Boa tarde.
Encontrei esse texto que mostra quando devemos usar média+/-desvio-padrão (dp) ou média+/-erro padrão (dp/raiz(n)). Para quem publica resultados estatísticos em artigos científicos, acho que vale a leitura. Caso haja algum erro nas assertivas, agradeceria os comentários.
Abraço, Luiz
"In describing a distribution, mean and standard deviation are usually suitable measures of the centre and spread. Unfortunately, many persons tend to present the mean and standard deviation as, for example, 44.3+/-17.0, with a note in the text that this signifies a mean plus or minus a standard deviation (...). This should not be done. Such an _expression_ can only be interpreted if the variable has a normal distribution. (...) On the other hand, the mean plus or minus a multiple of standard error (of the mean) has an interpretation (as long as the sample size is not too small), even if the variable does not have a normal distribution."
Em suma,
Mean +/- 1.96*standard deviations .... contains about 95% of observations (numa distribuição Normal)
Caso contrário,
Mean +/- 1.96*standard errors .......... is a 95% confidence interval for the mean (numa distribuição não-Normal)
Fonte: (Interpretation and uses of medical statistics - Leslie E. Daly, Geoffrey Joseph Bourke)
Link: http://books.google.com.br/books?id=AY7LnYkiLNkC&pg=PA100&lpg=PA100&dq=interpretation+of+standard+error+of+the+mean+.edu&source=bl&ots=_7tVCb4Ro2&sig=1V6XjJqdGJ1StBxTxDjj0KLQeYQ&hl=pt-BR&ei=d-WOSsjsFc-Ptgez1MnOBA&sa=X&oi=book_result&ct=result&resnum=3#v=onepage&q=&f=false
Attachment:
Standard deviations and standard errors.pdf
Description: Adobe PDF document