Video Game Sales
By: Adam Younes and Dianne Sebek
|
|
|
|
|
|
|
|
|
|
|
|
Purpose
This project was done to determine if a certain video game genre is preferred, and therefore bought, more frequently than other genres. There is no problem to be solved or issues that were raised for this project.
Background
Research
There is a common question that is asked in the video game community and that is what genre of video game brings in the most sales? There are many types of popular video games that people enjoy playing life action, combat, simulation, platform, and many more. There have been many studies done about the popularity of these types of games and how much in sales these genres bring in over an extended period of time. This is a relevant topic of discussion because video games play such a big part in modern-day entertainment and play a role In the economy. Popular video games can also reflect the attitudes of society and can also bring eachother together by getting everyone hooked and talking about the game. Therefore it is importamnst to analyze data about video games and the demand for each genre.
Click here to view raw data.
Data and Graphs
The raw data of the sampled genres in tabular form. Values are in millions of sales.
This table shows the observed, expected, and contribution to the chi-square of the sampled genres from the population.
Data and Graphs
Descriptive Statistics of the sample data (values are in millions of sales).
Probability Distribution of the Chi-Square for the sample. (The shaded probability is 190.561+, so small that it’s not visible).
Data and Graphs
A bar graph of the total number of sales (in millions) per video game genre. This proves that the large counts condition is met.
Significance Test
Step 1
This significance test consists of a chi-squared goodness-of-fit (GOF) test. to determine if action games are sold twice as much as every other video game genre.
The three conditions for a significance test are met:
✔ Random: A random population sample was taken using “Column Statistics” from Minitab.
✔ 10%: The sample size is less than 10% of the population. (1,650 < 16.598)
✔ Large Counts/Normal: All video game genres have an expected count of more than 5 sales.
Each category is a proportion of that genre from the total number of sales
Step 2, Hypotheses
Ho: The proportion of sales of action video games is twice as much as every other video game genre.
Ha: The proportion of sales of action video games are not twice as much as every other video game.
Significance Level (σ): 0.01
Step 3
The obtained test statistic for the sample is X2=190.561. This was obtained through Minitab using a Chi-Square Goodness of Fit Test for my sample (using a 2:1:1:1... ratio for action games to all other video game genres). Since there were 12 total genres in our sample, our experiment has 11 degrees of freedom (df).
Step 4, Calculating the P-value
Also using Minitab’s Chi-Square Goodness of Fit Test for the sample, the P-value obtained was 0.000. The distribution plot is shown below. (Shaded region is too small to see
Random Sampling
Suggestions
Our population was acquired through Kaggle and our sample was obtained through Minitab. From all of the data that was provided through the dataset, we isolated the information that was useful for our hypothesis. We used the columns that provided genre and global sales. Our sample size is 1,650 and our population of interest is video games. The population size was originally 16,598. This meets the random condition and 10% condition.
Something that we could have changed to make this analysis more accurate would be by condensing the time frame of the games that we were looking at. If we limited the games that we looked at to games that came out over the span of one year then we would have eliminated our confoudnmig variable
Weaknesses
A key weakness in our data is the fact that all video games have different release dates, so as one new game releases the previous games have more time to rack in sales while the other one is fresh off the press. Due to this confounding variable games with older release dates will have an advantage over the newer games when it comes to sales.
Conclusion
Since the P-value obtained from our significance test is less than the significance level (0.000 < 0.01), we have enough evidence to reject the null hypothesis. We conclude that out of the total number of video game sales, those the proportion of video games sold that are considered action games are not twice as much as every other genre.
