ITT you try to explain confidence intervals and I pedantically ridicule you for
| Vengeful menage | 02/17/19 | | mint chapel mediation | 02/17/19 | | Vengeful menage | 02/17/19 | | Razzmatazz house quadroon | 02/19/19 | | Ruby floppy hall giraffe | 02/17/19 | | Vengeful menage | 02/18/19 | | Ruby floppy hall giraffe | 02/18/19 | | Vengeful menage | 02/19/19 | | sienna wonderful knife | 02/18/19 | | Vengeful menage | 02/19/19 | | sienna wonderful knife | 02/19/19 | | Vengeful menage | 02/22/19 | | sienna wonderful knife | 02/23/19 | | Vengeful menage | 03/09/19 | | Histrionic narrow-minded garrison | 02/19/19 | | Vengeful menage | 02/22/19 | | hateful lemon kitty address | 02/19/19 | | Vengeful menage | 02/22/19 | | hateful lemon kitty address | 02/23/19 | | Vengeful menage | 03/09/19 | | Cheese-eating motley center foreskin | 02/22/19 | | Vengeful menage | 03/09/19 | | buff principal's office reading party | 02/22/19 | | Vengeful menage | 03/09/19 |
Poast new message in this thread
Date: February 17th, 2019 5:38 PM Author: Vengeful menage
... your pathetic layman's knowledge of statistics.
What does it mean when we say that a candidate will get 51% of the vote +/- 1% with a 95% confidence interval?
(http://www.autoadmit.com/thread.php?thread_id=4203709&forum_id=2#37798274) |
Date: February 17th, 2019 5:50 PM Author: mint chapel mediation
a lot of times this idea gets muddled for people because there is some machinery that has to be in place to fully get it.
you need the idea of a known or assumed probability distribution of outcomes (both the distribution for a single trial and the sampling distribution). the idea of success or failure as a binary event based on a subset of the outcomes. and finally the idea that randomness will mean that in some cases what is true will not be observed.
the p value gets at the point that, if a distribution is assumed for individual trials, the observed outcome would only be observed in p proportion of experiments, which is itself a collection of trials. it is a measure of how unlikely the observed outcome is with respect to the reference distribution.
confidence intervals are related to this idea and define an expected range of values that correspond to a given p value or confidence level.
(http://www.autoadmit.com/thread.php?thread_id=4203709&forum_id=2#37798332) |
Date: February 22nd, 2019 11:46 PM Author: Cheese-eating motley center foreskin
What does it mean when we say that a candidate will get 51% of the vote +/- 1% with a 95% confidence interval?
i had a good prof explain it to me a long time ago. it means the true value of the experiment will be within the interval of 50 to 52 percent 19 out of 20 times that we had ran the experiment. in other words, we will have captured the true value to within 1 percentage point 95 percent of the time we ran this experiment, so its very likely.. ie 95 percent likely, ie VERY LIKELY, that the value lies between 50 and 52 percent. there is a 5 percent chance though that the we fucked up and the value is not between 50 and 52. thats called the error margin, ie 5 percent.
or no wait, its something about being the sampling distribution being close to the real distribution.. i dunno, this shit is confusing.
(http://www.autoadmit.com/thread.php?thread_id=4203709&forum_id=2#37829240) |
|
|