Dr. Weinstein’s Experience With Peer Review
Posted by Jeff Id on May 31, 2009
I always prefer to put up articles which show actual experience rather than a hearsay report. Dr. Weinstein has taken the time to describe his own first hand experiences with peer review and confirmation bias in a far less politicized field. It is IMO one of the primary issues in play in AGW, as the Santer issue discussed at Climate Audit demonstrates. I definitely plan to discuss that one more, after SteveM’s correct and reasonable rebuttal paper is accepted by some journal.
Dr. Weinstein’s discussion has to do with the reduction of airfoil drag by manipulations of the boundary layer. – Something I also have experience in ;)
Some Limitations of the Peer Review Process and Its Effect on the AGW Issue
Leonard Weinstein, ScD
May 29, 2009
Monographs, books, technical papers, white papers, and other forms of published material that are to be widely disseminated should ideally be examined by independent reviewers (generally called peer reviewers) with reasonable knowledge of the subject area. The reviewer may or may not be as expert in the details of the subject as the author, but should be at least generally expert in a broader overlapping field. Frequently papers have material that covers more than one subject area and people within a narrow field may do poorly reviewing the broader subject. It is frequently best to have some reviewers that are knowledgeable people but not as close to the narrow field as the author, for a more independent and broader coverage.
The reviews, possibly resulting in suggested changes, do not assure the final material is correct. The lack of a review also does not mean the results are not correct and clear. It just increases the chance for errors to slip through.
If the paper is very complex or has numerous references, it may not be possible to do as good a review as desired in a reasonable time unless the reviewer is intimately familiar with the work. These reviews are often done within a small group of experts, who may even have collaborated on each others papers. The members of the overly narrow review group may have evolved toward a common understanding on the subject, and if the understanding is not correct, almost all of the limited number of experts in that field may make common mistakes that outside reviewers may not make. However, a reviewer outside that group may not be expert enough to be fully on top of the material. This is a problem without a clear solution, especially in fields with a limited number of experts. There are also occasions when errors in assumptions used in developing equations, or in interpretation of instrumentation result in erroneous conclusions that are repeated in many papers until the error is caught and corrected. For this reason all papers must be considered suspect until history vindicates or falsifies them. A paradigm shift on a subject may result from these types of errors being corrected at later times.
An example of the problems that may arise with peer-reviewed papers occurred on a project that I worked on for several years. The branch I was working with was examining possible methods to reduce turbulent skin friction over surfaces, with the goal of reducing aircraft and ship drag, and thus save fuel. We had developed several successful concepts that gave small but useful reductions in drag. These included “longitudinal microgrooves”, “large eddy breakup devices”, and for water flow, “air bubbles in longitudinal grooves”. An additional concept we worked on was a “compliant wall”. This last method appeared to be particularly promising as a method to damp out near wall turbulence by absorbing wall pressure fluctuations. Unlike the other methods, this last technique required a soft and movable wall, so it was more difficult to measure the drag of a test object. Several other groups all over the world also started work on this concept (bandwagon effect), and our group (and many of the other groups) obtained what we thought were favorable results in several tests. Dozens of peer-reviewed papers were published in main line journals (e.g., Journal of Fluid Physics, AIAA journal, etc.). The main experimental technique used to measure the drag for this type of model was a hot-wire Reynolds stress probe. Test results seemed to indicate a large reduction in drag. In order to independently verify the favorable result obtained with this technique, I designed and had built a large sensitive drag balance to directly measure the drag on a panel model. The direct measurement did not show the reduction in drag indicated by the hot wire probe. After much analysis, I found the cause of the problem. Reynolds stress drag measurements are proxy drag measurements. They are related to the drag under some special assumptions. In this case the critical assumption was that all of the vertical motion energy in the boundary layer was due to turbulent flow, and stayed in the boundary layer. It turned out that with a compliant wall, the flexible wall motion induced a vertical motion pressure wave that propagated as sound out of the boundary layer, so the required assumption was not valid.
Many of the other favorable tests had used the same type of probe with the same false positive result. Those that did not use the hot wire often gave results that were confusing, but many still passed peer review. It appears that since some of the other papers seemed to be getting the correct answers, they assumed they had it also and selectively chose the more favorable parts of their results for publication. The idea was they knew what the answer had to be so they just had to be selective in choosing data so they would be with the successful group.
When my paper was published showing the cause of the errors in the results, almost all of the compliant wall drag reduction efforts ended.
It appears to me that the AGW supporters are on a similar bandwagon. Initially the observation that the temperature had significantly increased (1 degree F) over the last 150 years, and the atmospheric CO2 content had increased significantly (over 30%) in the same period suggested a relationship. Previous analysis had indicated that CO2 is a greenhouse gas, and could cause some temperature increase. The effort to show that the temperature rise rate and level was unusual was pushed by several scientists, and initially it looked as if this might be true. After many other scientists jumped on the bandwagon, because they assumed the hypothesis was valid, they then had a personal stake (funding, prestige) to support the claims. Later data and analysis seemed to show the temperature increase was a local period of global warming (GW), but did not have a dominant component of AGW. This threatened the whole structure of the AGW position, and as is typical when a paradigm shift threatens to occur, was met with considerable resistance. This was further exasperated by the political and social involvement that had started a major drive to cut man made greenhouse gas production, and at a very high cost to society.
Recent efforts by even well respected authors to publish papers that falsify claims of AGW have had great difficulty passing peer review, apparently because they went against the current paradigm, not because they were shown to be in error. The misuse of claims of support for AGW by “peer reviewed publications”, and rejection of analysis because it is not peer reviewed is often used to try to discredit the opposition. This is not how science is done. The facts should speak for themselves.
New technology is changing some of the older paradigms of communication, and use of electronic media now allows very rapid and sometimes real time interaction discussing material and presenting results. Paul Coppin stated in a response to a recent blog “Technical blogs are rapidly replacing the “letters” sections of formal journals as the place where a public airing of a journal article or topic takes place. The very nature of blogs also provides a means for the lay populace to look inside both the science and the process of the science. In the past this has been generally closed to common scrutiny. In particular, science writers have access to information and opinion they had to previous dig out or interview for”. This crossover of discussion will surely increase. Since the purpose of science is (supposed) to advance knowledge, the real time open interaction may actually do a better job than the normal peer review process for much (but not all) material. Formal publications still should be made when possible for unique data or concepts, but discussions of the consequences are probably better done in the electronic medium. One major purpose of the formal publication is getting the author credit for the material. There is not presently an equivalent way to archive and give credit for a publication on the web. There needs to be such a structure made available. The draft of a publication may be put on a web site and left open for review and comments. This might actually be a more effective peer review process for some types of papers since a wider audience has access to look for flaws. This may be done in conjunction with some formal peer reviewing. The corrected final paper then may be archived, but with continuous access to later catch overlooked problems.