PaperdollDecay Posted November 6, 2011 Report Share Posted November 6, 2011 Hi there, I'm in dire need for some IA help. I am doing a chi-squared test to determine if there is any correlation between the length of power metal songs and the decade they were written and produced in. Here is my data: http://tinypic.com/view.php?pic=26284eu&s=5 I got feedback on my first draft, where my teacher concluded that I had too few values for a proper test. What I did was to organize the songs into three groups depending on the length of the songs; short, medium and long. But alas, too few values. I know how to carry out a chi-square test but I need help on how to organize the data so that I can actually do the test. Help. Reply Link to post Share on other sites More sharing options...
Bartosz jez Posted November 6, 2011 Report Share Posted November 6, 2011 There's no such corelation whatsoever and you cannot check it because of too many variables. The date simply cannot be the independent value. Reply Link to post Share on other sites More sharing options...
PaperdollDecay Posted November 6, 2011 Author Report Share Posted November 6, 2011 (edited) There's no such corelation whatsoever and you cannot check it because of too many variables. The date simply cannot be the independent value. I'm sorry, I don't really know how variables work within this test. And I won't be using the date itself, but the decade. By the way, love your Magritte display picture. Edited November 6, 2011 by PaperdollDecay 1 Reply Link to post Share on other sites More sharing options...
Drake Glau Posted November 7, 2011 Report Share Posted November 7, 2011 (edited) Don't use decades. I'd use the age instead because you can use age as an actual quantifiable value. Then you can run a test of age vs length to determine if the length of the songs is determined by how old they are. Use the exact ages and lengths for your data but then you can group the age into decades if you want. Lets say you have a 23, 15, 17, and 24 year old songs. That's years 1987, 1996, 1994, and 1986 respectively. You could group that data in the 90's and 80's by grouping your data into ranges of 12-21 for the 90's and 22-31 for the 80s, etc. Hope that made sense Don't use the mean lengths either. Record every length because you technically need 100 data points to even run a chi-squared test Titles/album names don't matter so you can omit them. You have them grouped sort of already, but you still need the actual age of how old each song is (even if they are the same age since they came from the same album) then do your decade grouping by grouping the ages like I said above. Edited November 7, 2011 by Drake Glau Reply Link to post Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.