When I had heard this statistic in the past, I then often thought "Well, maybe it's accurate but it must be out of date." As it turns out, the 92% statistic is certainly inaccurate.
Let's look only at the cases from 1980-1985 as recounted by Varvaro in his 1991 presentation. Vasoli states that only five of the cases received affirmative decisions. Two of these involved defective consent. One involved physical impotence and two were defects of form. From this, Vasoli concludes: "the Rota and the American tribunals were at odds in 95 percent of the cases where grounds were related to defective consent" (p. 61; 42/44 cases = 95%).
Vasoli's error stems from a misunderstanding of this comment from Varvaro: "Unless otherwise noted all decisions mentioned below were for validity of the marriage" (p. 28). In Varvaro's list of cases, Vasoli is correct that you will indeed only find five cases followed by the words "pro nullitate", and only two of them for defective consent (pp. 28-29). In addition to these five cases, however, the Rota ratified four decisions of American courts. One denied a further hearing of a case, thereby accepting the lower courts' decision. Three declared that the American sentence was not null (the decision of the lower court was allowed to stand). These are eight additional cases where the Rota concurred with an American court but the two words "pro nullitate" are not present: other Latin phrases are used to indicate the Rotal action. Adding the numbers, we see not a 42/44 ratio but 34/44. I think it's fairer to include all the cases (so, 34/47--72%) but even following Vasoli's deletion of three cases of agreement between the Rota and America results in not 95% disagreement but 77%. This is a significant difference but is this the actual "reversal rate"?
Some background information before answering that question. There are two, typical ways a case can go before the Rota: a Party appeals the decision of a lower court or the case automatically proceeds to the Rota after conflicting decisions of two lower courts. For example, Clark Kent introduces a nullity case before the tribunal of Metropolis. That court renders an "affirmative" decision. Lois Lane appeals the decision to the Rota. If she did not appeal the case, it must then move to Metropolis' appeal court, in Gotham. Let's say Gotham reaches the opposite conclusion: the marriage has not been proven invalid. Clark appeals that decision to the Rota. If the case first received a "negative" decision and then an "affirmative" decision in American courts, it would automatically proceed to the Rota with no appeal. It seems that Vasoli overlooked this, even though the point was made by Varvaro:
We also want to remind ourselves at this point--because we too easily forget it--that many of these cases reached the Rota because of a discordant decision given at the appellant level. Some of these decisions were negative in first instance and affirmative in second instance and therefore came to the Rota in third instance. Others were first instance affirmatives, reversed in the appellate jurisdiction, and appealed by the petitioners to the Rota (Varvaro, p. 25).
We can see, then, that "many" Rotal cases will have already received conflicting decisions in the USA: one affirmative and one negative. No matter what the Rota does, it is not really "overturning" or "at odds" with the American tribunals. It is reaching a final decision which will be in accordance with one lower court and contrary to the other. While the plain statistics of the cases listed by Varvaro reveal that relatively few marriages were declared invalid by the Rota, many of those cases already received a "negative" decision from an American court.
Since I do not have the text of all of the decisions before me (I only have summaries of some of them), I can't say how many fall into this category. Let's suppose that 50% received a negative and affirmative in the USA. Applying this to the numbers from earlier (34/44), we have a 17/44 ratio (39%) where the USA only said "affirmative" and the Rota said "negative." Again, this is only an estimate based on the summaries I have seen but, for the sake of argument, I think it is fair.
It is apparent that I have not yet actually addressed the 92% statistic. Vasoli arrived at this number by referring to decisions from 1980-1988 and 1991-1992. I have only addressed the cases from 1980-1985 and I will not examine the cases from 1986-1988 and 1991-1992, since I have already demonstrated that Vasoli misunderstood/misinterpreted at least about half of the data he used and therefore his conclusions are mistaken. Instead of 92%, I think the true statistic is clearly closer to 40%.
Is that a good number? No, I think it's too high. Are all American Judges/Tribunals always right and above suspicion? Of course not. But, if we are going to criticize American Judges/Tribunals, in the hope of improving them, let's use accurate information.