home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!gatech!news.byu.edu!ux1!fcom.cc.utah.edu!hellgate.utah.edu!caen!sdd.hp.com!wupost!darwin.sura.net!haven.umd.edu!purdue!mentor.cc.purdue.edu!pop.stat.purdue.edu!hrubin
- From: hrubin@pop.stat.purdue.edu (Herman Rubin)
- Newsgroups: sci.math.stat
- Subject: Re: sample size and invdference (was: Maximulikelihood estimates in LISREL)
- Message-ID: <57241@mentor.cc.purdue.edu>
- Date: 19 Aug 92 13:58:43 GMT
- References: <1992Aug18.152355.16237@news.cs.brandeis.edu>
- Sender: news@mentor.cc.purdue.edu
- Organization: Purdue University Statistics Department
- Lines: 31
-
- In article <1992Aug18.152355.16237@news.cs.brandeis.edu> mokaba@binah.cc.brandeis.edu writes:
- >could somebody explain in English what the true effect of sample size
- >is on maximulikelihood estimates. I have been playing around with
- >LISREL and have a little problem conceptualizing the effect of sample
- >size on the estimates I get using maximumlikelihood estimators. Are the
- >effects comming in becuase of the effect of sample size on the standard
- >error of estimates or is there something I am too dense to see.
- >I have also learn that the problem with large sample in LISREL is that they
- >lead to even small effect being significant... or something like that..
- >Furthermore, is a sample of 103 good enough for a full model estimation.
- >i.e running the Mesurement equations at the same time as the Structural
- >model.
- In English, a larger sample is very likely to have a more concentrated
- likelihood function. Among other things, this reduces the "standard error"
- of estimates.
-
- As for statistical significance, neither I nor anyone else has been able
- to come up with any good reason to use it; it is a leftover from the time
- when it was misinterpreted.
-
- As for a sample of 103 being good enough for ..., in some cases samples
- of 10 are quite adequate for purposes, and in some cases samples larger
- than the entire world population are inadequate.
-
- This is not an attempt at obfuscation, but the proper use of statistics
- is not that simple; see a good mathematical statistician.
- --
- Herman Rubin, Dept. of Statistics, Purdue Univ., West Lafayette IN47907-1399
- Phone: (317)494-6054
- hrubin@pop.stat.purdue.edu (Internet, bitnet)
- {purdue,pur-ee}!pop.stat!hrubin(UUCP)
-