Categories. Groups. Cliques.
People love to classify things. Walk into any store, search
any website. From the sections of the supermarket to the tags on this blog,
there is a common feature of clustering based on commonalities.
But this desire to categorize extends beyond things. People
love to classify people. We classify ourselves, our peers, even people we have
never met.
Take your friends, for instance. You have your friends you
call when something big happens, your “first responders” if you will. You have
your “casual friends” who you might meet up with every few weeks or so for
coffee. Then you have your friends who you can go months without talking to and
yet the friendship never deteriorates. But we have all experienced the agony of
that intermediate friendship that we don’t know how to categorize. You ask
yourself analytical questions to determine whether or not this is the type of
friendship where you can invite them to the movies the day of without scaring
them off.
So why do we like to categorize? It seems to make our lives
easier. When we place situations or people or foods or anything else for that
matter into a box, then we can treat all of those things in that box the same
way. We learn how to respond to one object in that box, and we then know how to
respond to all of the objects grouped in that box. We can learn what we like
and don’t like, what is important to us and what might be “dangerous” without
the hassle of fully and singly experiencing everything out there. As detailed
more in NPR’s Invisibilia episode“The Power of Categories,” categorizing the world around us begins at a young
age and provides us with comfort without requiring a change in ourselves.
I’ll offer another more personal example of how people use
categories to cope. My grandmother was born and raised in Thailand to a Chinese
family that had recently migrated there. She grew up in a neighborhood almost
completely populated by Chinese migrants, and daily saw the juxtaposition of
Chinese culture with Thai culture. My grandmother quickly learned how to
categorize in an effort to have ideal interactions with many different people;
Chinese people do this while Thai people do that. I have no doubt that it was
beneficial to do this in order to be broadly respectful, but at the same time I
always wondered if Person A might just do this because they were Person A.
Using statistical terminology, it seems like our world can
become ruled by discrete variables that we assign to anything and everything, but
aren’t continuous variables more realistic? Politically, people rarely identify
100% with one presidential candidate, or are 100% Democrat or Republican.
Morally, we are not 100% good or bad. Racially, few of us are 100% of one race.
Even medically, we have turned to more personalized treatment based on
individual needs. More recently, our society has seen a pushback against
established discrete social variables concerning gender and sexuality. There is
a cry out to recognize the continuous variables of identity that people truly
are, instead of continue on with the discrete variables that society has so
lazily established. As illustrated in the image below of a political spectrum,
the assignment of broad discrete categories ignores the intricacies of the
data.
It’s an overwhelming reality that we must venture out of our
comfortable categories and observe every new thing on its own and of its own,
but it’s a reality nonetheless. Who’s to say what the best approach is, whether
it is completely changing the current system or simply adding more categories
until they are nearly indistinguishable. But perhaps the first step is adding
new words to our vocabulary.
Spectrum. Blend. Continuous.
Just like categorizing people can have a detrimental effect to those wrongly categorized, data wrongly characterized can have a detrimental effect to the experiment. We categorize because it allows difficult concepts to be understood. Scientists categorize many things, and as I've begun my scientific career, I've wanted to categorize new complex biological pathways by incorporating mechanisms from other similar pathways. But categorizing leaves out the stochastic ability of the system of interest (a person, or a biological system). Humans are unique, and like the spectrum of political standpoints, they are hard to categorize. When we try to categorize people, and as scientists, when we try to categorize our data, we begin to lose some of the information that could end up being the next scientific breakthrough. Categorizing might even be an enemy of creativity.
ReplyDeleteI think the "spectrum revolution" has some interesting implications for demographic data and correlational research. As you said, everything from race to political affiliation to personality to gender are beginning to be conceptualized as continuous, but almost all historical data have treated these variables as discrete. How differently would survey data (or even the census) come out if these variables were analyzed as continuous? In the current environment, I find it fascinating to consider that what we have always analyzed as discrete variables may have just been binned by convention.
ReplyDeleteIt is very true that the human species tends to find the need to fit seemingly everything they experience into neat little categories, and this tendency certainly carries over to research as well. But just as in social categorization, important information is lost in research when continuous variables are categorized as discrete. I think politics in this election cycle makes a particularly good example. With an abundance people crossing party lines with their votes this election, it would be wiser to predict election outcomes based on where people lie along a political spectrum rather than just if they're registered as democrat or republican. With continuous variables, one can get a better look at sources of variation. Discrete, categorized variables provide little indication of such.
ReplyDeleteIt is very true that the human species tends to find the need to fit seemingly everything they experience into neat little categories, and this tendency certainly carries over to research as well. But just as in social categorization, important information is lost in research when continuous variables are categorized as discrete. I think politics in this election cycle makes a particularly good example. With an abundance people crossing party lines with their votes this election, it would be wiser to predict election outcomes based on where people lie along a political spectrum rather than just if they're registered as democrat or republican. With continuous variables, one can get a better look at sources of variation. Discrete, categorized variables provide little indication of such.
ReplyDeleteGreat points, by you and the previous commentors. Categorization has been considered an evolutionary survival trait, but now that we don't have to make decisions on what is "us" and what is "them" for safety as much anymore, it's time we start molding our thinking to be less discrete and more spectral. Instead of male/female, gay/straight/ scientist/artist, liberal/conservative we can recognize that forcing people into bins is an injustice to them as complex individuals.
ReplyDeleteI listed to this podcast on the evolutionary reasons behind categorization, mostly in context of race, a couple years ago: http://radio.seti.org/episodes/What_s_the_Difference_