WMOD00464

• Title: The Management of Weather Resources - Volume II
• Prepared For: The Weather Modification Advisory Board
• Date: 6/30/1978
• Weather Modification - Doc Type: Report

8-4

There is clear evidence of laudable sensitivity of the FACE investigators to the risks of
subjective bias (e.g., Woodley and Sax, 1976). However, risks remained to the extent that the
scientist was able on some days to predict the seeding decision before he then was responsible
for judging whether cloud conditions for that day remained acceptable for inclusion in the
study, i.e., for the distinction between A and B days; the scientist's decision may then have
been influenced by his knowledge of whether or not the day had been selected for seeding.
Similar risks also attend the definition of floating target, to the extent that any component of
that definition depends on subjective judgment. It is noteworthy that the authors of FACE pro-
vide data and analyses for A and B days combined, as well as only for B days, and for results
over a fixed target, as well as for a floating target. Stronger results are reported for B days
alone and for the floating target.

For an adequate confirmatory phase, which is yet to begin for FACE, it would be neces-
sary to guard much more effectively against possible bias in decision making as a consequence
of the scientists' knowledge, however gained, of whether seeding was employed on each day.
* NHRE *

In the National Hail Research Experiment, no effort seems to have been made to keep
knowledge of the treatment condition from project scientists.

* Colorado River Basin *

No attempts seem to have been made to withhold knowledge of the treatment decision
from the experimental personnel.

* Israeli II ·

Only preliminary reports seem to be available from the Second Israeli Cloud Seeding
Experiment of 1969-1975. From these reports, it is clear that the randomization schedule was
determined in advance of the season, and we presume that both decision making and data-
handling personnel knew (or could discover) whether a given day <.was to be seeded or not. All
results, however, are based upon a separation of days into seeded or unseeded, based solely
upon the randomization schedule. No important experimental decisions seem to have been
allowed during the course of an experimental day, at least none that could influence the results.
There remains some concern, of course, over the extent to which data-handling personnel were
aware of seeding decisions.

* Alberta hail *

In the Alberta hail project, on each experimental day, a sealed envelope was opened by
the "mission scientist", containing the randomly selected decision to seed or not to seed. The
mission scientist thus was aware of that decision as he decided the deployment strategies for the
various recording teams. The study thus is subject to possible bias due to influence of
knowledge of seeding on measurement procedures.
* Santa Barbara II *

It appears that in the Santa Barbara rain experiment II, with unconstrained randomization
implemented by the seeding technician, knowledge of "seed or no-seed" was available neither
to the scientific director nor to the analysis staff until after completion of the post-season
analysis.

* South Africa hail *

The Nelspruit (South Africa) hail project is a nonrandomized study, with conclusions
dependent upon comparison with historical records. The dependent variable is hail damage as