I would suggest that you please keep in mind the concept of the metamorphic array when considering what constitutes a "homogenous population".
Julie Vry
On 1/10/09 1:37 PM, "Félix Gervais" <[log in to unmask]> wrote:
Dear colleagues,
In geochronology, a common practice is to calculate the "weighted mean average" of a group of data It may be because of my young age, but I have not encountered any such statistical treatment in thermobarometric studies. Wouldn't it make sense to use such approach for different samples from a homogeneous population? This could potentially greatly reduce the large uncerntainties generally associated with thermobarometry. Obviously, one would have to conduct a rigorous statiscal analysis of the data by using the MSWD and probability of fit. An easy way to do this for geologist like me who are not virtuose of statistics, would be to use the Isoplot program of Ludwig. This is a widely used program in geochronology, but is basically a statistical solver. If a set of analyses came out with a very low probabilty of fit of, let's say below 0.05, then the data set could be judged inhomogeneous and discarded (or reduced in homogeneous subgroup). Alternatively, if the dataset has a high probability of fit then it is likely that the data set is homogeneous the number obtained geologically relevant.
Is anybody aware of such approach having been applied? Do you think it would be sound in the first place?
Best regards to all,
Félix Gervais
Post-doctoral fellow
McGill University
Montréal, Qc Canada
________________________________
Devenez un meilleur amigo grâce à Yahoo! Courriel <http://cf.promos.yahoo.com/courriel/visiteguidee2/><http://cf.promos.yahoo.com/courriel/visiteguidee2/>
|