With every cause comes a petition, or an online poll of some sort. Evidence that some number of people agree on something is intended to lend weight and legitimacy to political proposals. But does it really?
In 1915, a petition signed by 350 000 schoolchildren was handed to William Jennings Bryan, then the Secretary of State of the United States. It was to be delivered to the leaders of the European nations embroiled in a war in which the US, up to then, had remained resolutely neutral.
The petition called for peace among the warring nations. It did not end the war. On the contrary, two years later, the US itself would enter the hostilities.
In the New York Times of 12 August 1885, we find this delightful little article:
GALVESTON, Texas, Aug. 11. – In a suit at Longview the Hon. John W. Duncan, one of the attorneys in the case, argued that no value or reliance could be placed upon petitions promiscuously signed by citizens. Opposing counsel refuted Mr. Duncan’s argument scathingly, and characterized it as absurd and idiotic. The court and the jury were also against Mr. Duncan. They believed that petitions in any cause, if signed by good men, were entitled to respectful consideration. Chagrined and defeated, but not convinced that he was wrong, Mr. Duncan quietly went to work to prove that petitions were of no value. His ingenious zeal was rewarded yesterday, when in open court he read a solemn petition praying that the court would hang their fellow townsman, Luke Howard, who is one of the most respectable and enterprising citizens of the place. Mr. Duncan displayed the names of all the county officers and of 50 business men signed to this ludicrous petition. Among the signers were the brother-in-law and father-in-law of Howard. Every signature was proved to be genuine, and last night it was conceded that Lawyer Duncan had proved his argument.
This case would find an echo, 112 years later, in the 1997 science fair project of Nathan Zoner, a junior high student in fly-over America. Entitled ‘How gullible are we?’, the project involved a petition that gave a detailed description of the mortal dangers to humans and the environment of a chemical known as dihydrogen monoxide, also known as hydroxyl acid. It called for the immediate ban of this ‘unrecognised killer’.
Forty-three of his 50 classmates signed the petition against this substance, which is, of course, plain water. The petition travelled around the world. It, and others like it, attracted overwhelming support, illustrating in shocking terms the crazy causes one can get an ignorant public to support.
Marginal utility
Petitions are an inescapable feature of the modern political landscape. Websites such as Avaaz and Change.org have made a veritable industry out of online petition writing. The subject matter of these petitions is as varied and banal as the number of people writing them.
They are the polished turds of slacktivism, which since the rise of the internet more than a quarter-century ago has plagued the world with forwarded emails, viral videos, likes and reposts on behalf of causes great and small.
There are many issues with petitions, all of which make them only marginally useful as tools of activism.
The first is that one can never be sure that the content of a petition is a full and fair description of reality. Most are phrased in a manner that would predispose the reader to the cause of the petition writer, and bias them against the target of the petition. Many include outright falsehoods, or omit facts that could materially alter a reader’s opinion.
It is a chore to fact-check a petition, and when presented with one in a public setting, it is impossible. Yet do you want to be the guy who doesn’t sign a petition to save the children from the paedophile ring operated by global elites through pizza parlours and online furniture stores?
Petitions also are entirely one-sided. If an activist starts a petition in opposition to a proposed building development – say, a mosque, or an Amazon headquarters – it does not record those who are in favour of said development.
Sure, 57 000 people have signed a petition to block the redevelopment of a small golf course in Cape Town to make way for a new mixed-use development that would include a local headquarters for Amazon, but who are they, and do they represent a popular majority in a city of 4.7 million people? Have they considered all the issues that are not disclosed in the petition text?
I very often see petitions floating around and wonder where I get to voice my objection to the petition. The fact that one cannot express opposition merely confirms that a petition is an inherently biased instrument, that only has legitimacy if an extraordinarily large number of people support it.
When petitions shouldn’t work
Petitions sometimes do work. However, because of their inherent shortcomings, they shouldn’t. Take the SumOfUs petition in the wake of the shooting of Cecil the Lion, as an example. It petitioned US airlines to cease transporting animal trophies. Over a quarter of a million people, most of them likely members of the wealthy, foreign, urban elite, signed the petition. The airlines caved.
What this petition couldn’t tell them, however, was how many people were opposed to the trophy transport ban. It couldn’t tell them what African countries, which actually have responsibility for wildlife conservation, thought about it.
It didn’t consider potential counter-arguments, such as that prohibiting the transport of trophies would be a major financial blow to private game ranches and reserves, and even national wildlife reserves that rely on well-regulated hunting to fund their conservation activities.
Nobody told them that cracking down on trophy hunting could lead to a decline in wild populations and a rise in poaching, or that anti-hunting campaigns are often dishonest, and harmful to the very animals they claim to care about.
Petitions do have a limited role to play in raising public and media awareness, bringing public concerns to the attention of politicians or companies, or just rallying the troops for a broader campaign of activism. At best, however, they’re a first step in a campaign; a tactic, rather than a strategy.
They should never be a basis for public policy.
For every useful petition (like this one) that draws attention to a real issue, describes it fairly, and proposes a change that can realistically be effected through sustained engagement, there are a hundred that are badly written, tendentious, unrealistic, dishonest, and/or entirely pointless.
Polls
To counteract one of the major flaws of petitions, their one-sidedness, a lot of people prefer to do polls, in which respondents can select from two or more options. In principle, this should be an improvement upon petitions, but in practice, it seldom is.
Those reader polls you find daily on the websites of various supposedly respectable news organisations are entirely meaningless. They’re entertainment.
Because they’re written by journalists with no training in how to write survey questions, the questions are often leading, loaded, biased, or just plain confusing.
Even on the rare occasion where the question is well-written and the options are carefully chosen to convey a true reflection of the opinions of the respondents, they are fraught with other sources of bias.
The biggest of these is self-selection bias. A poll on a website is, by necessity, only answered by people who actually visit that website. If the site belongs to a newspaper, this might affect the political leaning of the audience. If the site belongs to an activist organisation or lobby group, it might attract only those people who take an active interest in the subject.
People are also differently motivated to answer questions. Those who feel strongly about an issue are more likely to respond to a poll than people who couldn’t care less. People who are directly affected by a proposed project or policy might be more inclined to respond to a question about it than people who are not.
These sorts of polls suffer from the same defects as convenience samples conducted in the real world, by people with clipboards at shopping malls. If you sample only people who happen to be shopping and happen to have the time to engage with the survey and happen to be sufficiently motivated to answer the questions, you have not constructed a representative sample of the population.
For a poll to be meaningful, and generalisable to the broader population, it must be based on a randomly-selected, representative sample.
Online polls can work well within companies, or for particular populations such as system administrators, but they cannot, ever, produce a randomly-selected, representative sample of the general population.
Vaccine mandates
I came across an interesting example last week. The campaign website Dear South Africa, which encourages public participation in public policy making, is conducting a poll on the question of mandatory vaccines. Do you, it asks, support mandatory vaccinations or passports – yes, no or not fully?
Under interim results, it shows (or showed, at the time of writing) a little pie chart that indicated just over 80% of the first 60-odd thousand respondents oppose mandatory vaccinations or passports, and less than 17% support them.
Let’s leave aside that mandatory vaccinations and vaccine passports (or certificates, as they are actually called in South Africa) are two entirely different things, and it is quite possible to oppose one and support the other.
The real problem is that such a survey is inherently biased. Respondents are chosen from only that subset of the population that can afford access to the internet, and has the leisure time to browse around and fill out surveys. It ignores, therefore, the majority of the poor, the majority of busy professionals, the majority of the working class, and the majority of older people, to name just four groups that aren’t likely to spend their days on slacktivism.
It also seems likely to me that those who are opposed to vaccines would be highly motivated to respond to such a survey, in the hope of making it seem like their opposition to vaccine mandates and certificates is widely shared.
What we do not have here is a random, representative sample of South Africans.
The effect of sample selection was brought home by Grant Fleming, a Twitter follower of mine in the financial services industry. Using a random selection from a database of 23 million contact numbers, he conducted two small tests via SMS. They both got about a 10% response rate, which is surprisingly large, and both indicated a significant majority (65% and 63%, respectively) in favour of mandatory vaccinations. (The young, interestingly, were more likely to support mandatory vaccination than the old.)
Now I don’t mean to suggest that these are conclusive results, by any means. After all, there is self-selection bias in who responds to an SMS survey, too. There was also no rigorous effort to make the sample as representative as possible across geographic locations, income brackets, and age brackets. Yet it is likely that even this flawed sample is far more representative of the general population than the visitors to the Dear South Africa website.
For a conclusive answer, we really need a professional survey house, with the skills and experience to conduct unbiased, representative population surveys. And even then, you’re stuck with the problem that the majority does not necessarily know what the most advisable policy is.
The point is to illustrate that sample selection can dramatically alter the results of a poll. Whereas Dear South Africa found 80% opposition to vaccine mandates, Fleming found 65% support.
Slacktivists achieve little
Perhaps the larger lesson here is that the vast majority of pie charts on the internet are wrong, and contributing to the making of those pie charts is, usually, a poor way to affect public policy.
To assess public opinion is not a trivial undertaking. The internet may have made it easier to reach a certain well-heeled, leisured, and engaged subset of the population, but it can rarely reach a representative sample of a target population.
This is why expert survey companies charge big money to conduct truly representative surveys, from which one can with some confidence extrapolate to the entire population.
Organisations that are truly committed to changing public policy, as I imagine Dear South Africa is, might consider building up their surveying skills, or commissioning such surveys from experts.
A quick-and-dirty online poll can only ever provide quick but dirty data.
The views of the writer are not necessarily the views of the Daily Friend or the IRR
If you like what you have just read, support the Daily Friend