Home > Aunt Pythia > Aunt Pythia’s advice

Aunt Pythia’s advice

November 10, 2012

I’d like to preface Aunt Pythia’s inaugural advice column by thanking everyone who has sent me their questions. I can’t get to everything but I’ll do my best to tackle a few a week. If you have a question you’d like to submit, please do so below.

Dear Aunt Pythia,

My friend just started an advice column just now. She says she only wants “real” questions. But the membrane between truth and falsity is, as we all know, much more porous and permeable than this reductive boolean schema. What should I do?

Mergatroid

Dear Mergatroid,

Thanks for the question. Aunt Pythia’s answers are her attempts to be universal and useful whilst staying lighthearted and encouraging, as well as to answer the question, as she sees it, in a judgmental and over-reaching way, so yours is a fair concern.

If you don’t think she’s understood the ambiguity of a given question, please do write back and comment. If, however, you think advice columns are a waste of time altogether in terms of information gain, then my advice is to try to enjoy them for their entertainment value.

Aunt Pythia

Aunt Pythia,

I have a friend who always shows up to dinner parties empty-handed. What should I do?

Mergatroid

Mergatroid,

I’m glad you asked a real question too. The answer lies with you. Why are you having dinner parties and consistently inviting someone you aren’t comfortable calling up fifteen minutes beforehand screaming about not having enough parmesan cheese and to grab some on the way?

The only reason I can think of is that you’re trying to impress them. If so, then either they’ve been impressed by now or not. Stop inviting people over who you can’t demand parmesan from, it’s a simple but satisfying litmus test.

I hope that helps,

Aunt Pythia

Aunt Pythia,

Is a protracted discussion of “Reaganomics” the new pick-up path for meeting babes?

Tactile in Texas

T.i.T,

No idea, try me.

A.P.

Aunt Pythia,

A big fan of your insightful blog, I am interested in data analysis. Seemingly, marketers I have recently met with tend to misunderstand that they can find or identify causation just by utilizing quantitative methods, even if statistical software will never tell us the estimation results are causal. I’m using causation here is in the sense of potential outcomes framework.

Without knowing the idea of counterfactual, marketers could make a mistake when they calculate marketing ROI, for instance. I am wondering why people teaching Business Statistics 101 do not emphasize that we need to justify causality, for example, by employing randomization. Do you have similar impressions or experiences, auntie?

Somewhat Lonely in Asia

Dear SLiA,

I hear you. I talked about this just a couple days ago in my blog post about Rachel’s Data Science class when David Madigan guest lectured, and it’s of course a huge methodological and ethical problem when we are talking about drugs.

In industry, people make this mistake all the time, say when they start a new campaign, ROI goes up, and they assume it’s because of the new campaign but actually it’s just a seasonal effect.

The first thing to realize is that these are probably not life-or-death mistakes, except if you count the death of startups as an actual death (if you do, stop doing it). The second is that eventually someone smart figures out how to account for seasonality, and that smart person gets to keep their job because of that insight and others like it, which is a happy story for nerds everywhere.

The third and final point is that there’s no fucking way to prove causality in these cases most of the time, so it’s moot. Even if you set up an A/B test it’s often impossible to keep the experiment clean and to make definitive inferences, what with people clearing their cookies and such.

I hope that helps,

Cathy

Aunt Pythia,

What are the chances (mathematically speaking) that our electorial process picks the “best” person for the job?  How could it be improved?

Olympian Heights

Dear OH,

Great question! And moreover it’s a great example of how, to answer a question, you have to pick a distribution first. In other words, if you think the elections are going to be not at all close, then the electoral process does a fine job. It’s only when the votes are pretty close that it makes a difference.

But having said that, the votes are almost always close on a national scale! That’s because the data collectors and pollsters do their damndest to figure out where people are in terms of voting, and the parties are constantly changing their platforms and tones to accommodate more people. So by dint of that machine, the political feedback loop, we can almost always expect a close election, and therefore we can almost always expect to worry about the electoral college versus popular vote.

Note one perverse consequence of our two-party system is that, if both sides are weak on an issue (to pull one out of a hat I’ll say financial reform), then the people who care will probably not vote at all, and so as long as they are equally weak on that issue, they can ignore it altogether.

AP

Dear Aunt Pythia,

Would you believe your dad is doing dishes when I teach now?

Mom

Dear Mom,

If by “your dad” you mean my dad, then no.

AP

Hey AP,

I have a close friend who has regularly touted his support for Obama, including on Facebook, but I found out that he has donated almost $2000 to the Romney campaign. His political donations are a matter of public record, but I had to actually look that up online. If I don’t say anything I feel our relationship won’t be the same. Do I call him on this? What would you do?

Rom-conned in NY

Dear Rom-conned,

Since the elections are safely over, right now I’d just call this guy a serious loser.

But before the election, I’d have asked you why you suspected your friend in the first place. There must have been something about him that seemed fishy or otherwise two-faced; either that or you check on all your friends’ political donation situations, which is creepy.

My advice is to bring it up with him in a direct but non-confrontational way. Something like, you ask him if he’s ever donated to a politician. If he looks you in the eye and says no, or even worse lies and says he donated to the Obama campaign, then you have your answer.

On the other hand, he may fess up and explain why he donated to Romney – maybe pressure from his parents? or work? I’m not saying it will be a good excuse but you might at least understand it more.

I hope that helps,

Aunt Pythia

Yo Auntie,

Caddyshack or Animal House?

UpTheArsenal

Dear UTA,

Duh, Animal House. Why do you think I had the picture I did on my zit post?

Auntie

Again, I didn’t get to all the questions, but I need to save some for next week just in case nobody ever asks me another question. In the meantime, please submit yours! I seriously love doing this!

Categories: Aunt Pythia
  1. November 10, 2012 at 5:53 pm | #1

    Dear Aunt Pythia,

    Your dad is doing dishes?!

    How do you explain this sort of statistically anomalous behavior? Has chaos theory just been proven?

    Beffudled in Aisle 4,
    Backup Ketchup

  2. FogOfWar
    November 10, 2012 at 8:11 pm | #2

    To RomConned–just make sure to doublecheck that it isn’t someone with the same name. Those online systems are only as good as the data people put in when they fill out the forms…

Comments are closed.
Follow

Get every new post delivered to your Inbox.

Join 892 other followers

%d bloggers like this: