Networks are endlessly interesting. They shape the world around us like almost nothing else, yet we spend very little time thinking about them or recognizing the non-obvious ways effects they have on our world. Because we tend to think in bell curves, we get tricked by networks pretty often. For instance, we’ve all had the experience of feeling like everyone is talking about something when it turns out to be pretty self-contained to your little group. Well, some folks at USC have spent some time looking into just how networks create that illusion, which they call the majority illusion:
The majority illusion occurs when the most popular nodes are colored. Because these link to the greatest number of other nodes, they skew the view from the ground, as it were. That’s why this illusion is so closely linked to the friendship paradox.
Interestingly, the majority illusion can even show up when a node isn’t globally popular:
That might seem harmless when it comes to memes on Reddit or videos on YouTube. But it can have more insidious effects too. “Under some conditions, even a minority opinion can appear to be extremely popular locally,” say Lerman and co. That might explain how extreme views can sometimes spread so easily.
As with most things related to how ideas spread throughout networks, the more popular something is seems to be the only reliable indicator to how popular it will become.
This critique of Zimbardo’s famous Stanford Prison Experiment is really fascinating. Basically the author, who writes intro to psychology textbooks, suggests that the experiment was flawed because it urged students to act in the way they thought typical guards and prisoners would act. Here’s an excerpt that captures it pretty well:
In a nutshell, here’s the criticism, somewhat simplified. Twenty-one boys (OK, young men) are asked to play a game of prisoners and guards. It’s 1971. There have recently been many news reports about prison riots and the brutality of guards. So, in this game, what are these young men supposed to do? Are they supposed to sit around talking pleasantly with one another about sports, girlfriends, movies, and such? No, of course not. This is a study of prisoners and guards, so their job clearly is to act like prisoners and guards—or, more accurately, to act out their stereotyped views of what prisoners and guards do. Surely, Professor Zimbardo, who is right there watching them (as the Prison Superintendent) would be disappointed if, instead, they had just sat around chatting pleasantly and having tea. Much research has shown that participants in psychological experiments are highly motivated to do what they believe the researchers want them to do. Any characteristics of an experiment that let research participants guess how the experimenters expect or want them to behave are referred to as demand characteristics. In any valid experiment it is essential to eliminate or at least minimize demand characteristics. In this experiment, the demands were everywhere.
I find stuff like this really interesting. I think most research is flawed in that it asks people questions they aren’t really prepared to answer and in turn forces them to come up with a conclusion. I thought about this a lot when I made Brand Tags and people were asking me to put up logos that no one had seen before so they could get feedback. I would always argue that this was measuring brand perception and if no one knew your brand they would just comment on your logo, which isn’t particularly helpful. Brands, ultimately, are the sum total of all the experiences one has and no one ever experiences one by just seeing a logo on a blank page. They hear about it, see it on a shelf next to another product, or any number of other contextual clues. Obviously this situation is pretty different, but I think it’s part of a very broad mistake research makes in not controlling for context (or lack thereof).
I’ve been thinking about big data lately. Mostly I’ve been trying to articulate why it’s a big deal, which I know, but isn’t often put succinctly. Recently I had a thought that the reason it’s such a big deal is because it means we can move away from using samples to infer to using actuals to understand. That seems really obvious, but it wasn’t a connection I had made before (though I sort of think it was obvious to everyone else).
Anyway, on the big data tip, I found this Wired piece on “long data” pretty interesting (even though I thought I was going to hate it based on the title). The gist:
By “long” data, I mean datasets that have massive historical sweep — taking you from the dawn of civilization to the present day. The kinds of datasets you see in Michael Kremer’s “Population growth and technological change: one million BC to 1990,” which provides an economic model tied to the world’s population data for a million years; or in Tertius Chandler’s Four Thousand Years of Urban Growth, which contains an exhaustive dataset of city populations over millennia. These datasets can humble us and inspire wonder, but they also hold tremendous potential for learning about ourselves.
Not the most in-depth piece in the world, but I like new concepts, and this is one.
AllThingsD has an interesting little story about some recent Forrester research on internet usage. The gist is that Forrester asked people how much time they spend online and the number went down from 2011 instead of up. This would be shocking if it were true, but it’s not, and presents an interesting case of the problem with self-reporting in research. The article sums it up like this:
“Despite the fact that they always have connected devices and are always online, they don’t really realize they’re online,” said Forrester analyst Gina Sverdlov. “They’re using Google Maps or checking in on Facebook, but that’s not considered online because it has become such a part of everyday life.”
It’s actually amazing to me we don’t spend more time talking about the issues with self-reporting because this stuff happens all the time. People classically overreport the good things about themselves and underreport the bad stuff (if you compared the number of New Yorkers who say they read the times with the traffic/newspaper sales something doesn’t add up). On top of that, though, we ask seemingly innocuous questions that actually turn out to be confusing for regular folks (I don’t know how I’d answer how much time I spend “using” the internet either).
There is something important here as the web continues to shift to an ambient medium that is just tied into our lives. I think for those of us that have been living online for awhile now it’s no surprise, but it’s interesting to see others catching up to that way of thinking.
The data here is sort of interesting, but I’m not sure I buy the premise:
Ask Americans if they are willing to spend more to buy American-made products, and nearly half say they are often willing to do this. But in the latest Economist/YouGov Poll, the country where a product is made trails price, quality, and even convenience, as an important factor in consumer decision-making. The public gives even less importance to a product’s brand, its impact on the environment, or the political leanings of the company that produces it.
Basically it breaks down that 76 percent say the price of the product is extremely important, 85 percent say quality is extremely important and only 31 percent say the brand is extremely important. I think the price thing is probably right, quality sounds like something people like to talk themselves into (if we cared that much then why do we buy so much Ikea furniture?) and brand seems like a classic case of people not understanding their own motivations. Sure, it might not seem like it, but when someone walks down the aisle and buys Crest instead of store-brand they’re choosing brand over quality and price and probably not even thinking about it. In fact, to most people I’d guess that brand equals quality.
[via Marginal Revolution]