Professional Documents
Culture Documents
Tucker Hammel
Abstract
The intent of this paper is to discuss whether it is feasible to determine if an indie game that is
designed to make money, will actually be able to do so. This paper reviews about some past
research and how it relates back to the question at hand. In addition, this paper discusses a
preliminary survey of 35 students conducted to determine if they would be dissuaded from
purchasing a game where it is stated that the intent is to make money. Ending with a section that
discusses how a larger study would be conducted, as the results of the research done up to this
point is largely inconclusive with regards to the initial feasibility topic.
Table of Contents
Introduction................................................................................................................ 4
Methods...................................................................................................................... 6
Results........................................................................................................................ 7
Is a regression the best model to determine expected profits?...............................7
Would the model have to be updated very frequently?...........................................8
Do consumers want to avoid purchasing a game that is designed to increase
profits?.................................................................................................................... 8
Conclusion................................................................................................................ 10
Recommendations.................................................................................................... 11
References................................................................................................................ 12
Introduction
Within the field of economics, there is a large void of studies regarding video games and
how to make a profit through this medium (Seo 2015). Larger companies tend to have access to
the funds required to determine the release schedule and current holes in the field to determine
what games are currently in demand and if a game in production will be profitable (Tschang,
2007). Unfortunately, for smaller companies or even sole individuals who desire to turn a profit
from the games that they produce, there are only high cost means in which they can determine
how their game will be received and this can only be used on a case by case basis (Tschang,
2007).
The intent of this paper is to look into the feasibility of developing a model which can be
used to determine the expected profits of a game in development, and if it is capable of being
changed, what should be modified to lead to higher expected sales. The model is intended to
include various interaction terms between genres, genres seasonality, the most statistically
significant variable for skill of production, prior to release advertisement, and a dummy variable
for if the game was created to make money, this would be in addition to any other variables
which appear within available data which have significant t-scores and influence over profit. But
the development of the exact model for profit is for a later study as I determined that it would be
better to first determine if the idea of profit maximization is unappealing to indie game
consumers or even possible to begin with. As such the limit of this paper is a look into past
research as well as the results of a self-made survey, all of which, looking into whether it is
worthwhile to attempt to develop a model which can predict profits in indie games.
The rest of this paper is focused on what I have found through my research and survey
results. Starting with methodology of my research, continuing to the results from said research, a
quick rundown on what conclusions can be made, and wrapping up with how to proceed should
the desire exist to do so.
Methods
In finding past research, the main restriction which I was placing on the papers I was
looking for was if I could apply the information to the video game genre, limiting the goods
which could be used to those with the intent of entertainment. Beyond that restriction however, I
desired to answer a few questions with regards to the model that would be developed and used
those questions to lead my research. The first criteria question which I based my research on was
if I would have to recreate the model shortly after creating it. The second criteria, while I hold
the method highly and have some confidence that it can be, having evidence or backing on if
regressions can be used this way, would also be useful in determining if the model is even
possibly going to work. The final question I had desired to answer was if there were significant
reservations among consumers to purchase games which are designed to make money.
As I believed that there was a better chance on finding answers to my first two questions
through other peer reviewed articles I decided to attempt to answer my third through a survey of
35 Hamline University students.
Results
Is a regression the best model to determine expected profits?
While speaking with certainty is difficult with respect to whether or not it is the best,
what a regression does (reduce sum of squared error between the predicted and observed values
across all available data points), suggests that it is a good fit for this model (Wooldridge, 2015).
There are a few nuances that must be upheld however, including but not limited to, proper
creation of the model, setting aside data so that it can be confirmed to be predictive, and it should
be used only as a causative model if a set of rules are upheld.
With regards to the creation of the model, there are a few things that can or must be done
to different variables to make sure that there is logical reasoning behind its existence within the
model. The most important thing which would have to be done for this model is to make sure to
include interaction terms. Including interaction terms would show how a newly created variable,
which is dependent on two or more other variables, influences profits. As touched on in the
introduction, the basest interaction terms I would be using is largely between binary variables for
genres and different months to denote seasonality of a specific genre. The reason to include this
seasonality variable for each genre is to make sure that as much variation is explained as possible
from those variables when there is an expected relationship unique to the combination of the
variables (Wooldridge, 2015). Of course there are a few things to be on the lookout for when
developing the model, the most important of which is to make sure to not include variations of
variables that do not logically make sense to increase the models R2 or its explained variation.
To make sure that there is in fact some use for the model besides to explain variation
within the dataset that the model is based off of, having a separate set of data to which the model
can be used to attempt to predict the profits of a new set of games, is a common and important
part of the model creation process. There are a number of restrictions to regressions which must
be kept in mind, most of which are tackled in the model creation process but there are a number
which are inherent to various parts of things, the only one which I believe that needs to be noted
is the fact that if the variables values are not normally distributed, the predictive ability of the
model falls apart. The only thing which can be done to combat this is to increase sample size and
have the law of large numbers allow us to operate as if it is normally distributed, unfortunately
there is not a set sample size which fixes different levels of non-normally distributed values and
can only be determined to not be useful after running the regression.
Would the model have to be updated very frequently?
Given that the games industry is in fact vary volatile with regards to its desirable genres
and how people decide to purchase goods (Tschang, 2007). It is safe to assume that at any
singular time that the model is run with relevant data, the output can only be used for a short
period of time. That however does not mean that the model itself is not useful, merely that it will
have to be re-run with current data to determine the optimal changes to make or what the
expected profits will be, at the time of deliberation.
Do consumers want to avoid purchasing a game that is designed to increase profits?
From my survey which was responded to 35 times, I found that the majority of those who
held a significant likelihood of purchasing a game, held only a slight reservation to purchasing a
game which is explicitly created to make a profit (figure 1). Only two would be included in a
larger survey, as they are the only ones who had responded saying that they had previously had a
purchasing history of indie games, and as such this conclusion and all other conclusions made
from this survey, are in no way statistically significant with regards to my initial feasibility
question.
Conclusion
It is difficult to determine whether or not it would truly be a good idea to continue with
the development of a model to determine the expected profits of a game. On one hand, if the
development of the model goes smoothly, all you have to do to determine expected profits is run
the model again with current data. On the other hand, it is entirely possible that the results are
inconclusive, it is possible that a mistake was made when creating the model or when retrieving
data, or that no one within the games audience is open to the idea that the game was developed
with profits in mind.
The creation of the model is, in itself relatively easy once the data is received, it may take
time to think through everything that must be done to make it as thoroughly capable as possible
but should just be a matter of time and logic to get it done properly. This model can be used
multiple times, but it would have to be re-run after some time has passed to make sure the
coefficients are staying current.
From the perspective of someone who desires for the creation of the model, I would say
that it is both a good idea and feasible to create the model. Still, I understand the reservations
that others may hold with the albeit inconclusive, negative impact on sales due to using the
model. In the end I believe that there is a little bit more research that must be done in order to
determine if this model is in fact a good idea to follow through with.
Recommendations
Before collecting the data and starting on the model, the only recommendation that I can
think of at this point in time is to create a new survey, strictly for those who have purchased indie
games and are the market base for the games that would be using the model. When the survey is
re-run, determine if there is in fact a significant impact on desire to purchase a game because it is
developed to make money and attempt to quantify it so that it can be used to determine if the
model once developed will have any positive influence.
References
SEO, Y., BUCHANAN-OLIVER, M., & FAM, K. (2015). Advancing research on computer
game consumption: A future research agenda. Journal Of Consumer Behaviour, 14(6), 353-356.
doi:10.1002/cb.1557
Li, M., Jiang, Q., Tan, C., & Wei, K. (2014). Enhancing User-Game Engagement Through
Software Gaming Elements. Journal Of Management Information Systems, 30(4), 115-150.
doi:10.2753/MIS0742-1222300405
Tschang, F. T. (2007). Balancing the tensions between rationalization and creativity in the video
games industry. Organization Science, 18(6), 989-1005. doi:10.1287/orsc.1070.0299
Diamantopoulos, A., & Winklhofer, H. (2003). Export sales forecasting by UK firms: Technique
utilization and impact on forecast accuracy. Journal Of Business Research, 56(1), 45.
Wooldridge, J. M. (2013) Introductory Econometrics. Boston, Massachusetts: Cengage Learning.