Experimenting in the social sciences
One of the most challenging tasks in social science research is acquiring data in order to confront and test theories about how the world works. Traditionally, data acquisition in the social sciences has relied much more on observation than it did experimental design and trial (this has been especially true for fields like economics, political science, and sociology alike). My goal in writing this post is to familiarize readers with the unconventional idea of using virtual environments in networked video games to run controlled experiments for gathering behavioral data. This brute force tactic allows the researcher to answer hard-hitting social sciences policy questions that would be otherwise inaccessible in the absence of large-scale experimental data.
Most of the criticism in approaching social science issues in the confines of an experimental laboratory setting hinge on the fact that subjects may not face the proper incentives to act in a way they would normally (while not under scrutiny). I argue herein, that not only do people generally face the right incentives in these virtual worlds, but that they also behave accordingly (rationally) as if they were making a real decision so that their actions in the virtual environment closely mimic the behavior of the agent in the real world.
So, part of this blog post is about virtual worlds, as seen in the title of the post. I will abuse the definition of the term “virtual world” by associating it strictly with video games (and virtual reality), but more on this in a bit. The other portion of this post is about REAL trade-offs, so let’s first start by discussing what I mean by a real trade-off.
Trade-offs and costs
Consider the following two simple examples of decisions:
#1) You decide to go to graduate school instead of working at a salary at a job during those years while in school.
#2) You decide to spend the next two hours mining for gold in World of Warcraft instead of studying for a test you have the next day.
Notice that in either decision, something had to have been given up (traded-off). Economists have a name for such things – opportunity costs. These are the values of the best foregone alternatives – the “next best” options – we give up every time we make a choice. In #1, it was the foregone salary from not working – easily measured in dollars (or currency of whatever country you are in). In #2 it was the time that could have been spent studying - a subjective cost that depends on the unique characteristics of an individual’s psychological preferences (how much value YOU place on your study time may be different than for someone else). Take note that decision #2 has an important characteristic – the decision maker is facing a cost of giving up time (something real) to harvest more currency (a virtual asset) in a video game. We will discuss more on why this is important in a minute. For now, the general idea that gets us where we need to go in the discussion is, for the most part, intuitive: the economic cost of something is the value of what you must give up to get it, but back to this idea shortly.
Economists, scarcity, and all that jazz
So who are economists you ask (maybe)? You might imagine a group of grumbly old dudes in suits with glasses and calculators huddled around a table taking painstaking note of market data. Unfortunately, sometimes this isn’t far from the truth. However, for the most part, the typical mistake is confusing us (That’s right! I’m an economist!) with accountants, portfolio managers, analysts, or Wall Street traders (not to say that we aren’t qualified to hold these positions and many times do!). However, the scope of what we study as economists is much broader in context than what a typical “suit” task of structuring a financial asset portfolio or minimizing a tax obligation might require. We’d like to be thought of as scientists since we too, try and explain how the world works, but our field of study is far from being perfected.
Most generally, economists study scarcity. More specifically, economics is the study of how scarce resources are allocated (distributed) in an environment where people behave rationally. That might seem like a lot of terminology, but let’s use a very specific idea to define rationality at this junction (there are many different ways rationality is defined across disciplines, so I am of course, “abusing notation”).
For now, let’s define an individual as rational if the individual makes decisions in their own self-interest. One consequence of the imposition of this definition is that incentives matter: if the next action an individual can take will benefit him more than it costs him, then it is in the individual’s best interest (he is incentivized) to take the next action (economists call this
“weighing” of benefits and costs of the next action, “thinking at the margin”).
Why is scarcity interesting? Well, suppose (counterfactually) that it wasn’t a constraint of our analysis -that is, suppose resources were not scarce. If this were the case, then there would be an unlimited amount of resources available to everybody and the question of how to best allocate them is moot (since nobody would ever go hungry!). The fact of the matter is that scarcity is a constant constraint we all face every day. It is impossible to be in two places at the same time, so the nature of time and physics, itself, introduces an unavoidable dimension of scarcity (and one of the most important that will be discussed in the future). The implications of scarcity on behavior are crucial. Do you think you would play Call of Duty the same way you do normally if instead, you had an infinite amount of health and never had to face the fear of being killed? How about in real life? Would your behavior change?
Back to virtual worlds
Ok. So economists study scarcity under the assumption that people are, for the most part, rational (we’ll see later that this isn’t always the best assumption). Got it. Now let’s get back to virtual worlds. There are two major ways in which virtual worlds may differ from our actual world in terms of scarcity.
First, virtual worlds can, in many cases, can be parameterized (built) in a way that makes them limitless (in terms of size, resources, time horizon, etc.). Have you ever jumped over the flagpole in Super Mario Brothers for the NES (I know - I’m old school)? The world never ends – it spans to the right…FOREVER (however, time runs out of eventually)! Have you ever notice how, in the past, video games with automated “sellers” (there are too many examples) never seem to run out of inventory no matter how much of an item you’ve purchased (not to mention prices don’t change either)? Understanding how people behave in these contexts is also important in understanding the role games play in the development of our character and how we face difficult decisions, scarcity or not.
The second atypical element of virtual worlds that requires attention is the fact that producing another virtual good in a video game requires little to no real cost (In terms of material inputs, the marginal cost of producing another virtual good is near zero). Putting another hammer in the virtual store’s inventory is a simple matter of changing a few lines of code – it can be automated at little to no cost. Putting another hammer in a store in real life requires that the hammer actually be built with raw materials (wood, steel, labor, technology, etc.), costing both time and resources (real resources!). So from the “supply side,” virtual market will function fundamentally different than their tangible counterparts.
So why study these virtual worlds if scarcity is not always present? The answer is again, two-fold (surprise!). First despite some elements of virtual worlds being unbounded, there is one major element of scarcity that is common to all games: the opportunity cost of leisure – the subjective value of what you could be doing with your time instead of doing what you are doing now (like playing video games in decision #2 or perhaps reading this blog post?). This is a cost that defines how people value their leisure. It is important because it is different (heterogeneous) across individuals and is useful in explaining how much time someone is willing to allocate to an activity (say, video games). We will come back to this issue later as it turns out to be a crucial determinant of how much value a player may place on a particular gaming experience.
Secondly, just because scarcity isn’t present in all dimensions of a game doesn’t mean it isn’t present in some. In fact, only recently has scarcity been exploited in the game space as a means to satiate the public’s desire for alternatives to visual realism. For years, the traditional focus of “realism” in games has been superficially relegated to graphics and visual grandiloquence.
However, the advent of the social media boom has certainly turned the tradition on its head. Realism in social interaction has never been exploited on a large scale like it has been today – playing games with friends in a large network means facing trade-offs between cooperating and competing with them, and studies have shown that the multi-player/social element of games is what players are generally most interested in.
My (our) goal
As our games converge in a way that makes our experience richer, the question of how our behaviors adapt has yet to be resolved. Virtual environments can tell us very interesting things about how players interact when they are being monitored (or not) by others when individual decisions and their consequences are interrelated. Perhaps the most interesting aspect of utilizing a virtual environment as an experimental setting is the ability to control the parameters of the environment as the experimenter. That is, virtual environments give researchers the ability to change or shock isolated elements of the environment to analyze how players and aggregate variables respond. This can be as easy as manipulating a few lines of code. These environments, despite their “virtual” nature, provide an excellent atmosphere for conducting controlled behavioral experiments on both the micro (small) and macro (large) scales provided players face real trade-offs and have incentives to act strategically. Virtual environments in video games provide us with a rare and once previously inaccessible opportunity to ask some hard-hitting questions in the social sciences that have yet to be resolved through observation and theory alone. My goal is to illustrate (through several more of these posts & some published research) that data generated from video game players’ experiences using virtual worlds as the experimental settings provides both a compelling and viable method for engaging in frontier behavioral research on a much larger scale than what has ever been done before.
If you like what you hear and want more (or want me to elaborate on any points raised herein), feel free to email me ([email protected]">[email protected]) or follow me on twitter (@SteveLevkoff) to suggest future discussion topics.
Dr. Levkoff is a behavioral scientist and the chief economist at Nerd Kingdom. He also holds joint appointments as a lecturer in both the Department of Economics and the Graduate School of International Relations & Pacific Studies at UC San Diego.