## Some uDG2004 Statistics

Member
Posts: 201
Joined: 2002.06
Post: #1
I calculated the percent correlation of some of the numerical data (the data involving text is a bit harder to manipulate). I made sure to keep the numbers negative for indirect correlations.

Code:
```Gra        Aud        Ori        Sto        Pol        Pts        Overall Gameplay    44.32%     59.20%     28.61%     7.34%      45.78%     28.21%     69.15% Graphics               82.00%     25.35%     15.16%     58.08%     41.88%     81.09% Audio                             39.65%     24.81%     56.62%     44.76%     92.62% Originality                                  26.22%     16.75%     28.57%     53.16% Story                                                   0.09%      15.40%     28.04% Polish                                                  43.19%     63.22% Contest Pts                                                        52.60%             Pri        Age        HCo        Inc Gameplay    14.79%     -0.03%     -0.05%     0.24% Graphics    14.73%     2.72%      -0.10%     1.10% Audio       14.36%     0.94%      -0.16%     1.10% Price                  0.41%      -0.55%     0.00% Age                               -10.14%    0.09% Hardcore                                     1.81%```

To explain a few of the types of data, Price(Pri) is the shareware price that voters thought the game would be worth, Hardcore(HCo) is a 1-4 scale of how hardcore of a gamer the voter thinks he is, and Inc stands for spendable income.

I have been looking at numbers for a while now, so I don't really feel like interpreting anything right now. I will just show my results.
Moderator
Posts: 767
Joined: 2003.04
Post: #2
You may want to check this thread

I have all the stats I calculated here: http://www.reversecode.com/powermacx/udg04stats.html

As you can see, my correlation results and yours have very different values, did you do the correlation on ALL the data at once? I first grouped it by game, so that may be the cause.
Member
Posts: 201
Joined: 2002.06
Post: #3
How did I fail to notice that thread?

Anyway, the first table was created using data from the average scores of each game. The second was created using all the raw data at once. In addition, my results are reported in percent correlation instead of correlation coefficient. I did that because I think it is a bit more meaningful.

Oh well. It looks like you did most of the dirty work, and very nicely reported as well. I wish I had seen that before.

[Edit: I just realized that you also reported a lot of the quantitative data in the form of tables rather than correlations, such as casual-hardcore. I used the 1-4 values as data in a correlation, but you sectioned them into tables of data.]