Staticulation on rainfall from the BOM

The out­look for ‘excep­tion­al­ly low rainfall’—that is, drought in its usu­al meaning—in the BOM’s own mod­el­ing for South­ern Aus­tralia has a ‘cen­tral ten­den­cy’ of no change or slight­ly few­er droughts (see Sec­tion 5.2.2 quot­ed here). The mod­eled results show increased drought (‘excep­tion­al­ly low rain­fall’) only in the ‘tail’ of the dis­tri­b­u­tion: below the 10th per­centile of results.

What that means, exact­ly, is that increas­ing drought is com­plete­ly untyp­i­cal of any of the mod­el results for South­ern Aus­tralia because it occurs in few­er than one-tenth of all mod­eled out­comes. Sta­tis­tics gives no war­rant to char­ac­ter­ize the ‘typ­i­cal’ results of the tail. They’re in the tail of the dis­tri­b­u­tion pre­cise­ly because they’re a-typ­i­cal.

2 Comments

  • David (not Eric) Jones was one of the 11 authors of the CSIRO/BoM Drought Excep­tion­al Cir­cum­stances Report (DECR). So far as can be seen from the pub­lic record, he and the BoM played no part in the sub­se­quent devel­op­ments. There is no evi­dence that the CSIRO checked with BoM before ini­tial­ly deny­ing David Stock­well access to the data sup­port­ing the mod­el­ing in the DECR, or before CSIRO sub­se­quent­ly reversed that stance. We don’t know whether David Jones and the oth­er BoM authors were con­sult­ed about the sim­pli­fi­ca­tion (dumb­ing down?) of the draft DECR in order to meet DAFF’s require­ment for it to be acces­si­ble to a non-pro­fes­sion­al read­er­ship.

    The use of the DECR in pol­i­cy­mak­ing in Aus­tralia with­out due dili­gence has now become one of the exam­ples cit­ed in the study “Check the Num­bers: The Case for Due Dili­gence in Pol­i­cy For­ma­tion” by Bruce McCul­lough (Pro­fes­sor of Deci­sion Sci­ences at Drex­el Uni­ver­si­ty in Philadel­phia) and Ross McK­itrick (Asso­ciate Pro­fes­sor of Eco­nom­ics at the Uni­ver­si­ty of Guelph), which has just been pub­lished by Canada’s Fras­er Insti­tute (avail­able at http://www.fraserinstitute.org/researchandpublications/publications/6511.aspx : the dis­cus­sion of ‘Droughts in Aus­tralia’ is at pps. 25–26).

    The review of drought assis­tance in Aus­tralia is pro­ceed­ing in the light of the assess­ment in this and two oth­er reports to Gov­ern­ment. It is to be hoped that the cri­tiques pub­lished on David Stockwell’s blog will now receive some fur­ther inter­na­tion­al atten­tion from experts at Cli­mate Audit.

  • Damn… Thank you for cor­rect­ing that slip Ian, and my apolo­gies to Dr Eric Jones (a for­mer colleague)whose name some­how crept into my head while I was writ­ing this.

    And thanks for the link to the Fras­er Insti­tute doc­u­ment. It looks like an intrigu­ing sum­ma­ry of a prob­lem that I see hap­pen­ing quite a deal in Aus­tralian pub­lic health pol­i­cy, too. The ‘noble-cause’ cor­rup­tion prob­lem that Ayns­ley Kel­low has writ­ten about seems to me to be spread­ing to the man­age­ment of ‘risk fac­tors’, helped in part by the obscu­ri­ty of our very pal­try nation­al health data. Not only is the nation­al data (based on ‘self-report­ed’ health) of low qual­i­ty, but some of the pri­vate­ly col­lect­ed data (such as the 2001 and 2005 sur­veys by the mod­est­ly named “Inter­na­tion­al Dia­betes Insti­tute”) are being held close­ly and selec­tive­ly pub­lished.

    The IDI’s data, for exam­ple, is being used to pro­mote the idea that there is an ‘epi­dem­ic’ of obe­si­ty and type II dia­betes in Aus­tralia. But they have not pub­lished basic sta­tis­tics from their sur­veys, so we don’t know impor­tant facts about the dis­tri­b­u­tion (of Body Mass Indices, dia­betes) in the pop­u­la­tion such as the cen­tral ten­den­cy (they pro­vide a sta­tis­tic rep­re­sent­ing the change in the pop­u­la­tion mean but only age-strat­i­fied data for the rest) or, per­haps more impor­tant, changes in the vari­ance of the dis­tri­b­u­tion over time (which may be respon­si­ble for increased pro­por­tions in the upper tail of the dis­tri­b­u­tion).

Leave a Reply

Your email is never shared.Required fields are marked *