| happydolphin said: @wh1pL4shL1ve_007. My pleasure.
I see what you're saying about casuality, even if they were directly proportional (hypothetical), you wouldn't be able to tell which caused which (HW sales causes more SW releases, or SW releases causes more HW sales). One thing I wanted to ask you. What if we had the following instead for the SW releases trend: A column chart with 6 values for each year: the number of: 1) AAA Nintendo games: 3D Mario, 3D Zelda 2) Big Nintendo hits: Mario Kart Wii, Wii Sports, Wii Play 3) Moderately successful Casual games (Nintendo and 3rd party): Trauma team, cooking mama 4) AAA 3rd party games: MH Tri, Silent Hill: Shattered Memories. 5) B 3rd party games: The Conduit, Madworld 6) Shovelware: Babyz, horsez, etc. Would that give us more data to work with? Would the ratio of Nintendo Big Hits to shovelware be an important factor in the analysis? What if Nintendo released the M+ in a time where the number of yearly games (total and shovelware for that matter) was dwindling? Would the shovelware from previous years affect the result, still causing saturation and buyer confusion? |
Not only we couldn't tell which caused which, but we wouldn't be able to tell if they even caused each other. I remember hearing about a study that showed there was a strong positive linear correlation between ice-cream sales and number of deaths by drowning. You couldn't say "oh, ice-cream caused the drownings because they ate it then went swimming" or "the drownings caused the ice-cream sales, they bought it to console themselves after they drowned (horrible example :P)". In reality the most likely thing is more people swam in the season when more people bought ice-cream: the summer.
As for the chart, I believe all we would get is a very skewed chart (Nintendo simply can't make enough AAA games to balance out the shovelware). Perhaps if we had a chart that shows the proportions of games sold in each category, it would be much more manageable, but I can't seem to find any relevance to this particular study.
In my opinion, the best way to express the data would be a least-squares linear regression line, with ŷ representing the predicted SW sales for a given year (n) and x representing hardware sales of year n-1. Having two charts with different intervals on the axes makes it much more difficult to accurately see the correlation between the two.
