Ears, camera angles, and the "null hypothesis"

Today I want to explore how camera angles and the like can make similar subjects look different.

But first, credit where it is due: around June of 2010 (I think), Palin's Peyton Place posted this graphic at Flickr to make the argument that "Ruffles" was simply Trig with his ears fixed:

The note in the middle of the graphic refers to surgery being done; the Earwell and Ear Buddies procedures that I mentioned in an earlier post do not require surgery, but other than the use of the word "surgery," I totally agree with the argument the author is making. 

To see this graphic and a few related ones, go to: http://alturl.com/nnzn6. In one related graphic, the author argues (despite rejecting an earlier swap) that a baby swap must have happened at the Republican National Convention, based on differences in Trig's ears in two photos. I think she (or he) is wrong on that. Here is the graphic:

First note that the photos at the top of the graphic were taken at very different angles: the first one was shot very much at an angle below Trig's ear, while the second one is more straight on. Trig's ears, in fact, do bent outward at the top, as the second photo shows – but you cannot see the bend in the first photo because of the extreme angle. Moreover, Trig's sweater is pushing up the bottom of his ear in the second photo, making earlobe comparisons problematic. But my heavens, look at the rectangular shape in the middle! Even though the ears were shot at different angles, that very distinctive shape tells me this is the same child (or twins – but that way madness lies.)

The apparent difference in the ears in those photos can be accounted for by the different angles at which the photos were taken.

Next, here is the 3-panel illustration I posted earlier to show the progression of Trig's ears:

Note that the second and third pictures show an irregular shape mid-ear that looks (in the middle picture, at least) somewhat like a heart. The key to resolving the difference between that mid-ear shape in the second and third pictures is to understand that that the photos were taken from much different angles. Here are the two photos again, starting with the Bailey picture:

Note that this photo was shot more from the front than the side, which would cause that space in the middle to appear shortened. Now here is the VP debate shot:

This shot is straight onto the ear. So to get a sense of how the two ears are truly equivalent, you have to "rotate" in your mind's eye the middle photo. Therefore, I see no reason to question that the second and third photos are the same baby. And the stunning similarities at the tops of the first and second pictures convince me those are the same – at least they are stunning when you "rotate" the middle one mentally. The first picture is also fairly straight on, as you can see:

As I hope the foregoing has made clear, comparisons ears or other body parts can be made difficult by different camera angles. But that's not all. Different lighting can change apparent skin color, for example. Lenses with different focal lengths can seem to "lengthen" or "shorten" a given subject. For example, is the baby at Mat-Su Medical Center too "big" to be the baby in the picture just above? Answering that would be very difficult without having a common reference point, which is lacking, plus the lack of technical information concerning the photos makes it hard to even guess.

Let me also bring up the idea of the "null hypothesis." Whenever you form a hypothesis that something happened or routinely happens, the null hypothesis is that it did not happen. For example, a hypothesis is that smoking causes cancer. The null hypothesis is that it does not. We can "reject the null," as social scientists say, because of the mountain of evidence over the years that smoking does in fact cause cancer.

In the case of the baby-swap hypothesis, the null is that Palin did not present different babies as Trig in 2008. To reject the null, you would need very compelling evidence. Typically in the social sciences, you need to be at least 95 percent confident before you can reject the null and assume your hypothesis meets the standard threshold for "significance" – termed "statistical significance" when numbers are involve. We don't have precise probabilities assigned to the events surrounding the baby-swap hypothesis, but I think we all have a gut-level sense of what it means to be 95 confident that something is true: that's pretty darned confident! 

Now I know some people will not accept my judgment when I say I am 99 percent confident that the three ears in the composite above come from the same child. That's fine. If you don't, then my question to you is this: are you at least 95 confident the ears come from  at least two different children? If your answer is not a resounding yes, then you probably should put the baby-swap hypothesis on the back burner. If you can't make a great case for that hypothesis, you may start to appear like a tin-foil hat wearer to people who are sophisticated about research. And in light of the counter-arguments and evidence I have present, I don't think an unbiased person could be 95 percent confident the pictures come from different babies. (Bias, by the way, can come simply from a desire not to change your mind once you have staked out a position.)