Below is a critique of a scholarly research paper, "Testing Theories of American Politics: Elites, Interest Groups, and Average Citizens"(TTAMP). The paper concludes, based on statistical analysis, that voters are helpless, and their desires are only achieved when they happen to align with the desires of the powerful, i.e., the superrich and the corporate lobbies. Putting the results of a week of review into a bumper sticker, the paper is invalid, and its conclusion is without merit.
What is scariest about the TTAMP research is this: it would be easy to write a paper only slightly different, that draws a conclusion almost exactly the same, that is just as invalid, and that cannot be refuted without years of research. This is how statistical analysis, masquerading as scientific insight, becomes the handmaiden of fantastical, paranoid belief systems that cause grievous harm not only to our policy making, but even to the very fabric of civil society.
This is a very dry writeup. If you sensibly choose to read no farther than the Executive Summary, the main takeaway is this: when complex statistical analysis violates either common sense, simple logic, or basic laws of nature, disregard it. Or accept it at risk of your intellectual soul.
---------------------
Executive Summary
The paper "Testing Theories of American Politics: Elites, Interest Groups, and Average Citizens"(TTAMP) concludes, based on statistical analysis of legislation, that voters are helpless, and their desires are only achieved when they happen to align with the desires of the powerful, i.e., the superrich and the corporate lobbies.
TTAMP includes a careful, thoughtful list of limitations and qualifiers on their work. However, here in this critique, several even more serious issues are identified with TTAMP that suggest key conclusions -- including the conclusion about voter helplessness -- are invalid. This critique identifies significant reasons to suspect the following statements are true:
Significant Issues with TTAMP
Issues unidentified or disregarded by the authors include:
Failure to distinguish statistical correlation from causation
This is the bane of statistics and it applies here. An example the authors implicitly identify, without acknowledging its broader applicability, is the relationship between government and corporate lobbying. Most papers assume that corporations lobby to get special favors. The authors acknowledge a theory proposed by another investigator that corporations lobby, not to seek special favors, but because the government blackmails corporations into supporting them with the threat of harmful legislation. Whether the legislators are blackmailing the companies, or the companies are bribing the legislators, the statistical correlation is the same. But the causation is completely inverted. Experts who are intimate with multiple industries and their interactions with the government see numerous examples of both. For readers familiar with the evolution of the software industry, one finds that software companies universally ignored the government until the government, on behalf of other powerful lobbies, started imposing legislation, at which point the companies got involved simply for self defense. Of course, once you have an enormous and expensive lobbying machine, built for defense, it seems inevitable that you would eventually start using it for offense as well, flipping the cause-effect relationship. In the end, it becomes difficult to distinguish defense and offense because the power flows are so complexly interrelated: a corporation eventually defends itself from harmful legislation by promoting helpful legislation.
A more general driver of non-causative correlation occurs when both of the correllated systems are caused by an underlying phenomenon. An example in the context of legislative power: citizen majorities may desire legislation that conflicts with forces other than corporations and superrich. The Bill of Rights is such a force. A naive look at polling numbers on gun control would suggest that the gun manufacturers have a lobby more powerful than the will of the citizens. But a more important obstacle to the democratic will is the Second Amendment. Gun laws have been passed by the will of the people despite the whole gun lobby (including entities much more powerful than the manufacturers), only to be struck down by the Supreme Court. The same is true in the field of political campaign speech limitations, which runs straight into the First Amendment. If the superrich and corporations tried to overcome the Constitution they would have the same problems, but one would expect such groups -- especially businesses -- to avoid engaging in such cost-ineffective, futile, and profitless advocacies. As noted in the research, corporations have their greatest success when advocating a position on a matter which no one else cares about or notices. Legislation that assaults the Constitution would flout this self-evidently sensible, profit-optimizing strategy.
Another force which the will of the people sometimes runs against is the force of correctness. For some policies, there is an actual right answer. Such right answers are more likely to be identified and pursued by experts than by any of the traditional power blocs. Consider the setting of interest rates by the Fed. It is generally considered favorable to the common citizen to have lower interest rates, to reduce the burden of their debts. Meanwhile it is considered favorable to the supperich to have high interest rates, to increase their interest on savings. Since the interest rate is adjusted upwards as often as it is adjusted downwards, statistical assessment would say that the superrich overpower the will of the people about half the time. But this not a causal relationship. In fact, Congress intentionally set a firewall between the passions of all powers (including even themselves) and the rate setting process, leaving it to a team of experts. Even experts are not always right with complex problems only partly understood even by people who have studied the problem their entire lives. But Congress understood that a popular vote to specify the interest rate would be no more sensible than having a popular vote certify the design of a new aircraft. Moreover, the Founders intended this type of governance by expertise to be included as part of the system. They intentionally avoided allowing a mechanism for direct popular referendum because they wanted a representative body, with more time to deliberate, to assess the merits of all legislation. They intended for such deliberation to use expert testimony to help make their decision, even if that decision conflicted with the will of the majority.
Role of the Super-Expert
The opinion of the super-expert, as distinct from the citizens, the superrich, and the corporations, needs to be considered in future research. This will be harder than it appears because corporations represent not only a major source of corrupt self-interest but also a major source of expertise on industrial policy. It is often impossible to distinguish where the wisdom of corporate expertise leaves off and the ruthlessly harmful profit-maximization begins. Consider a plausible scenario with the $15 minimum wage. As Congress deliberates, a series of experts including liberal economists testify that jobs will be lost and unemployment will rise. A washing machine manufacturer testifies that they will have to move their factory to Mexico. Walmart demonstrates on a spreadsheet the 30,000 jobs they will have to eliminate. In this case, the government may sensibly decide to thwart the will of the people in order to protect the people from harm. Here again expertise, not power, could play the decisive role, even though simplistic statistics suggest the victory of corporate power.
Loss of Objectivity On Data Representing the Superrich
It seems likely that the authors lost objectivity about the validity of the data-set intended to represent the will of the economic elite, i.e., the one-percenters, i.e., the superrich. As the authors observe, the data-set is actually based on the views of the top 10% of the population in terms of income. But the authors repeatedly claim, without evidence, that the influence of the superrich is most probably greater than that represented by the data. But this is not merely unproven, it is most probably false. For the top 1% to have more influence than that represented by the top 10% that includes them, the will of the 90%-99% percentiles must be inversely correlated with the will of the 1%. In the whole of this study, the only negative correlation is between businesses and people, and even that negative correlation is tiny. The authors' claim of much greater superrich influence seems farfetched. Yet the authors state it repeatedly, suggesting that, probably subconsciously, they are attempting to use the ad nauseum persuasion technique to convince readers (and possibly themselves) that their data is more valid than it actually is. In practice, the thoughtful reader must conclude the data-set represents the will of those who, with incomes of $146,000 and more, are "barely elite", and the superrich wield only an unquantifiable subset of that influence.
Uncredible perfection of the data representing the will of the people
The irrelevance of the will of the majority represented in the flat line on Figure 1 is simply too perfect. It does not pass a back-of-the-envelope calculation or a sniff test. The will of a 90% majority is visually indistinguishable from the will of the 0% minority, with nary a bump or hiccup in its linear precision. This makes it strikingly different, not only in slope, but also in noiseless simplicity, from the other lines representing other influences. Consider carefully the extremity of the claim made by this straight line. To achieve this level of perfection, there must be so few examples of majorities succeeding against the will of elites and corporations that such examples should be almost impossible to identify. Even mere anecdotal numbers of examples are enough to invalidate this perfection. Yet a number of examples of the people's will prevailing pop to mind almost effortlessly. So here are a couple examples:
Limitations on freedom of speech of tobacco companies: in the face of absolute opposition from the supremely powerful tobacco industry, even flying in the face of the Bill of Rights Freedom of Speech, the people prevailed.
Ban of DDT: The outrage of common citizens overcame virtually the entire collection of corporate powers, for whom DDT was a godsend.
War in Afghanistan: The majority of the people favored going to war with Afghanistan after 9/11. This was not in the interest of any corporate lobby -- even the military industrial complex generally finds actual war unfavorable because funds that could be turned into profits through the purchase of more weapons is redirected to paying for logistical support for troups in the field.
Marijuana Legalization: Once marijuana is uniformly legalized, a corporate lobby defending its legality will inevitably arise. But today there is no vested corporate interest in the matter. Since legalization is currently driven by referendum, it is strictly a matter of simple majorities at the polls.
Every kind of air pollution control and water pollution control, up to and including the recent EPA rulings on carbon dioxide emissions, is a counterexample to the flat line. In each of these cases, the will of the people faced off against the desires of the most powerful corporate lobbies and still won.
Just these examples seem likely to cause the curve to be less level than the flat line resulting from the analyzed data. The simplest explanation for this perfect absence of variation in the line is simply that the data set is as uncorrelated with people's desires as it is uncorrelated with legislative outcomes. In other words, the data-set is invalid. All conclusions based on this data-set should be disregarded until a more reliable data-set is identified.
Uncredible nonexistence of influence of Citizen Groups
Similar skepticism needs to be applied to the conclusions drawn about special interest groups comprised of concerned citizens. Even anecdotal evidence should be enough to show that the authors' assessed absence of citizen interest groups is in error. Such evidence includes:
The NRA: While attention was drawn above to the way in which the Second Amendment interferes with the will of the people, in fact few conflicts with the Constitution arise because of the acknowledged might of the NRA. The NRA is generally given much of the credit (and the blame) for flipping the House from Democratic to Republican in 1992. Claims that this was the result of corporate lobbying by gun manufacturers seem unlikely: gun manufacturers have none of the size, revenues, or employee headcounts of a General Motors. They would be overrun easily if the NRA switched sides and endorsed a ban on firearms.
The AARP: The AARP's fierce defense of every form of government largesse for senior citizens prevails regardless of its impact on the long term health of the national debt, deficit, and economy.
The banning of Alar and similar outbursts: a flash crowd of concerned citizens arose, based on sensationalistic media coverage, to ban Alar. No corporate or supperich interest was served by this, and it flouted the experts. It was a pure victory of the concerned citizen. Though it also suggests that the influence of media sensationalism should be included in future research.
Both Pro-Choice and Pro-Life legislative initiatives: There is no corporate or superrich interest in the legislative status of abortion. The state of this issue is decided entirely by the rise and fall of majoritarian pluralities (i.e., citizen interest groups) in conflict with each other.
Are such victories by citizen interest groups democratic or non-democratic? The authors seem to acknowledge that such influence would be democratic ... if only their statistics showed such influence existed. It is certainly different from the simpler democratic concept of a majority vote. But if the 30% of most concerned citizens on one issue get their way on that issue, another 30% get their way on another issue that is their greatest concern, and another 30% get their way on yet a different issue that is their most important consideration, then everyone is happier with the overall outcome than they would be if simple majorities rejected all their greatest desires. If this is not democracy, it is actually something better. And if people who neither care nor think about any governmental action never get the outcome that drops from their lips when asked by a pollster, is that a failure of democracy, or simply a sensible result of disinterest?
Unidentified influences on legislature comprise over half of all influence
It is clear that at least one source of influence, at least as powerful as either corporate lobbies or the interests of the economically barely elite, remains unidentified. Even when 100% of corporate lobbies share consensus, only 35% of their desires are met. Even when 100% of the barely elite share consensus, only 35% of their desires are met. Even if superrich desires were fully uncorrelated with corporate desires, and all the superrich shared full consensus on everything, and all corporations shared full consensus on everything, this leaves 30% of legislative action driven by no identifiable force except random Brownian motion. Since legislative action takes significant, even dramatic effort with substantial costs, this seems unlikely. The strong probability is over half of all influences on legislation remain unidentified.
Conclusion
Where did the other 50% of influence go? Given the other flaws in this research identified in this paper, a simple hypothesis is that the missing influence is wielded by democratic forces, in the shape of some combination of simple majority rule and citizen interest groups. As the authors note, further research is needed. This critique suggests that no conclusion from TTAMP should be considered valid unless verified by analysis of more trustworthy data sets.
>> All Blog Posts <<
What is scariest about the TTAMP research is this: it would be easy to write a paper only slightly different, that draws a conclusion almost exactly the same, that is just as invalid, and that cannot be refuted without years of research. This is how statistical analysis, masquerading as scientific insight, becomes the handmaiden of fantastical, paranoid belief systems that cause grievous harm not only to our policy making, but even to the very fabric of civil society.
This is a very dry writeup. If you sensibly choose to read no farther than the Executive Summary, the main takeaway is this: when complex statistical analysis violates either common sense, simple logic, or basic laws of nature, disregard it. Or accept it at risk of your intellectual soul.
---------------------
Executive Summary
The paper "Testing Theories of American Politics: Elites, Interest Groups, and Average Citizens"(TTAMP) concludes, based on statistical analysis of legislation, that voters are helpless, and their desires are only achieved when they happen to align with the desires of the powerful, i.e., the superrich and the corporate lobbies.
TTAMP includes a careful, thoughtful list of limitations and qualifiers on their work. However, here in this critique, several even more serious issues are identified with TTAMP that suggest key conclusions -- including the conclusion about voter helplessness -- are invalid. This critique identifies significant reasons to suspect the following statements are true:
- The data set intended to represent the will of the people is invalid.
- The authors lost objectivity with respect to the data set intended to represent the will of the economic elite, i.e., the "one-percenters".
- The data set intended to represent the one-percenters in fact principally represents the will of the "barely elite", i.e., people making $146,000 or more per year.
- The will of the "super-expert" needs to be included in future work. Sometimes there is an actual correct answer, based in laws of physics and economics, that stands outside the control of groups of people. And once in a while the political process responds to the need for correctness -- a need only defended by the super-expert.
- The claim that citizen interest groups have almost no influence is false.
- At least 50% of the total influence in government policy creation remains unidentified, i.e., the influence of the barely elite and corporations accounts for less than half of all influence.
- A simple interpretation is that the missing 50% (possibly much more) of influence is wielded by democratic processes.
Significant Issues with TTAMP
Issues unidentified or disregarded by the authors include:
- Failure to distinguish statistical correlation from causation
- Disregard of the role of the super-expert
- Loss of Objectivity On Data Representing the Superrich
- Uncredible perfection of the data representing the will of the people
- Uncredible nonexistence of influence by concerned citizen groups
- Unidentified influences on legislature comprising over half of all influence
Failure to distinguish statistical correlation from causation
This is the bane of statistics and it applies here. An example the authors implicitly identify, without acknowledging its broader applicability, is the relationship between government and corporate lobbying. Most papers assume that corporations lobby to get special favors. The authors acknowledge a theory proposed by another investigator that corporations lobby, not to seek special favors, but because the government blackmails corporations into supporting them with the threat of harmful legislation. Whether the legislators are blackmailing the companies, or the companies are bribing the legislators, the statistical correlation is the same. But the causation is completely inverted. Experts who are intimate with multiple industries and their interactions with the government see numerous examples of both. For readers familiar with the evolution of the software industry, one finds that software companies universally ignored the government until the government, on behalf of other powerful lobbies, started imposing legislation, at which point the companies got involved simply for self defense. Of course, once you have an enormous and expensive lobbying machine, built for defense, it seems inevitable that you would eventually start using it for offense as well, flipping the cause-effect relationship. In the end, it becomes difficult to distinguish defense and offense because the power flows are so complexly interrelated: a corporation eventually defends itself from harmful legislation by promoting helpful legislation.
A more general driver of non-causative correlation occurs when both of the correllated systems are caused by an underlying phenomenon. An example in the context of legislative power: citizen majorities may desire legislation that conflicts with forces other than corporations and superrich. The Bill of Rights is such a force. A naive look at polling numbers on gun control would suggest that the gun manufacturers have a lobby more powerful than the will of the citizens. But a more important obstacle to the democratic will is the Second Amendment. Gun laws have been passed by the will of the people despite the whole gun lobby (including entities much more powerful than the manufacturers), only to be struck down by the Supreme Court. The same is true in the field of political campaign speech limitations, which runs straight into the First Amendment. If the superrich and corporations tried to overcome the Constitution they would have the same problems, but one would expect such groups -- especially businesses -- to avoid engaging in such cost-ineffective, futile, and profitless advocacies. As noted in the research, corporations have their greatest success when advocating a position on a matter which no one else cares about or notices. Legislation that assaults the Constitution would flout this self-evidently sensible, profit-optimizing strategy.
Another force which the will of the people sometimes runs against is the force of correctness. For some policies, there is an actual right answer. Such right answers are more likely to be identified and pursued by experts than by any of the traditional power blocs. Consider the setting of interest rates by the Fed. It is generally considered favorable to the common citizen to have lower interest rates, to reduce the burden of their debts. Meanwhile it is considered favorable to the supperich to have high interest rates, to increase their interest on savings. Since the interest rate is adjusted upwards as often as it is adjusted downwards, statistical assessment would say that the superrich overpower the will of the people about half the time. But this not a causal relationship. In fact, Congress intentionally set a firewall between the passions of all powers (including even themselves) and the rate setting process, leaving it to a team of experts. Even experts are not always right with complex problems only partly understood even by people who have studied the problem their entire lives. But Congress understood that a popular vote to specify the interest rate would be no more sensible than having a popular vote certify the design of a new aircraft. Moreover, the Founders intended this type of governance by expertise to be included as part of the system. They intentionally avoided allowing a mechanism for direct popular referendum because they wanted a representative body, with more time to deliberate, to assess the merits of all legislation. They intended for such deliberation to use expert testimony to help make their decision, even if that decision conflicted with the will of the majority.
Role of the Super-Expert
The opinion of the super-expert, as distinct from the citizens, the superrich, and the corporations, needs to be considered in future research. This will be harder than it appears because corporations represent not only a major source of corrupt self-interest but also a major source of expertise on industrial policy. It is often impossible to distinguish where the wisdom of corporate expertise leaves off and the ruthlessly harmful profit-maximization begins. Consider a plausible scenario with the $15 minimum wage. As Congress deliberates, a series of experts including liberal economists testify that jobs will be lost and unemployment will rise. A washing machine manufacturer testifies that they will have to move their factory to Mexico. Walmart demonstrates on a spreadsheet the 30,000 jobs they will have to eliminate. In this case, the government may sensibly decide to thwart the will of the people in order to protect the people from harm. Here again expertise, not power, could play the decisive role, even though simplistic statistics suggest the victory of corporate power.
Loss of Objectivity On Data Representing the Superrich
It seems likely that the authors lost objectivity about the validity of the data-set intended to represent the will of the economic elite, i.e., the one-percenters, i.e., the superrich. As the authors observe, the data-set is actually based on the views of the top 10% of the population in terms of income. But the authors repeatedly claim, without evidence, that the influence of the superrich is most probably greater than that represented by the data. But this is not merely unproven, it is most probably false. For the top 1% to have more influence than that represented by the top 10% that includes them, the will of the 90%-99% percentiles must be inversely correlated with the will of the 1%. In the whole of this study, the only negative correlation is between businesses and people, and even that negative correlation is tiny. The authors' claim of much greater superrich influence seems farfetched. Yet the authors state it repeatedly, suggesting that, probably subconsciously, they are attempting to use the ad nauseum persuasion technique to convince readers (and possibly themselves) that their data is more valid than it actually is. In practice, the thoughtful reader must conclude the data-set represents the will of those who, with incomes of $146,000 and more, are "barely elite", and the superrich wield only an unquantifiable subset of that influence.
Uncredible perfection of the data representing the will of the people
The irrelevance of the will of the majority represented in the flat line on Figure 1 is simply too perfect. It does not pass a back-of-the-envelope calculation or a sniff test. The will of a 90% majority is visually indistinguishable from the will of the 0% minority, with nary a bump or hiccup in its linear precision. This makes it strikingly different, not only in slope, but also in noiseless simplicity, from the other lines representing other influences. Consider carefully the extremity of the claim made by this straight line. To achieve this level of perfection, there must be so few examples of majorities succeeding against the will of elites and corporations that such examples should be almost impossible to identify. Even mere anecdotal numbers of examples are enough to invalidate this perfection. Yet a number of examples of the people's will prevailing pop to mind almost effortlessly. So here are a couple examples:
Limitations on freedom of speech of tobacco companies: in the face of absolute opposition from the supremely powerful tobacco industry, even flying in the face of the Bill of Rights Freedom of Speech, the people prevailed.
Ban of DDT: The outrage of common citizens overcame virtually the entire collection of corporate powers, for whom DDT was a godsend.
War in Afghanistan: The majority of the people favored going to war with Afghanistan after 9/11. This was not in the interest of any corporate lobby -- even the military industrial complex generally finds actual war unfavorable because funds that could be turned into profits through the purchase of more weapons is redirected to paying for logistical support for troups in the field.
Marijuana Legalization: Once marijuana is uniformly legalized, a corporate lobby defending its legality will inevitably arise. But today there is no vested corporate interest in the matter. Since legalization is currently driven by referendum, it is strictly a matter of simple majorities at the polls.
Every kind of air pollution control and water pollution control, up to and including the recent EPA rulings on carbon dioxide emissions, is a counterexample to the flat line. In each of these cases, the will of the people faced off against the desires of the most powerful corporate lobbies and still won.
Just these examples seem likely to cause the curve to be less level than the flat line resulting from the analyzed data. The simplest explanation for this perfect absence of variation in the line is simply that the data set is as uncorrelated with people's desires as it is uncorrelated with legislative outcomes. In other words, the data-set is invalid. All conclusions based on this data-set should be disregarded until a more reliable data-set is identified.
Uncredible nonexistence of influence of Citizen Groups
Similar skepticism needs to be applied to the conclusions drawn about special interest groups comprised of concerned citizens. Even anecdotal evidence should be enough to show that the authors' assessed absence of citizen interest groups is in error. Such evidence includes:
The NRA: While attention was drawn above to the way in which the Second Amendment interferes with the will of the people, in fact few conflicts with the Constitution arise because of the acknowledged might of the NRA. The NRA is generally given much of the credit (and the blame) for flipping the House from Democratic to Republican in 1992. Claims that this was the result of corporate lobbying by gun manufacturers seem unlikely: gun manufacturers have none of the size, revenues, or employee headcounts of a General Motors. They would be overrun easily if the NRA switched sides and endorsed a ban on firearms.
The AARP: The AARP's fierce defense of every form of government largesse for senior citizens prevails regardless of its impact on the long term health of the national debt, deficit, and economy.
The banning of Alar and similar outbursts: a flash crowd of concerned citizens arose, based on sensationalistic media coverage, to ban Alar. No corporate or supperich interest was served by this, and it flouted the experts. It was a pure victory of the concerned citizen. Though it also suggests that the influence of media sensationalism should be included in future research.
Both Pro-Choice and Pro-Life legislative initiatives: There is no corporate or superrich interest in the legislative status of abortion. The state of this issue is decided entirely by the rise and fall of majoritarian pluralities (i.e., citizen interest groups) in conflict with each other.
Are such victories by citizen interest groups democratic or non-democratic? The authors seem to acknowledge that such influence would be democratic ... if only their statistics showed such influence existed. It is certainly different from the simpler democratic concept of a majority vote. But if the 30% of most concerned citizens on one issue get their way on that issue, another 30% get their way on another issue that is their greatest concern, and another 30% get their way on yet a different issue that is their most important consideration, then everyone is happier with the overall outcome than they would be if simple majorities rejected all their greatest desires. If this is not democracy, it is actually something better. And if people who neither care nor think about any governmental action never get the outcome that drops from their lips when asked by a pollster, is that a failure of democracy, or simply a sensible result of disinterest?
Unidentified influences on legislature comprise over half of all influence
It is clear that at least one source of influence, at least as powerful as either corporate lobbies or the interests of the economically barely elite, remains unidentified. Even when 100% of corporate lobbies share consensus, only 35% of their desires are met. Even when 100% of the barely elite share consensus, only 35% of their desires are met. Even if superrich desires were fully uncorrelated with corporate desires, and all the superrich shared full consensus on everything, and all corporations shared full consensus on everything, this leaves 30% of legislative action driven by no identifiable force except random Brownian motion. Since legislative action takes significant, even dramatic effort with substantial costs, this seems unlikely. The strong probability is over half of all influences on legislation remain unidentified.
Conclusion
Where did the other 50% of influence go? Given the other flaws in this research identified in this paper, a simple hypothesis is that the missing influence is wielded by democratic forces, in the shape of some combination of simple majority rule and citizen interest groups. As the authors note, further research is needed. This critique suggests that no conclusion from TTAMP should be considered valid unless verified by analysis of more trustworthy data sets.
>> All Blog Posts <<