I understand the importance MeSH had in the past when google scholar (and similar tools) did not exist.
However, with all the websites that enhance your searches, like Mendeley, Academia.edu, and Google is there any reason to stick to MeSH terms? Or should I use keywords at my will?
For example, suppose I want to submit a paper that discusses some theoretical elements of structural equation model (a group of statistical methods), however there is no MeSH term that quite represents that. The closest one, would be "linear models". In this case, should I use a created keyword like "structural equation model", or sould I stick to "linear models" wich is more general, but is listed by pubmed.com?