Dear all,
I am a rookie in this listserver, and I my doubts surely will seem very silly for you, but I need your help. I am within my PhD in marketing and I have these doubts:
(1). Confidence interval of a population parameter have meaning under a random sampling?
(2) Under a convenience sampling the population estimate of a, for example a mean, is biased. Statistical inference have not meaning here, but the use of confidence interval could be helpful?
(3) The quasivariance of a sample have to be corrected by the finite population correction. This is not done by most softwares (only SUDAAN or STATA). But it is easy to calculate by hand. The goal for doing this is not overestimate the standard error of the mean.
(4) T test of the mean difference of two subgroups is only feasible if the sample is randomly selected.Correct?
(5) If I have a convenience sampling of two subgroups, can I use the T test? I think T test could be valid for analysing if the means are different (altough I know that these means are biased). So, my results are not generalizable to the population, but my results would be valid for my sample.
Any advice would be really appreciate.
Thank you so much
Jose.
---------------------------------
LLama Gratis a cualquier PC del Mundo.
Llamadas a fijos y móviles desde 1 céntimo por minuto.
http://es.voice.yahoo.com
|