File talk:Gdp20-40.jpg: Difference between revisions
Jump to navigation
Jump to search
imported>Nick Johnson No edit summary |
imported>Richard Jensen (the goal is show deviation from long-term trends) |
||
| (2 intermediate revisions by 2 users not shown) | |||
| Line 1: | Line 1: | ||
I'm not sure that the treadline is appropriate. It is not a very good predictor for the points shown... Maybe even misleading?--[[User:Nick Johnson|Nick Johnson]] 15:34, 18 April 2007 (CDT) | I'm not sure that the treadline is appropriate. It is not a very good predictor for the points shown... Maybe even misleading?--[[User:Nick Johnson|Nick Johnson]] 15:34, 18 April 2007 (CDT) | ||
**it's just the line of best fit (generated by Excel). [[User:Richard Jensen|Richard Jensen]] 16:30, 18 April 2007 (CDT) | |||
: But its such a bad fit that it's meaningless. Is it a polynomial fit? If so, what order and why? Or, is it an exponential fit? If so, why? IMHO, these things add a lot of meaning to the graph, which you may not have intended. If it's in an article about the new deal, why not fit the treadline to points starting at 1933? It's inappropriate to have such a simple treadline that covers pre-crash, crash, and new deal... This is an example of where a best fit reflects the data very poorly. --[[User:Nick Johnson|Nick Johnson]] 14:00, 19 April 2007 (CDT) | |||
::it's the least squares trend (the most common one in econ). The object is not to show how close is the fit but just to show DEVIATIONS from long-term trends between the wars. (three straight lines would make a good fit--prosperous 1920s, crash 1929-33, recovery 1933-40) so I'll fiddle with that. [[User:Richard Jensen|Richard Jensen]] 22:08, 19 April 2007 (CDT) | |||
Latest revision as of 22:08, 19 April 2007
I'm not sure that the treadline is appropriate. It is not a very good predictor for the points shown... Maybe even misleading?--Nick Johnson 15:34, 18 April 2007 (CDT)
- it's just the line of best fit (generated by Excel). Richard Jensen 16:30, 18 April 2007 (CDT)
- But its such a bad fit that it's meaningless. Is it a polynomial fit? If so, what order and why? Or, is it an exponential fit? If so, why? IMHO, these things add a lot of meaning to the graph, which you may not have intended. If it's in an article about the new deal, why not fit the treadline to points starting at 1933? It's inappropriate to have such a simple treadline that covers pre-crash, crash, and new deal... This is an example of where a best fit reflects the data very poorly. --Nick Johnson 14:00, 19 April 2007 (CDT)
- it's the least squares trend (the most common one in econ). The object is not to show how close is the fit but just to show DEVIATIONS from long-term trends between the wars. (three straight lines would make a good fit--prosperous 1920s, crash 1929-33, recovery 1933-40) so I'll fiddle with that. Richard Jensen 22:08, 19 April 2007 (CDT)
