Palmer Drought Index Overlay Chart

For a junk scientist, there's nothing like using deficient data to ensure that their desired results are baked in, which is why it is so often so difficult to get their data! Fortunately, in this case the data they used was available and proved to be sufficient to demonstrate that their resulting analysis would automatically qualify as pseudoscience.

But was that outcome the result of incompetent analysis on their part? Or was the originator of the example deliberately seeking to produce that outcome?

We had a unique opportunity to find out. Over a year ago, in the comments on their web site, we challenged their belief that the presence of severe drought in Kansas was in no way responsible for any portion of the lackluster performance of the state's economy, which is especially remarkable in that such a claim could be considered the economics equivalent of climate change denial. We advanced the argument that the incidence of severe drought from late 2010 through mid-2014 would explain a significant portion of the impairment of Kansas' economic performance since 2010, and particularly in 2012, which was a direct challenge to their rather unique theory that the adoption of certain fiscal policies by the state's governor since they took office in January 2011 were solely responsible for that outcome, and which for whatever reason, would seem to be a theory that they cannot abide having challenged with contradictory evidence.

A little over a month later, they produced their first junk science example in a sad attempt to specifically refute our argument.

We let that example go by for months without comment in any forum, so if they were to ever correct the record on their own, they had more than ample time to do so. In fact, it wasn't until we wrote the allegorical post "How to Bake a Highly Deficient Cake" that we ever publicly addressed any aspect of the pseudoscience they had chosen to put on display as the pseudoscience it is.

Finally, we decided enough time had passed and we called them out on their site again for the first time in months with little more than six words, which included a link to our "deficient cake baking" allegory.

That single action on our part sent them into a manic frenzy of activity where they spent a good portion of the next 27 hours (and actually longer) actively engaged in trying to refute the possibility that drought explains any part of how Kansas' economy performed after 2010.

We do not believe that any of that activity would have occurred if there were not some gross deficiency in their previous analysis. If it had been legitimate, there would not and should not have been any need for it, nor any need to produce the new example that we'll be discussing today.

By challenging them for having used "baked data" to ensure their previous results, we succeeded in compelling them to use actual state-level GDP data to do their analysis, which gives us the means to see how they handled the data in their subsequent analysis to support their claims. What follows is the relevant portion of the analysis they produced that reinforces their previous climate change denialism. The data they used was originally published by the BEA on 2 March 2016 and is the exact same data we'll be using in our discussion today.

BEA KS/US GDP Log Ratio and Palmer Drought Index Overlay Chart

The astute observer will note that there is a pretty poor correlation between relative growth of Kansas and the drought index. More formally, run a regression over the 2005Q3-2015Q3 period:

ΔyKSt = -0.001 + 0.106 ΔyKSt-1 + 1.300ΔyUSt + 0.0006PDSIt + ut

Adj-R2 = 0.45, NObs = 41, SER = 0.009, DW=1.65. bold denotes significance at the 10% msl, using HAC robust standard errors.

Note that the drought index coefficient is not statistically significant. I’m sure with enough work, I could get it to be statistically signficant[sic] in some specification. In any case, it doesn’t pop out easily.

This is the first bit of comedy the originator of this new example of junk science has provided for our entertainment. They have gone from the position that there is absolutely no correlation between the incidence of drought and Kansas' economic performance since 2010 to perhaps some statistically significant correlation that would "pop out" with "enough work" to reveal. Oh, if only it weren't so difficult! Oh, if only they had more time to spend on the problem than a minimum of 27 hours!

Want to see how easy it is?

Here's an idea that might automatically come to the mind of any competent analyst, if not the originator of today's example of junk science. Why not drill down into the state-level GDP data to directly extract the state's real Agriculture output for the period in question on its own, seeing as it is the exact data that would both show the real effect of drought on the state's agricultural output and which is completely omitted from the Federal Reserve Bank of Philadelphia's state coincident indices that was used in their previous "baked" analysis, and then see if there is any significant correlation?

It's not like it's hard. At all. By our estimate, it would take less than five minutes to execute if one had already done all this other "analysis". Especially seeing as that all one would need to do is to substitute the top line of the BEA's real GDP data that was used in the analysis, indicating the sum of the contributions to GDP from all industries in the state, with the specific data for Agriculture, which is just two lines below it in the BEA's state level GDP data....

Real GDP by state (Kansas) - Source: BEA, 2 March 2016 Release

Of course, our time estimate assumes competence on the part of the analyst. For someone who may not be particularly analytically gifted, if you know what we mean, such a simple substitution of a line of data might represent a tremendous time commitment that it would most certainly be an almost insurmountable obstacle to their progress. Then again, for something that seems an awful lot like incompetence, the results they published are indistinguishable from junk science, where directly relevant data that might contradict their preferred narrative has been very directly ignored.

Let's next look at a chart we threw together using the exact same data that was available at the time the originator of the example we're now discussing created it, but which was clearly ignored. It shows the same data for the Palmer Drought Severity Index (PDSI) on the left-hand scale, but now shows the real contribution of Kansas' agricultural sector in terms of constant 2009 U.S. dollars on the right hand scale. Judge for yourself whether any portion of these two sets of data are correlated during the period of the drought.

Kansas Statewide Average Palmer Drought Severity Index and Real GDP (Agriculture, Forestry, Fishing & Hunting), 2005-Q1 through 2015-Q3Update 10 September 2016: We accidentally posted an earlier version of the chart above, which we've replaced with the intended final version now shown. Click here to see the original.

We can quickly see pretty darn quickly that there is a pretty darn strong correlation between the drought severity index for Kansas and the contribution of agriculture to the state's real GDP, which is confirmed by the actual correlation coefficients we calculated (Pearson) and determined to be statistically significant (t-value)! Moreover, we see that the incidence of severe drought in Kansas in the period from the end of 2010 to mid-2014 directly coincides with a period when growth in the state's agriculture industry stalled out compared to its pre-drought trajectory, indicating a significant reduction in the contribution to GDP growth from the agricultural sector of the state's economy.

There are additional problems. In their attempt at a correlation analysis, they used the "Kansas/US GDP log ratio", in which they incorporated the state's entire GDP for all its industries, which includes all the industries other than agriculture that are much less impacted by the presence of drought conditions. Since the combined contributions for all industries are what produce the state's total GDP, the effect of using this combination of all economic sectors is to suppress or conceal the full impact of drought on the state's agricultural sector, which in the absence of drought would have added significantly more to the state's total real GDP.

We can illustrate that effect much more directly in the following chart, in which we substituted Kansas' real GDP for all industries for the state's agricultural contribution to its GDP in our previous chart and recalculated all the correlations. It took us less than five minutes.

Kansas Statewide Average Palmer Drought Severity Index and Real GDP (All Industries), 2005-Q1 through 2015-Q3Update 10 September 2016: We accidentally posted an early draft version of the chart above, which we've replaced with the intended final version now shown. Click here to see the original.

In this chart, we can see from the correlation factors that apply during the most severe period of drought in the state of Kansas that its impact has been suppressed to indicate almost no correlation by confounding it with of the economic contribution of all other industries, particularly during the most severe period of drought that so clearly affected the performance of the state's agriculture industry between January 2011 and June 2014. That outcome is a very different result from what we saw when we only considered the contribution of the state's agriculture-related GDP.

We can also confirm that the state's drought-stricken agriculture industry made up a significant percentage of the impairment of the state's GDP, contrary to the claims of the originator of today's junk science example. [Note: The state-level GDP data since then has been significantly revised, particularly in the period from 2014-Q2 to the present, which now confirms that the negative impact from its severe and prolonged drought has continued to weigh on the state's economic output, which is a direct consequence of how much it caused the state's agricultural sector to shrink in very real terms, lowering the state's real GDP by somewhere between $4.4 and $4.7 billion according to current estimates.]

That concludes today's example, which is far from the last produced by its originator. In fact, the techniques they used to cook their analysis in this analysis would appear to be their preferred M.O. to generate their desired and predetermined results, producing several additional examples that share a similar deficiency with today's example over just the last month and a half, although they have also managed to check off several other categories in the junk science checklist as well during that time.

On a more personal note, we've heard that the originator of today's example of junk science is upset that we keep returning to the subject of drought affecting Kansas economy and wonders why we do. From our perspective, that's a lot like O.J. Simpson complaining about people always asking him about the murders of Nicole Brown and Ron Goldman, only in this case, the metaphorical dead bodies that are piling up are being used to arbitrarily dismiss real evidence that explains a large part of reality both more simply and more convincingly than their preferred narrative, which would appear can only be propped up using their pseudoscientific practices.

Speaking of which, to date, they have not retracted any of their findings that are based on their pseudoscientific practices, which we can confirm extends back over a number of years and would also appear to cover many aspects of their published analyses.

It's not like our repeat offender couldn't have taken an alternative course of action where they could have avoided such unpleasantness as this public exposure, but it is far too late for that now, as they apparently cannot restrain themselves from engaging in bad behavior. From our own example, when we make mistakes in the analyses we present, we address them and even thank those who point them out. It works pretty well for us, but then, there's a difference between unintentional mistakes and premeditated ones, which if you follow each of the links above, you'll find we've demonstrated applies to the repeat offender.

Labels:

About Political Calculations

Welcome to the blogosphere's toolchest! Here, unlike other blogs dedicated to analyzing current events, we create easy-to-use, simple tools to do the math related to them so you can get in on the action too! If you would like to learn more about these tools, or if you would like to contribute ideas to develop for this blog, please e-mail us at:

ironman at politicalcalculations

Thanks in advance!

Recent Posts

Indices, Futures, and Bonds

Closing values for previous trading day.

Most Popular Posts
Quick Index

Site Data

This site is primarily powered by:

This page is powered by Blogger. Isn't yours?

CSS Validation

Valid CSS!

RSS Site Feed

AddThis Feed Button

JavaScript

The tools on this site are built using JavaScript. If you would like to learn more, one of the best free resources on the web is available at W3Schools.com.

Other Cool Resources

Blog Roll

Market Links

Useful Election Data
Charities We Support
Shopping Guides
Recommended Reading
Recently Shopped

Seeking Alpha Certified

Archives