The tag is described as being for structural equation modeling, but a proportion (my hunch: 25%) of its use is for (econometric) simultaneous equation models.

Is it worth clarifying, or splitting the tag?

  • $\begingroup$ On the basis that, sadly, nobody reads the documentation splitting it seems more likely to achieve anything. $\endgroup$ – mdewey Jun 13 '16 at 21:06
  • $\begingroup$ Clarifying never hurts. You could also go through the threads & remove it from those not about SEM. (A few at a time, during slow periods, etc., as usual.) $\endgroup$ – gung - Reinstate Monica Jun 14 '16 at 2:34
  • $\begingroup$ @Gung - I've been doing that when the questions appear for a while. I feel a little bad that I don't have a tag to replace it with. $\endgroup$ – Jeremy Miles Jun 14 '16 at 16:50
  • $\begingroup$ I think it's fine to just delete an inappropriate tag & replace it with nothing. $\endgroup$ – gung - Reinstate Monica Jun 14 '16 at 17:22
  • $\begingroup$ Although it may not be used this way at the moment, it's also worth noting that SEM is commonly used as an abbreviation for "standard error of the mean" (e.g. Wikipedia does so) which is another potential source of confusion for people adding a tag. $\endgroup$ – Silverfish Jun 21 '16 at 0:12

In my opinion, splitting seems unnecessary since simultaneous equation models and structural equation models are essentially the same thing -- more precisely, simultaneous equation models are a special case of structural equation models (in the structural equation model world, so-called simultaneous equation models would be called path analysis models). This is according to, among others, Bollen 1987, who claims that this fact is well known. So anyway, to have two separate tags would seem to be splitting hairs, and I think it's actually rather convenient (for our purposes on this site) that they work out to have the same acronym and thus tag.

With that said, I suppose it couldn't hurt to clarify the sem tag description to point out that some people using this tag will only have the special case of simultaneous equation models in mind.


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .