Lesley Jackson (X) (asker) Oct 25, 2011:

Yes, I went with "standard error" (SE)... I'm convinced, thanks to what I found in Wikipedia (quoted below) and to DLyons and Benjamin for further clarification.

Benjamin A Flores Oct 25, 2011:

Uso de ET en la especialidad En las tablas que aparecen en http://www.psicothema.com/pdf/3519.pdf
Aparece la siguiente NOTA: B= coeficiente; E.T.= error estándar; p= probabilidad; OR= odds ratio; I.C.= intervalo
de confianza al 95%

In this document
http://tolstoy.newcastle.edu.au/R/e2/help/06/09/0930.html
One can see "standard error OF coefficient" I agree with @DLyons about the values, but heaven knows what the sample is, or what is it they are plotting... Hope to be of help

DLyons Oct 25, 2011:

@Lesley My best guess (and it's only that) is there's a model with dependent variable y (measuring state of health) and independent variables x1 (age), x2 (years).

The model is y = b0 + b1*x1 + b2*x2.

It's not entirely clear what x1, x2 are but I'd imagine (person's age) and (no of years with depression). Your coefficient (-0.33) is the (constant) multiplier of one of these variables. The coefficient is highly significant (P<0.001) so its SE should be small - that's consistent with it being 0.01.

So, without more context, I read your text as saying:
Coefficient = -0.33
ET of coefficient = 0.01 [and ET is standard error]

Lesley Jackson (X) (asker) Oct 25, 2011:

No other context / clues... there is just this one instance and no clues in the running text (definitely NOT one of the best papers I've seen in this journal).

In my simplistic understanding of a coefficient as a "constant" in an equation (a linear regression analysis, in this case), I'm wondering if ET (figure in parentheses) might be a factor in how the coefficient was arrived at??

A coefficient isn't a "result"... so I'm pretty sure ET has nothing to do with standard deviation/error---although Wikipedia has this statement: "In regression analysis, the term "standard error" is also used in the phrase "standard error of the regression" to mean the ordinary least squares estimate of the standard deviation of the underlying errors."

Would this explain ET and is this how it would be reported? (in parentheses next to the coefficient for each variable) Thanks!

DLyons Oct 25, 2011:

More context? It seems like the beta1 or beta2 coefficient, but the reporting format is unusual. The P value looks really low for this sort of study
but could be consistent with (0,01) representing the SE of a beta!

Does it have anywhere another entry Coeficiente (XXX) [YYY (ZZZ)]
(especially where the XXX is "AT")?

Joseph Tein Oct 25, 2011:

Multiple possibilities! Hi Lesley,

This could mean so many things ... I hope someone who is expert in statistics will tell you what it stands for in this case. The tremedica site has over 30 possible meanings for this ... of those, "error típico" has a statistical flavor to it, but I don't know if this is what we're looking for here. Also look at this site (if you don't know it already): http://www.globalacronyms.com/et