A Bayesian Approach to the Paleomagnetic Conglomerate Test

Abstract

The conglomerate test has served the paleomagnetic community for over 60 years as a means to detect remagnetizations. The test states that if a suite of clasts within a bed have uniformly random paleomagnetic directions, then the conglomerate cannot have experienced a pervasive event that remagnetized the clasts in the same direction. The current form of the conglomerate test is based on null hypothesis testing, which results in a binary “pass” (uniformly random directions) or “fail” (nonrandom directions) outcome. We have recast the conglomerate test in a Bayesian framework with the aim of providing more information concerning the level of support a given data set provides for a hypothesis of uniformly random paleomagnetic directions. Using this approach, we place the conglomerate test in a fully probabilistic framework that allows for inconclusive results when insufficient information is available to draw firm conclusions concerning the randomness or nonrandomness of directions. With our method, sample sets larger than those typically employed in paleomagnetism may be required to achieve strong support for a hypothesis of random directions. Given the potentially detrimental effect of unrecognized remagnetizations on paleomagnetic reconstructions, it is important to provide a means to draw statistically robust data‐driven inferences. Our Bayesian analysis provides a means to do this for the conglomerate test.

High-Resolution Late Devonian Magnetostratigraphy From the Canning Basin, Western Australia: A Re-Evaluation

“…The hightemperature components do not show as clear a trend, though clusters can be identified in the upper and lower hemispheres (Figure 5). A paleomagnetic conglomerate test was applied to all the tilt-corrected high-temperature directions using the Bayesian approach described by Heslop and Roberts (2018). When the high-temperature lower-hemisphere components were flipped (adding 180 °to the declination and reversing the sign of the inclination), the Bayes Factor and p(H A |R) value were both 2.58 × 10 −28 .…”

Paleomagnetism indicates that primary magnetite in zircon records a strong Hadean geodynamo

“…2C). The high-unblocking temperature data correspond with a uniform (random) distribution with positive support in the Bayesian approach of Heslop and Roberts (33), whereas no viable data clusters that might indicate partial magnetic overprinting are seen using the approach of Bono et al (24) which searches for clusters in data that might represent partial magnetic overprinting (SI Appendix).…”

An Initial Collision of India and Asia in the Equatorial Humid Belt

“…The value ρ obs = 0.001 is less than the = 0.190 ( N = 25) and indicates no preferred direction in the conglomeratic clasts. Heslop and Roberts (2018) developed a Bayesian approach to the conglomerate test that is more nuanced than Shipunov et al. (1998).…”

“…The value ρ obs = 0.001 is less than the  crit = 0.190 (N = 25) and indicates no preferred direction in the conglomeratic clasts. Heslop and Roberts (2018) developed a Bayesian approach to the conglomerate test that is more nuanced than Shipunov et al (1998). Our conglomerate test yields a Bayes Factor (B = 6.058) and p = 0.858 which yields uniform positive support for the conglomerate test (see Table 1 of Heslop & Roberts, 2018).…”