================================================================== The gateway between this list and the sci.stat.edu newsgroup will be disabled on June 9. This list will be discontinued on June 21. Subscribe to the new list EDSTAT-L at Penn State using the web interface at http://lists.psu.edu/archives/edstat-l.html. ================================================================== . On Thu, 3 Jun 2004, 6stephen sands wrote:
> I should be so grateful if you please could give me a simple answer > for my following question. I have a sample which accounts for about > 90% of the population. There was no possibility to include the rest > (10%) because of lack of data. From a statistical point of view, Is it > advantageous to have this large proportion which nearly the whole > population. Answers depend on several aspects. 1. What sizes are we discussing? It's one thing if you have a sample of 9 from a population of 10; it's quite another if it's a sample of 9 million out of a population of 10 million. 2. What is the population? What do you want the population to be? (You mention "firms". You might, e.g., be thinking of all of the real estate firms in the state of Delaware as your population, and you have data for 90% of them. But you might have wanted (as the reference population) all the real estate firms in the U.S. Or you might want to consider all the real estate firms that there could have been (whether in Delaware or in the whole U.S.): a conceptual population that is much larger than the number of extant firms.) > How does this influence the inferences and conclusions? How different > the way in which I look to P values in this case from that in which I > have a sample of less percentage? (Let'’s assume that the missing > firms from my sample are represented by similar firms that have > similar characteristics). Unless you also assume that the missing firms are a random sample of the population, or are otherwise arguably representative of the population, the assumption you propose is inadequate. The missing firms (and the hypothetically similar firms) may be extraordinary in some sense -- firms of extremely small size, for example, or at an extreme end of the distribution of one or more other variables. But a detailed answer would depend on what the population of interest really is. > (Statistical analysis used includes Kruskal-Wallis, Mann-Whitney and > Logistic regression). I would highly appreciate it if you please > could state your answer in simple terms for a person who is not good > at statistics? I don't know whether this qualifies as "in simple terms"; but the question itself is perhaps less simple than you had assumed. Hope this helps some. -- DFB. ------------------------------------------------------------ Donald F. Burrill [EMAIL PROTECTED] 56 Sebbins Pond Drive, Bedford, NH 03110 (603) 626-0816
