On 14 Jan 2003 10:10:23 -0800, [EMAIL PROTECTED] (dennis roberts) wrote:
> rich ... i really don't want to spend my time arguing with you about my > little handout made for my class that i shared with folks IF they thought > they could get some benefit out of it ... - I will keep answering questions, so long as you make progress - > > but, i did take the basic notion, which i think is good, from moore and > mccabe's (which i acknowledged) "intro to the practice of statistics" 3rd > edition ... which says in at least two places: > > p443 ... " ... the confidence interval for a population mean will have a > specified margin of error (m) when the sample size is ... " > > and then uses the formula that i put in the handout ... though, here they > make an assumption of the population SD and DO use z ... rather than t that > i used The Basic Notion is fine. Your personal variation, claiming 95% confidence in attaining a 95% Confidence Interval, was not fine. The "z" is like what I said in my last post, "I gave you the 100% assurance/ power version: variance is fixed (for example, polls)." SO, you say that Moore and McCabe used z; THUS, there would be 100% assurance on p443. You can't blame that precedent for where you claimed 95% assurance, even though you started with the estimate of the variance based on a small sample. > > p506 ... when they talk about the SAME notion of CIs ... but, in the case > when we don't know the SD ... and use the sample estimator for it ... they > talk about a t confidence interval and further say: " ... so the margin of > error (m) for the population mean when we use t that data to estimate mu is > t * (s/sqrt n) ... " > > t is what would be necessary for building a regular kind of 95% (or other) CI I don't read any invalid claim by M and McC in what you quote here; nor any precedent or justification for the way you worded yours. > > if you want to argue with moore and/or mccabe ... that's fine ... but, i > have not heard anyone on this list ... really argue vociferously that what > THEY did ... was in error along the lines you have been fussing with my > handout about ... Eventually, I will get to the local library where they have the textbook on 2-hour reserve. I suspect M and McC wrote it all just fine - technically speaking. DMR has not shown me their fault. I don't know if I can will be able to offer M & McC any way to "idiot-proof" it -- I expect that they would be interested -- because even this local form of Interactive-Text-Learning can't seem to progress very fast. Will someone else pitch in, and say that I do make sense, or else that I've lost them, too? -- Rich Ulrich, [EMAIL PROTECTED] http://www.pitt.edu/~wpilib/index.html . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
