ASA: Bias in Science, Part 2

Randy Isaac Wrote:

In part 1 I spoke of prejudicial bias, in which there is a tendency for a prejudice, or an a priori desire or preference for a particular result, to influence the analysis and the outcome of a scientific investigation, and a scientific bias, in which there is a tendency for anomalous results, namely those not expected on the basis of established scientific knowledge, to be rejected, particularly if the results directly contradict previously well-documented results.

In this post, I’d like to take a closer look at Baumgardner’s paper[…]mgardner.pdf which elicited the concern a few weeks ago that it might be rejected by peer-reviewers due to an inappropriate bias. Specifically, Vernon Jenkins wrote on April 4, 2005: “There can be little doubt that Baumgardner et al would be more than happy to publish these findings in peer-reviewed form if a relevant journal could be found to accept their work. However, the sad truth is that a paper challenging the accepted uniformitarian paradigm - irrespective of its intrinsic quality - invariably meets with editorial and reviewer hostility.”

Read more at Bias in Science, Part 2, on the Calvin Reflector


Syntax Error: mismatched tag at line 11, column 2, byte 821 at /usr/local/lib/perl5/site_perl/mach/5.18/XML/ line 187.

I can’t reply to that post on the reflector, but IMHO he made one pretty serious error.

… there is no allusion to why this method is adequate to remove any contamination. The only concession made to this problem is that a value of 0.077ppc is subtracted from the data “believed to be from sample process contamination.” But there is no discussion of how this value is determined and why it can justifiably be considered to be constant for all measurements. This is a very large value, ranging from 14% to 44% of the measured values in Table 2. If the subtracted background is so large, it must be justified and the relevant error analysis must be applied. But no such discussion is present in the paper. This is not acceptable scientific methodology.

What he’s reviewing is a poster, rather than a full-bore paper. You just can’t get everything on a poster. Yes, the reasons for using that value of background should have been discussed, if only as a pointer to further information; but it’s not that big a deal. However, if we look at the full “paper” presented at the International Conference on Creationism 2003, MEASURABLE 14C IN FOSSILIZED ORGANIC MATERIALS: CONFIRMING THE YOUNG EARTH CREATION-FLOOD MODEL, there are four lengthy paragraphs discussing the derivation of that value:

Table 2 summarizes the results from ten coal samples prepared by our RATE team and analyzed by one of the foremost AMS laboratories in the world. These measurements were performed using the laboratory’s ‘high precision’ procedures which involved four runs on each sample, the results of which were combined as a weighted average and then reduced by 0.077±0.005 pmc to account for a ‘standard background’ of contamination believed to be introduced by sample processing. This standard background value is obtained by measuring the 14C in a purified natural gas. Subtraction of this background value is justified by the assumption that it must represent contamination.

The background standard of this AMS laboratory is CO2 from purified natural gas that provides their background level of 0.077±0.005 pmc. This same laboratory obtains values of 0.076±0.009 pmc and 0.071±0.009 pmc, respectively, for Carrara Marble (IAEA Standard Radiocarbon Reference Material C1) and optical-grade calcite from Island {sic - see note} spar. They claim this is one of the lowest background levels quoted among AMS labs, and they attribute this low background to their special graphitization technique. They emphasize backgrounds this low cannot be realized with any statistical significance through only one or two measurements, but many measurements are required to obtain a robust determination.

The laboratory has carefully studied the sources of error within its AMS hardware, and regular tests are performed to ensure these remain small. According to these studies, errors in the spectrometer are very low and usually below the detection limit since the spectrometer is energy dispersive and identifies the ion species by energy loss. The detector electronic noise, the mass spectrometric inferences (the E/q and mE/q2 ambiguities), and the cross contamination all contribute less than 0.0004 pmc to the background. Ion source contamination as a result of previous samples (ion source memory) is a finite contribution because 50-80% of all sputtered carbon atoms are not extracted as carbon ions and are therefore dumped into the ion source region. To limit this ion source memory effect, the ion source is cleaned every two weeks and critical parts are thrown away. This keeps the ion source contamination at approximately 0.0025 pmc for the duration of a two-week run. Regular spot checks of these contributions are performed with a zone-refined, reactor-grade graphite sample (measuring 14C/12C ratios) and blank aluminum target pellets (measuring 14C only).

The laboratory claims most of their quoted system background arises from sample processing. This processing involves combustion (or hydrolysis in the case of carbonate samples), acetylene synthesis, and graphitization. Yet careful and repeated analysis of their methods over more than fifteen years have convinced them that very little contamination is associated with the combustion or hydrolysis procedures and almost none with their electrical dissociation graphitization process. By elimination they conclude that the acetylene synthesis must contribute almost all of the system background. But they can provide little tangible evidence it actually does. Our assessment from the information we have is that the system background arises primarily from 14C intrinsic to the background standards themselves. The values we report in Table 2 and Figure 3 nevertheless include the subtraction of the laboratory’s standard background. In any case, the measured 14C/C values are notably above their background value.

Of course, that doesn’t obviate the fact that their “science” is crap, but there’s no need to accuse them of errors they didn’t make. There’s plenty of errors that they did make.


Note: The authors use the correct “Iceland spar” in other papers and other parts of this paper, so the typo is obviously just that.

There was an interesting comment sent to the TalkOrigins feedback by Craig Morton which will be in the May Feedback.

He wrote:

Hi T.O, some further references for the section on 14C in coal. I had cause to look into this myself (prompted by an email discussion) and also contacted Prof. Gove. Some of their work has been published and more has been submitted and it may be worth adding to the 14C in coal discussion page.

R.P. Beukens, H.E. Gove, A.E. Litherland, W.E. Kieser, X.-L. Zhao. (2004) “The old carbon project: how old is old?”, Nuc. Instr. Methods. Phys. Res. B 223-224: 333-338

They reference some work to be submitted in this article ([5] X.-L. Zhao, A.E. Litherland, H.E. Gove, W.E. Kieser, to be submitted.) and a paper I found particularly useful:

G. Alimonti et al. “Measurement of the 14C abundance in a low-background liquid scintillator” (1998) Phys. Lett. B 422:349-358

This has a section on the formation of 14C in petroleum, ranking the causes (with references) as:

1. 17O n,a 14C

2. 14N n,p 14C

3. 13C n,g 14C

4. 11B a,n 14C

5. direct 14C emission from tripartition of 226Ra

They specifically indicate that cause 4 (alpha transition of 11B to 14C) is dependent on the level of boron in the material. Boron is, of course, quite common in many coals and forms a large part of the fly ash at power stations.

About this Entry

This page contains a single entry by PvM published on May 30, 2005 8:15 PM.

Privileged Planet: Law and chance eliminated? was the previous entry in this blog.

Teach the uhh… our controversy is the next entry in this blog.

Find recent content on the main index or look in the archives to find all content.



Author Archives

Powered by Movable Type 4.381

Site Meter