Mr. Know - It - All

Installment 2

Dear Mr. Know-it-all,
I am a newlywed husband and my wife is driving me crazy with this stupid game where she asks me to choose which shoes, or some other article of clothing I like better, then wears the other pair. I can't believe she is playing mind games already. What can I do to make her wear what I pick?
Hrut Herjolfsson, Iceland


hen offered this position, Mr. Know-it-all was assured that the questions would be technically hard, thought provoking questions like Q. "why is the sky blue?"; now I find that most people don't care about that stuff and instead want me to help their marriages (A. because sunsets are red - couldn't leave my faithful readers just hanging). Despite my first inclination to just make up technically hard, thought provoking questions, I feel that I have to help someone in need, and brother, you're in need. So with my apologies to Dr. Laura, I'll explain the simplicities of men and women.
You've made a common assumption that the opposite sex thinks just like you. There's a reason we say "the opposite sex", and not "the just like us sex that happens to have that other set of sexual organs". You think that when your wife asks your opinion between two choices, she is at a loss and needs your help, since a man only asks for help when he's at his wits end and knows he can't do it himself. So to a man, this question is an invokation of heirarchy, and you've been made the boss on this question. But to a woman, this question is reaching out for connectedness, a discussion of equals, and most importantly, a confirmation of a decision she has already made. If she really needed help, she'd ask another woman who's judgement she trusts, not you, who can't be trusted to leave the house without at least a cursory fashion check (she: you're wearing that out? In public?).
The way to approach this question is the way another woman would, and that is to point out one positive for each choice, and then say "But what do you think?". This reconnects your wife to the decision and allows her to then signal (listen to her!) which choice she likes. Don't be like a man, and render your verdict Solomon like (he: Red, and that's final) and then get all huffy when your authority is flouted (she: Red? Are you blind? Red doesn't go with this outfit at all! He: What are you talking about? If you didn't like it, why did you ask me in the first place?). Whichever one she seems to like, pick. Don't argue, that defeats the whole purpose of this exercise, which is to be included, to get closer through agreement, to bond. If you didn't get it right at first, don't argue, but be persuaded (he: What was I thinking? Taupe would be perfect!. You're so beautiful, I just got confused there for a moment). And don't refuse to participate, that's felt as a rejection (he: Don't ask me, you never listen anyway! She: Don't you care about me?).
Einstein liked thought experiments, so here's one to illustrate. Imagine you take your wife out fishing. You keep telling her to be quiet, or she'll scare all the fish away (hah, like the fish care). If she understands men, she'll be quiet, because she knows you just want share the experience together, and not intrude on and disrupt the experience with extraneous thoughts. Men bond by the sharing of the experience itself, and women by sharing their reaction to the experience and connecting it with others. So when you simply render a decision about "which one do you like", it's like going fishing with a buddy at his invitation and then you talking about what the neighbors are doing the whole time.

NOTE: Any place I've claimed a gender difference, these are trends and tendencies, not absolutes, and I make no claim, either implied or explicit, about which is better or more important.

Dear Mr. Know-it-all,
Which would you rather have: the ability to smell things you see on TV, or a lifetime supply of lettuce?
Wondering in Queen Maud Land


 can understand your preoccupation with food and eating given that you live in the frozen wastelands of Antarctica. I was loath to answer this question at first, but given that it actually did come from a reader, and given that oddly enough we are often confronted with conundrums such as this one in our life, I decided to go ahead and print the answer. So, on to my answer, and more importantly, the reasoning behind it, which oddly enough, will come first.
How to compare such different outcomes is our first task. For the lettuce, we can make some quick estimates. In one sense, I already have a lifetime supply of lettuce since I already consume as much as I like and probably always will. So, if you take the lettuce, really the only effect is to put money in your pocket. If you figure that you normally eat a head of lettuce a week, and figure that the average price of lettuce over your lifetime at 2 dollars a head, this would work out to roughly a hundred dollars a year, for my entire life. Just to put this into perspective, it is less than what I pay for internet service. Let's say that you could eat as much lettuce as you could stand, and therefore could manage to choke down a head of lettuce a day, you are then talking in excess of 700 dollars a year. So we'll take this as our maximum dollar figure on the lifetime supply of lettuce.
Now let's turn our attention to be able to smell whatever is on the TV. This is something I can't do right now, and it doesn't look like any technological advance will allow me to do this during my lifetime. What dollar figure can you put on this ability? Frankly, in my view it is priceless. If I could pay 700 dollars a year (which isn't that much larger than my cable bill) to be able to smell anything on TV, I would. This would be an extension of the quality of my life (assuming I got to choose the channel, that is). When you compare this with simple money, which is what lettuce represents to me, the choice is both obvious and simple. I'll take the smell, thank you very much.
Often we are confronted with the choice of simple money or something else, such as an experience or a skill, and let me say you should choose the something else most every time. Frankly, that's what money is for. Unless, of course, you're stuck in a frozen wasteland without food in which case it would be much better to have the lettuce and live, than to smell the TV and die happy.

Dear Mr. Know-it-all,
I read the newpapers alot and they are always printing some poll result. What I want to know is how can a poll of 700 people adequately represent a nation of near 300 million? It seems to me you need a lot more than that. Am I right?
April Storms, Dayton Ohio


ou've asked a vital question about an increasingly popular opinion shaper. The short answer is it depends on how accurate you want the results. The long answer has several parts, not all of which you touched on.
First, none of the polling groups claim that their results have no margin of error. Usually you see a number of 3 or 4 percent claimed. That is to say, if a poll says that 75 percent of Americans believe G. Washington really cut down a cherry tree with a margin of error of 4 percent, the real answer is that from 71 to 79 percent believe our first president was a tree killer, with no value within that range being more likely than another. Of course, you never see poll answers reported this way. In fact, quite often you'll see that a later poll with the same margin of error will claim that 78 percent now believe in an ax-wielding Washington, with the claim that his numbers have gone up. Sadly, the second poll does not support such a claim, because the second answer is within the error of the first. In fact, Big George's numbers may have actually dropped from 79 to 74 percent. Because of the accuracy of the polls, we just can't tell. Keep that in mind when you read headlines based on polls.
Second, the quoted margin of error comes from a purely mathematical analysis. That is to say, probability theory tells you what your error is for a random sample of a given size for any distribution. The size of the distribution doesn't matter. So, pollers assume that there is a certain distribution of opinion, and that if they randomly sample the distribution, then they arrive at the mathematical derived error. All well and good, but we are dealing with people. Anyway, there is a reason polls are of small size relative to the number of people they represent. Because mathematics tells us that as we increase our sample size, the decrease in error does not keep up, changing your sample from 100 people to 1000 people does not make you 10 times more accurate. Since changing your sample from 100 to 1000 people probably increases your cost by a factor of 10, you see the drive to limit the sample to the smallest possible number while retaining some accuracy.
The minimum inaccuracy occurs when the sample is perfectly random. As any lack of randomness leads to greater uncertainty, the question becomes just how random is your random survey. Part of the answer to this is the question of whether there is a single distribution out there (Americans), or are there several (men vs. women, blacks vs. white, wealthy vs. poor). For example, if you do a phone survey, do people with phones have a different opinion that those without? Today, that is probably true, but it caused a problem in the thirties when Reader's Digest did a phone survey and predicted that Roosevelt would loose in a presidential election. Because mostly wealthy people had a phone in those days, and they didn't support Roosevelt, the phone poll didn't get a random sample, or to look at it another way, they didn't sample the two distributions (wealthy and not) properly.
The search for randomness leads down many paths. Pollsters have worry if people who answer polls (either in person or over the phone) have a different opinion than those who don't? Or if you ask people at a certain time of day, do those who are available then have a different opinion than those who aren't? There are, of course, the more obvious possilbe distributions. For instance, there was a large racial split over the OJ Simpson verdict. If a poll were taken about that, it would have to insure that both blacks and whites were adequately sampled and then their relative numbers accounted for. So, the pollsters have to stay up late at night worrying about whether any particular question has a difference in opinion based upon race, upon sex, upon geography, urban vs. rural, you get the idea.
Of course, people don't belong to just one possible group. So do you worry that you've adequately sampled the white female rural homosexual poor distribution? That leads me to the conclusion that quite frankly, any well constructed poll has an unknown error, but certainly much larger than what is quoted by the polling organization.
And if you throw in question bias, well, now you have to take into account who did the poll. Question bias is the phenomenon that you can change a poll by not only how you phrase a particular question - "Are you not outraged that the barbaric and vicious practice of killing innocent, intelligent whales is still allowed?" - but by the order in which questions are asked or even by how a pollster introduces himself and the poll. Consequently, any pollster who knows what she is doing can obtain a more favorable poll if so desired.
So taking all of that into account, I think we can safely say that there are lies, damned lies, statistics, and then polls.

Return to the Index for Mr. Know-It-All

Return to The Murphy Nexus ^TM

---

This page last updated 6 May 1998

---

Contents copyright Kevin Murphy 1998. All rights reserved.