course=one of four courses, numbered 1 through 4.  

myballoonsleft=# of balloons or the power that I had when the game ended from 0 to 3.

opponentballoonsleft=# of balloons that my opponent had left.  The game ends when one person has 0 balloons out of 3.

result=w-win or l-loss.  If myballoonsleft=0, it will be a loss.  If my balloonsleft=1-3, the result will be a win.

matchwinpercentage=percentage of the games out of 20 that I won. 

opponentrating=a rating based on how my opponent's skill compares to mine, separate from matchwinpercentage.

time=length of the match, measured in units.  The real time in seconds is about 4.5 times this amount.  The units were measured off a video of a game recording.  The video which recorded the screen occassionally froze the screen causing the actual time of the game to appear slightly slower.  Unfortunately, this seemed to be random and could not be measured, so I left the time in units.

This experiment was performed on a video game called Super Mario Kart, released in 1992.  The purpose is to determine whether the course played has a significant impact on the length of the match.  This is important because I need to manage my free time accordingly and would like to avoid playing long or "boring" courses.  I have reason to believe the course times vary because there are more barriers on certain courses, which help a player avoid losing balloons and losing the game.  Course 1 should be the shortest because there is open space everywhere.  Course 4 should be the longest because it has the most walls. Courses 2 and 3 vary in their wall positioning, but the length of these games could be similar to 4. 
picture:http://www.planetnintendo.com/mariokart/smk/battlemode.html

I've separated the games by matches with five different opponents.  I played each opponent 20 times, 5 times on each course, approximately in course order.  I rated my opponent's 1 through 10 based on the many times I've played them.  I avoided measuring with any horrible players because I'm not likely to play any of them.  There are no 9s or 10s because I haven't found these players yet.

Match win percentage may be the most important factor in predicting the length of a game.  Players may play slower and take less risks in close matches and take more risks if they are either behind or ahead by many games.  This is not as accurate as measuring the win percentage after each individual game but similar, and I didn't want to add a complicated/psychological variable. 

If balloonsleft=0 for either my opponent or myself, the game is over.  There is no point comparing the course or time with 0 balloons because one player always will have 0 balloons.  If I have all 3 or 2 of my balloons left, I tend to be aggressive, shortening the match if I'm successful.  If I have only 1 balloon, I'll play defensively.  I'm successful most of the time, so I usually stay aggressive.  I believe better opponents will use a similar strategy.

There is a random factor which is the character's attacking item.  A red shell item tracks the opponent effectively, and a green shell can only move linearly.  Both items occur equally about 1/3 of the time.  The remaining 1/3 are other, less effective items.  There is a chance that one player will randomly receieve weak items while the other player receives mostly reds, but this is only noticeable every 10 games or so.


Once again, the variable I'm most interested in is the course, 1-4, in relation to the time.  You do not have to analyze the other variables extensively unless they lead to a more accurate analysis.