221
edits
Changes
Jump to navigation
Jump to search
Line 130:
Line 130: + + + +
Talk:Essay:Best New Conservative Words (view source)
Revision as of 13:44, July 27, 2010
, 13:44, July 27, 2010→Selection bias
: Selection bias exists in any study. The issue is not whether there is selection bias (there always is), but whether the selection bias is so great that it disqualifies the results. Unless there were a strong underlying pattern of increase by century, it would be almost impossible even with high selection bias to attain the resulting pattern of doubling by century.--[[User:Aschlafly|Andy Schlafly]] 10:51, 25 July 2010 (EDT)
: Selection bias exists in any study. The issue is not whether there is selection bias (there always is), but whether the selection bias is so great that it disqualifies the results. Unless there were a strong underlying pattern of increase by century, it would be almost impossible even with high selection bias to attain the resulting pattern of doubling by century.--[[User:Aschlafly|Andy Schlafly]] 10:51, 25 July 2010 (EDT)
::*''Selection bias exists in any study. '' But most scientists try to avoid it (even in the social sciences), and try to monitor its effect. They most certainly should not embrace it as a way to make their point (that is, they are called on it when they do so...)
::*''Unless there were a strong underlying pattern of increase by century, it would be almost impossible even with high selection bias to attain the resulting pattern of doubling by century.'' But [[Essay:Conservapedia's Law|Conservapedia's Law]] doesn't claim that ''their is a increase by century''. No, it explicitly states that ''conservative insights increase over time at a geometric rate, as in 1-2-4-8-16-etc. For example, there is a doubling in effective new conservative terms per century.'' While their may an increase over the centuries, the rate of this increase (doubling, i.e. an increase by 100% by century) is an artefact of the way you perform your search: That is, even if the real rate is 70% , 130% - or 83% (the [[maximum likelihood estimator]] for your current set of words taken into account the year of their creation), you end up with a perfect fit of 100% - unless you have enumerated ''all'' conservative words at least for one century.
[[User:RonLar|RonLar]] 09:44, 27 July 2010 (EDT)
===An unbiased test===
===An unbiased test===