Just this week a student in one of my online classes answered a test question with a real-world example of trying to manipulate people using numbers and graphs, which they had to debunk. The example they chose (from the course text, Levitin’s Field Guide to Lies) is solid gold; and they directed me to an actual video clip of the lie-in-use, “in the field” as it were. After looking at it more in that light, I realized there was even more going on here than Levitin was pointing out. It was bonkers in so many ways I had to do a complete write up of it here, so you can follow along and be well primed to spot these kinds of deceptive tactics yourself whenever they are used.

Because I think people just gullibly assume stats and graphs presented to them are at least somewhat honest. Maybe they are making a fallacious argument from the graph. But they aren’t even imagining they would be dishonest not just once or twice or a little bit, but outrageously and in nearly every aspect of their presentation. We don’t expect people to be so brazen in their attempt to lie and deceive. The liars in today’s example are anti-choicers. Which should not surprise. Lying for Jesus is probably the most common Christian practice in public discourse, outfrequenting even prayer (and certainly giving alms). It has been the number one Christian industry from even their earliest days. I’ve documented this too many times to count by now. But let’s tour one more example, for funzies.

That Time When Jason Chaffetz Lied to the American People

Back in 2015, U.S. Representative Jason Chaffetz was sitting a hearing on how evil Planned Parenthood was, interviewing their president, Cecile Richards, and making (as usual) a bunch of disingenuous and dishonest points, and trying to speak over his own witness and testify for her instead (and a lot else is going on there we could poke frustrated fun at, like that Chaffetz didn’t know—or pretended not to know—that a mammogram and a breast exam are not the same thing, or that Pap smears are not breast exams). But the centerpiece of this fiasco was a “graph” he presented Richards with, which caught her off guard, so she was too flustered to tear him to shreds over it. Her lawyer brought attention to the fact that the graph is Christian propaganda (it says right on it “SOURCE: Americans United for Life”; yes, the infamous AUL) and not from a real source anyone should trust, which is a fair enough point. But that’s not even the least of what went wrong here: the graph is dishonest in so many different ways it is actually a model teaching tool for critical thinking and how and why numeracy is crucial to not being a dupe.

Richards was numerate enough to have torn it down, but she was blindsided and didn’t have the time—which is the first dishonest tactic on display here: not even giving someone enough time to critically examine the information being presented, and instead just railroading them with what it “proves” without first having passed even rudimentary tests of whether it proves anything at all. But to get you centered, you should watch the original two-minute clip (archived on C-SPAN) showing the graph and the context of its presentation and use. You can then jump to the punchline if you want, and read this apt takedown by Columbia University statistics professor Andrew Gelman (shout out to my alma). But either way I’ll walk you through the essentials and expand.

This is the graph Chaffetz hurled unexpectedly at Richards:

If you want to play a game, before reading anything below, see how many distinct lies you can catch in this graph. I will give you one assist: the numbers are not wrong; every number on this graph is correct. And the source line doesn’t count, because though damning, it is nevertheless true. Now, how many lies is this graph using to deceive you? Write down the number you counted up.

-:-

Here we go.

  • First: It only shows the end-data points for 2006 and 2013, yet draws lines as if through data points in every intervening year. But those data points don’t exist. The graph’s shape is therefore fake.
  • Second: It “communicates” that between 2006 and 2013 Planned Parenthood gradually started providing more abortions than “cancer screening & prevention services,” that its purpose thus completely flipped upside down from screening to aborting. But pay closer attention. The y-axis has been invisibly flipped instead: on the left, the larger number is at the top; but on the right, the larger number is at the bottom. Yet the graph signals the top-right-side number is larger, when it’s not. In fact the lines should never cross: cancer screenings should remain several times more numerous than abortions all the way across the graph (as that is what its own printed numbers say).
  • Third: The graph is anamorphic. Cancer screenings halved while abortions increased by barely twelve percent; yet the image shows a straight swap, as if abortions doubled while screenings halved, and as if there were almost as many abortions in 2013 as screenings in 2006. Neither is the case.

Now, you might have missed the next three, because they require external knowledge, but…

  • Fourth: The graph conflates correlation with causation. It was built and contextualized to signal that Planned Parenthood has gradually shifted policy toward abortions over screenings; but what is left out of this graph is the actual cause of the halved screening frequency: that after 2006, national standards changed regarding the recommended frequency of “cancer screening,” due to research showing that, for example, screening for cervical cancer annually (likewise breast cancer) causes more harm from false positives and so biannual or triannual screening came to be recommended instead. So in a very basic sense, the graph lies about the cause of the shifting frequency of those specific procedures. (Which is the lynchpin of the entire deception here; more on that in a moment.)
  • Fifth: It does this twice. Not only does it falsely signal this halving of tests was a Planned Parenthood mission to abandon testing, but it falsely signals the 12% increase in abortions was also a Planned Parenthood mission to increase the national abortion rate. In fact in the period depicted the number of abortions in America declined, indeed by over 20%. So why were more abortions being performed by Planned Parenthood? Because anti-choice efforts were closing clinics and limiting access in exactly those years. The result was that Planned Parenthood increasingly became the only place women could seek abortion care. In other words, anti-choice activism caused this number to go up, not Planned Parenthood’s conniving.
  • Sixth: More particularly, this graph lies (through deceptive wording) about what is being counted. Abortions register in our mind as single-patient events, so the graph misleads us psychologically to extend that same understanding to the admittedly vague category of “cancer screening & prevention services,” as if we are counting the same thing. We are not. The reason cancer screenings halved after the new recommendations is because the new recommendations called for screening half as often. In other words, their “cancer screening & prevention services” number here is counting tests, not patients. In fact the same number of patients (more or less) were receiving these cancer screenings. They were just gradually being tested every other year instead of every year. So the number of tests halved, but not the number of people being tested. This change in what is being counted thus creates an incongruity with a counting of abortions. These are not commensurate things to measure at all. If the graph had counted patients, both lines would be nearly parallel, and screened patients would show as always roughly ten times more numerous than Planned Parenthood abortion patients across the entire period, directly refuting the entire argument intended by the graph.
  • Seventh: Putting those last three points together, the biggest lie being told here is by selection-and-omission: the choice itself to only graph abortions against cancer screening tests is a lie. Because Planned Parenthood provided near five million services in each of those years, so you should ask, why aren’t the other three million services on this graph? Or, if you didn’t know this external information and were just looking at the graph, you should still ask, “Are cancer screenings the only other non-abortion services Planned Parenthood provides? Why are no other services being shown?” We’ll get to that. But the short answer is: because then the entire graph’s argument would be refuted. There was no substantial change in the number of non-abortion services they provided between 2006 and 2013.

Maybe you caught more lies than those seven (if so, please detail any additionals in the comments). But seven is more than enough. The authors, endorsers, and users of this graph are liars. Not just mistaken. Not just goofs. Liars. They know they are lying. They know the graph is contrary to reality. The number of abortions never exceeded the number of cancer tests—it never even came close, in any year. The number of patients likewise remained the same—always ten times more women getting screened than seeking abortions. They know the national number of abortions went down in this period, not up; and the increased recourse to Planned Parenthood for abortions services was caused by antiabortion activism, not “Planned Parenthood.” And they know the actual non-abortion services they provided didn’t change meaningfully, and that there were ten times more non-abortion patients served in both years. All of this information is being kept from you with a manipulation of data and imagery.

We thus get to see what a real chart would look like thanks to the curtain being yanked away by Dr. Gelman:

Here we have a correct and real y-axis, and resulting visual honesty. Gelman is using their data points, so he still drew straight lines across all years, but he admits he had to with the information given—a better graph would draw lines through real data points for every year, so you could see the actual rate and shape of change in both numbers. Gelman also adds some otherwise-hidden causal information (for screenings, he notes the cause was changes in national standards, not in Planned Parenthood’s mission). He could have added more—like why there was a slight rise in abortions there when nationally abortion numbers went down, or that the “screenings” line is not commensurate with the “abortions” line because it is measuring tests and not patients, or that lines are missing for all the other services PlanP provides. Indeed, the best graph would simply count patients—and then both lines would be nearly flat, and the screening line about ten times above the abortion line all the way through (and above, if it included “all other patients served”).

Lessons

Basic principles of numeracy are essential to critical thinking. It’s not enough to just question a source. You need to be able to question even the source’s argument. The graph Chaffetz lied to Richards’ face with presents a visual argument. The argument’s conclusion is that Planned Parenthood is nefariously abandoning its meritorious work (like cancer screening) to deliberately scale up abortions instead. It reaches this conclusion through various visually-presented logical steps.

For example, Gelman quotes Zachary Roth pointing one of these out:

[The image] makes it seem like in 2006, Planned Parenthood performed far more cancer screening and prevention services than abortions, but that by around 2010 it performed an equal number of both, and by 2013 it performed far more abortion services than anti-cancer services.

The issue is important because as part of their effort to defund Planned Parenthood, Republicans have portrayed it as primarily an abortion provider, while the group’s defenders have said it mostly performs other women’s health services, like cancer screenings.

In other words, the argument is:

  1. Planned Parenthood claims it performs more cancer screenings than abortions, but in fact it now performs more abortions than cancer screenings, and so is really just an abortion provider.
  2. If Planned Parenthood is really just an abortion provider, it should not receive any public support, and shutting it down will do no harm to honest patients.
  3. Therefore, Planned Parenthood should not receive any public support, and shutting it down will do no harm to honest patients.

Obviously one can easily refute the major premise as well (that shuttering abortion care is good and not bad), but that’s not the premise this particular graph is arguing for. It’s arguing visually for the minor premise: that Planned Parenthood is really just an abortion provider (and is lying when it says it’s not). One gets this from its headline and its visual rhetoric, but also its context (as you can tell from the video, for example; and Chaffetz isn’t misusing it there: it is used the same way in the original document that produced it, as we’ll find in a moment). The premise is “defended” with an image of two arrows crossing each other, claiming what Roth points out. The claim is false, as Roth also noted. And ironically, it can be proved false by anyone who looks at the actual numbers on the graph (Richards struggled to see those in the hearing; you can see her squinting and not succeeding). The visual contradicts the numbers. The visual is lying. These liars presumably want to claim plausible deniability by insisting they had the right numbers, so that if you call them on the lie, they can try to pivot away from that into some other talking point. But a lie is a lie. And lying is lying.

Other lessons may be familiar, like the essential canard that correlation does not entail causation. Showing a correlation graphically to argue for a causal relationship is a logical non sequitur. You need more information than that. Conversely, omitting crucial yet positively known information that clearly signals the actual cause (and that cause isn’t what you are claiming it is) is lying. Why did abortions slightly increase at PlanP? Does that match or buck a national trend? And if it bucks it, why is that? “Is” it because PlanP is an evil abortion-for-profit enterprise? Or is it because antiabortionists destroyed its peer providers forcing it to take over more patient load? Why did screenings halve at PlanP? Does that match or buck a national trend? And if it matches it, why is that? “Is” it because providers don’t take their other services seriously and just use them to figleaf their real operation as abortion factories? Or is it because standards changed so that in in fact test frequency halved but not tested patients? Cause-correlation fallacies are ubiquitous. Don’t let that slide.

Other lessons of course relate to all the ways visuals can deceive, like invisibly flipping a y-axis, distorting proportions, and even using “red” for the “bad” abortion and “pink” for the “nice and fluffy” cancer screenings. Definitely be on your guard against that. Don’t be a dupe. But the most important lesson—because it may be one you think of less and are thus less on your guard against (as possibly Richards wasn’t)—is this: always ask when arithmetical arguments like this are being attempted at you, “What are you actually counting?” Of course, also, who counted, and how were they able to count this, and how reliable is that way of counting, or finding the things to count? And, ideally even, has anyone else independently counted this and come up with the same or similar result? Or, worse, come up with a concerningly different result? But the key factor here is how these liars used a change in test frequency to dupe you into thinking they were counting a decline in patients served by these tests.

After all, what does “cancer screening & prevention services” mean exactly? Services—how? Patients, or medical visits? Because those aren’t the same thing (the same patient can spread care over multiple visits, or get a lot done in one visit). Procedures performed, or procedures coded? Because those aren’t the same thing (the same procedure can be cross-coded multiple times or grouped under a single service code). For example, if they had said “pap smear tests performed,” that would clue you in to what is actually being counted, and that would make it possible for you to investigate questions like, “Did pap smear tests decline in frequency everywhere, or only at Planned Parenthood? And in either case, why?” Of course you should already have done this for the abortion data. Immediately you should have been skeptical of the narrative that abortions rose 12% between 2006 and 2013, and thus have checked if they actually did—at which you would find they didn’t, but went down 20%, requiring inquiry into why Planned Parenthood was picking up more of the load, which would explode the entire argument being foisted on you.

But I’ll bet you were hoodwinked entirely by the conflation of cancer tests and tested patients, because no information was provided that would alert you to question that. Which is the most important lesson here: when you get duped (and you will get duped), it will be by tactics like this, where they don’t clue you in at all to what information is being concealed from you, so you won’t be able to defend yourself against any resulting disinformation. The only way to arm yourself against this attack is to always remind yourself the attack exists as a strategy, and you should be alert to how you can catch it. And this is way more important a point than I am even making out here: because almost all apologetics, for any false claim or view, consists of omitting information that, when reintroduced, refutes it—so you must always go look for what information is being left out. In my case, once my student drew my attention to the video clip (which I hadn’t looked at before), I immediately ran some keyphrase searches to find out if anyone else had already debunked it and thus already checked this stuff—hence I found Gelman, and the rest of the thread from there. I then could research the history of changing test standards, the history of Planned Parenthood’s isolation as increasingly, in many places, the only available abortion provider left, and everything else.

Ultimately, after all that, I still wanted to know where even these liars’ numbers came from. So then I tracked down the original committee presentation, where this “slide” is on page 6. Which led to a dead URL I could recover on archive.org, which led to the original Christian lobbyist package, which originated the graph, but still does not cite a source for its numbers. It mentions relying on PPFA’s “annual reports,” though not specifically for these numbers. But I figured, why not check there? So I hunted down their online report archive, but it only goes back a few years. A little more digging and I found the report for 2006, and indeed the numbers are there that we find on the dishonest graph: page 6 says over a million Pap tests, almost a million breast exams, over forty thousand colposcopies, and a few thousand tissue removal procedures, for a total of “2,007,371” instances of “cancer screening and prevention.”

Then, of course, I found the report for 2013. Indeed it shows that number fell to 935,573 (page 18), by adding up 378k Pap tests and 487k breast exams (about half of the numbers from 2006), and now tens of thousands of HPV vaccinations (because that had just been introduced in 2006), over thirty thousand colposcopies, and a few thousand tissue removals. They also, of course, add up millions of other services, as I mentioned before: pregnancy tests, prenatal services, STD testing, contraception, and a smattering of other things like adoption counseling. Which of course showed no major change in quantity over time (and in fact still show roughly the same numbers today). There is a reason our liars chose only to count cancer screening tests. Indeed, the number of those that Planned Parenthood performs has since gone down even more–though for a different reason now: the Affordable Care Act. Cancer screenings are now more widely available to the population, because uninsured people are at an all time low, thus abrogating the need of independent clinics like Planned Parenthood for those specific services, because they’re now routine components of most people’s Health Management plan. So we can note that after 2013, the uninsured rate dropped by about 30% due to the ACA, which roughly matches the drop seen in cancer screenings at PlanP.

So the takeaways here are: never simply trust a conservative Christian source (lying and misrepresentation are their normal mode of discourse); visual arguments are still arguments (and as such can still represent fallacies or lies); numeracy is crucial to critical thought (and to just being a savvy person generally, rather than a dupe); and when presented with any numbers argument, we must understand how to ask (and investigate) what is being counted, who is counting it, how are they counting it, and how does that affect what we make of the count. And, of course, we must never fall for causation-correlation fallacies, but always ask, “Why does the data present this way? What bigger phenomena might explain this? What other data or sources could I check to get a sense of whether this is normal or weird or what is causing it?”

But the most important lesson of all is: never trust proven liars again. Once they have demonstrated their utter disrespect for the truth (and thus for you), you should condemn them as useless bloviators whose opinions and assertions no longer matter to any useful discourse. The Americans United for Life are simply liars. And liars deserve no place at the table. Much less at a congressional committee hearing.

§

To comment use the Add Comment field at bottom, or click the Reply box next to (or the nearest one above) any comment. See Comments & Moderation Policy for standards and expectations.

Discover more from Richard Carrier Blogs

Subscribe now to keep reading and get access to the full archive.

Continue reading