COMMUNICATIONS TO THE EDITORS
Bias and Probity
We in INR's Office of Research and Analysis for Africa have read with interest your recent article on "Policy Bias." 2 As a sometime contributor to the INR papers from which it quotes and for the past five months INR analyst for the Portugues African territories, it is perhaps appropriate that I attempt to comment on the views it puts forward. While I cannot claim to be as recent a newcomer to the field of intelligence as the author, a graduate of one year's experience, I have not yet lost the feeling of wonder and trepidation with which one must approach the task of intelligence evaluation.
Let me begin with the specific and proceed to the general. The article asserts that Research Memorandum RAF-21 of January 31, 1962 ("The Portuguese Overseas Reforms: An Appraisal of the First Six Months") seems "to have been written in support of policy." In one sense, I concur wholeheartedly. INR deliberately shaped the paper to support policy by answering a high-level request for evaluation of the reforms thus far undertaken by the Portuguese in their overseas territories. These reforms resulted in large part from direct pressure by the U.S. Government, which felt that drastic changes were required to improve Portugal's image in the world community and to construct a base upon which Angola could become politically and economically viable in the future. An evaluation of the reforms was important to the policy maker primarily as a means of assessing the extent to which they were contributing to these objectives.
It seems to me that any evaluation of these or other reforms must have some such standard of reference. A charge of bias could be brought if the standard of reference were concealed; but the authors of this Memorandum took care to indicate at the outset that the evaluation was being made in the context described. To this extent, therefore, the paper supported policy exactly in the way intelligence must if it is to be useful.
To say that the evaluating was done in the context of U.S. policy is, however, far from admitting that we supported policy by coloring the facts and the evaluation in favor of the policy. We firmly believe that this should not be and has not been the case. To support its charge of bias, the article presents passages out of context and ignores other passages which note Portugal's positive efforts, the physical problems facing it in embarking upon this ambitious program, and its accomplishments to date. Indeed, one wonders what would have happened if, in order to avoid seeming anti-Portuguese, we had evaluated the reforms against the standard of Portugal's own extremely optimistic claims when they were introduced. I think that in the final analysis we may have been kinder to the Portuguese-and equally objective-in doing it the way we did.
The article further attempts to show that INR's evaluation has consistently been at variance with that of the U.S. consul in Luanda. I trust there is no suggestion that INR would be less biased were it to accept one source's evaluation as definitive? In any case, a key point incorporated into the introductory section of the Memorandum was drawn almost verbatim from a Luanda report: "Reports from the overseas provinces indicate that the status of the African still has not changed significantly, despite earnest attempts to expand educational facilities." Furthermore, nearly all recent reports from Luanda, Lourengo Marques, and even Lisbon have tended to concur in the basic INR assessment--i.e., that Portugal has shown neither the resources nor the capacity for implementing far-reaching reforms. This is, I feel sure, not an "antiPortuguese" position; the article itself at one point implies that it is anti-Portuguese not to mention Portugal's lack of resources as an impediment to reform. The INR evaluation is made on the basis of present and past performance-the only valid evidence by which we can judge intention and capacity.
In effect, the article seems to be saying that because we do not in every paper point out that poor little Portugal is doing its best and is suffering as well as inflicting unhappiness, we are biased. Every paper cannot have the whole story in it. As it happens, the issue of Portugal's resources has been carefully studied and much thought given to ways of meeting the problem. But intelligence would surely be doing itself damage if it refused to analyze any one facet of a subject in the light of explicitly stated assumptions and reference points. The reforms could be written about in terms of Portuguese capacity, Portuguese will, African receptivity, human rights, Latin American relations-and any number of other reference points. We wrote about them in the light of two prospects-improved status for the African and eventual self determination-that chiefly concerned the United States at the time, and we carefully explained that we were doing so. I do not believe this approach is biased.
Perhaps the article is basically concerned over the concept of policy-oriented research which underlies INR's production. This concept does not involve corrupting data to make policy look good. It does mean a constant attention to the unspoken estimates underlying policy and an examination of these for their accuracy. Thus if U.S. policy was based on an expectation that the reforms would produce certain results, the research analyst must ever be re-evaluating the likelihood of this expectation's being fulfilled.
We perceive no reason why a basic dichotomy should exist between policy and intelligence. Frequently the policy maker receives from the intelligence community an indigestible compendium of all possible considerations and consequently feels constrained to shape his own estimate of a particular situation. If the community is to play a meaningful role, we believe it must be prepared to present not only such undoubtedly necessary round-ups but also selectively focussed papers that bear directly on specific and limited policy questions in terms of implications, alternatives, and outlook. This is another thing than permitting policy to color intelligence evaluations.