Letters

In the spirit of Dr. Garfield’s remarks in your July 25 issue, may I bring up problems I have with some statistics in articles in that issue? On page 18, you show an analysis of research output 1979 to 1987, state by state. The number of states marked in red versus those in black is suggestive that something is wrong, and the actual figures add up to a net increase of 7.6% for the 50 states and the District of Columbia. This may reflect an increase in co-authorship across stat lines

Sep 5, 1988
Roland Hirsch

In the spirit of Dr. Garfield’s remarks in your July 25 issue, may I bring up problems I have with some statistics in articles in that issue?

On page 18, you show an analysis of research output 1979 to 1987, state by state. The number of states marked in red versus those in black is suggestive that something is wrong, and the actual figures add up to a net increase of 7.6% for the 50 states and the District of Columbia. This may reflect an increase in co-authorship across stat lines from 1979 to 1987. The percentages in each year should have been based on the sum for the states in that year and not on the net number of articles. On page 19, your headline states “Collective bargaining seen as boon to science salaries.” This may well be true, but there is nothing in the text to suggest that the study is valid. Perhaps the institutions with collective bargaining had higher salaries even before their collective bargaining agreements were instituted.

Finally, I believe that the chart on page 20 showing average NIH grant size has the caption reversed. The current dollars should be the upper, heavy line, and the constant dollars should be the lower, thin line.

Despite these comments, I must state that The Scientist is an interesting and informative publication, and I have subscribed to it at the government agency which I am joining.

ROLAND F. HIRSCH
111 Orchard Drive
Gaithersburg, Md. 20878

[Ed note: Reader Hirsch has a keen eye. Just as the percentage shares for any gwen year in the page 18 map add up to more than 100% because of coauthorship, the apparent increase of 7.6% percent between years does indeed reflect an increase in coauthorship. Using the total sum for the states would have eliminated these conse- quences of coauthorship, as Hirsch points out, although the data are still valid as presented. We agree that, on page 19, collective bargaining is not the only explanation for the difference in salaries. The article may have made the connection appear stronger than the evidence warrants. On page 20, the caption indeed was reversed. We noticed the mistake and thought we had corrected it, but the change was unfortunately not made.]