[ RadSafe ] significance level of cancer in Swedish Balkan personnel

[James Salsman]

Ivor Surveyor wrote:

 > James Salsman  quoted from a paper by  P. Gustavsson, M. Talbäck,
 > A. Lundin, B. Lagercrantz, P.-E. Gyllestad, and L. Fornell,
 > "Incidence of cancer among Swedish military and civil personnel
 > involved in UN missions in the Balkans 1989­99," Occupational and
 > Environmental Medicine, vol. 61 (2004) pp. 171-173:
 >   http://oem.bmjjournals.com/cgi/content/full/61/2/171
 >
 >... the authors are clear that statistical significance is reached.

I think Dr. Surveyor meant to write, "is not reached."  However,
with an overall risk ratio 95% confidence interval of 0.9 to 1.7,
it seems clear that technically the results are insignificant at
p=0.05 but would probably be significant at p=0.1.

This kind of a borderline significance level is a reminder that
declarations of significance should always be accompanied by the
level involved, or better yet, simply state the level at which
the results would be significant.

In any case, this certainly calls out for further study, as the
authors note.  Time will tell.

Sincerely,
James Salsman