the eight plots, revised

(Standard disclaimer about how I understand that this is not of interest to the usual regular or irregular reader of this blog.) Kieran and the mysterious "Steve M" caused me to rethink the eight plots I posted the other day. Kieran was right that what I said about the graphs wasn't consistent with the graphs themselves; instead of changing the graphs, though I've thought more precisely what I said about them. Steve M thought I should think more precisely in terms of the potential outcomes framework, which is what I actually meant to be doing, only I forgot about the "precisely" part. What I think I want to say is characterized by this table:



I is the blue line in the graph, I′ is the red line. I is the “natural/default/control” entity (e.g., genotype), and I′ is the “intervened/treated” entity. E is the “natural” environment, and E′ is the “intervened” environment. y is the outcome, so y(I|E) would be the outcome for the "natural" entity in the "natural" environment. "---" in the table means that the answer can vary in ways which would imply that the graph would be drawn differently but I’m taking these variations for purposes here as being minor variations of the same plot.

Anyway, the plots themselves may be characterized like this:

(1) the genetic cause has an effect that is the same across the environments and the environmental cause has no effect
(2) the genetic cause has no effect and the environmental cause has an effect that is the same across the genotypes
(3) the genetic cause has an effect that is the same across environments and the environmental cause has an effect that is the same across genotypes
(4) the genetic cause only has an effect if the environmental cause is present
(5) the genetic cause only has an effect if the environmental cause is absent
(6) the genetic cause has an effect that is reduced by the environmental cause
(7) the genetic cause has an effect that is increased by the environmental cause
(8) the genetic cause has an effect that is reversed by the environmental cause

And, because the original point was not so much the eight plots per se as the fact that the figure in stories about genes v. environment, person v. situation, and agency v. structure all as matters of "internal" versus "external" causes, this would be an example of the eight plots using the example of people (given that drunkenness isn't actually a situation, one could substitute "at parties").

(1) A is more outgoing than B, the same whether they are sober or drunk
(2) A isn’t more outgoing than B, but they’re both more outgoing when drunk
(3) A is more outgoing than B. They’re both more outgoing when drunk, to the same degree
(4) A isn’t usually more outgoing than B, but A is way more outgoing when you get them both drunk
(5) A is more outgoing than B, but B gets so much more outgoing when they both get drunk that then they’re about the same
(6) A is more outgoing than B. They’re both more outgoing when they’re drunk, B more than A so they’re more similar
(7) A is more outgoing than B. They’re both more outgoing when they’re drunk, A more than B so they’re even more different
(8) A is more outgoing than B. Unless you get them drunk, then B is more outgoing than A

I think I can name pairs of people who would satisfy all eight of these stories, actually.

after the tone

So a question I ask in a variety of contexts, including sometimes on this weblog, is "Coincidence, or causality?" This question has a special self-absorbed counterpart: "Quirk, or incisive exemplar of a fundamental flaw in Jeremy's character?" To wit: I've let the voicemails on my cel phone pile up again. I've got like 17. I didn't check my messages one day, left my cel at home the next, and next thing I know there were 8 messages for me. Who wants to deal with 8 messages? But then I keep putting off listening to them, and new people keep calling (or old people keep calling again), which just makes me not want to deal with it even more*, which just makes it pile up all the more. Am I alone in this? If I'm not alone, is it because we share the same fundamental character flaw?

Sorry if I haven't returned your call. Don't take it personally. You know that if it's anything important you should e-mail me, right?

(Okay, this is crazy, even by my standards. I am not going to post this until I have listened to all my messages... There, done. No family members dead for days, great. Later messages did include--verbatim, from separate people--"Darling, where are you? Why don't you ever answer your phone anymore?", "Jeremy, every part of your overall communicative apparatus has failed me at this point..." and, eek, "Jeremy, answer your [expletive] phone..." Ah, well, the boy detective is back on top of The Case of The Vibrating Treo.)

BTW, I've disconnected my landline at home. I decided that ~$80/month was altogether too much to pay when a high-speed home Internet connection (the main reason I had it in the first place) might perhaps be obtained through certain uncostly means.

* This might be different if I could listen to my e-mails from the most recent message backwards, like with e-mail. Why doesn't voicemail offer this feature?

don't let's sleep

Almost 6AM, and here I am: sort of reading an AJS article (Bearman, Moody, and Stovel's "Chains of Affection")*, sort of reading through Wikipedia entries on neoclassical economics, and sort of watching the YouTube video for "Don't Let's Start" over and over again.

It's hard not to watch "Don't Let's Start" and not feel a bit of rousing glee, perhaps even as viewings careen into the double digits, but still this deliberate flipping around of days and nights seems wrong. These were not unusual hours for me in graduate school, but, let's face it, graduate school is way back in the rear view mirror at this point. Ah, well, the luxuries of twisted self-experimentation that one can do when one is unmarried, childless, and doing a post-tenure postdoc.

* "Chains of Affection" (a deservedly award-winning article) is based on a saturated sample based in a "midsized midwestern town." The pseudonym for this town is "Jefferson City." This is what investigators from the coasts do to those of us in the Midwest. You wouldn't see researchers doing a study of a "midsized northeastern town" and use the pseudonym "Albany" or of a "midsized west-coast city" and use the pseudonym "Sacramento." See, there is this state in the Midwest, called "Missouri." The capital of Missouri is "Jefferson City."