Timeline for Is any quantitative property of the population a "parameter"?
Current License: CC BY-SA 4.0
8 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Sep 23, 2019 at 3:51 | history | edited | Alexis | CC BY-SA 4.0 |
added 141 characters in body
|
| Jul 5, 2013 at 16:06 | comment | added | Nick Cox | Much food for thought here. As a quick reply: My answer was not intended to imply that we are in Liberty Hall where anything goes. The context for the Tukey quotation I welcome, as my point of view is that it is usual that the customary hypotheses do not hold in so far as all models are approximations not exactly matched by the data. So far from biting, that clause underlines the value of the different point of view. In general, I am not attempting, nor qualified to produce, more abstract and more mathematically refined formal definitions. | |
| Jul 5, 2013 at 15:32 | comment | added | whuber♦ | I agree with Tukey although you might think from my answer to this thread that I am one of the "ossified" statisticians he challenges. The problem is that you have taken his quotation out of context. Tukey is specifically addressing the question of how to evaluate the properties of procedures "when the hypotheses on which they are customarily developed do not hold." This in no way changes the definitions of things like parameters, estimators, and estimands. In particular, a parameter still is not "whatever we are estimating." | |
| Jul 5, 2013 at 14:03 | history | edited | Nick Cox | CC BY-SA 3.0 |
added 1480 characters in body
|
| Jul 5, 2013 at 13:50 | comment | added | Nick Cox | I will expand my answer to address this. | |
| Jul 5, 2013 at 13:44 | comment | added | whuber♦ | Much of this looks at odds with the standard statistical literature, especially your definition of parameter. It appears to confound the processes of finding a procedure to calculate an estimate and identifying what is to be estimated. The latter--choosing the estimand--is a matter for the scientist or investigator to determine. The former is then selected by the statistician to have desirable properties among all possible procedures for estimating the estimand. There are also technical issues; suffice it to say that a parameter is more restricted than an arbitrary estimand. | |
| Jul 5, 2013 at 8:22 | history | edited | Nick Cox | CC BY-SA 3.0 |
deleted 3 characters in body
|
| Jul 5, 2013 at 8:09 | history | answered | Nick Cox | CC BY-SA 3.0 |