|
J. \~Vildl. Manage. 63(.3):1999
<br />a very logical way to make decisions in the fare
<br />of uncertainty It allows for incorporating be-
<br />liefs, data, and the gains or losses expected from
<br />possible consequences of decisions. See Wolf-
<br />son et al. (1996) and Ellison (1996) for recent
<br />overviews of Bayesian tnethads with an ecolog-
<br />ical orientation.
<br />STATISTICAL SIGNIFICANCE TESTING • Johnson 771
<br />BAt:,~N, D. 1966. The test of significance in psycholog-
<br />ical research. Psychological B+dletin 66:4?3-117.
<br />BARNAx~, G. 1998. Pooling probabilities. iVew Sci-
<br />entist 157:47.
<br />BAUERNFEIND, R. H. 1968. The need for replication •
<br />in educational research. Phi Delta ~Kappan 50:
<br />126-128.
<br />BAYES, T. 1 ~ 6,i. An essay toward solving a problem in -
<br />the doctrine of chances. Philosophical Transac-
<br />tions of the Royal Society, London 13:370-418.
<br />BERGER, J. O. 1985. Statistical decision theory and
<br />Bayesian analysis. Springer-Verlag, Berlin, Ger-
<br />man..
<br />, ~~xD D. A. BERRY. 1988. Statistical analysis
<br />and illusion of objectivity. American Scientist~76:
<br />159-165.
<br />\D VL DEC.A~IPADI•. 1987. Testing precise
<br />hypotheses. Statistical Science 3:317352.
<br />• AxD T. SELLI:E. 1987. Testing a point null
<br />hypothesis: the irreconcilabiliri• of P values and
<br />ew~dence. Journal of the American Statistical As-
<br />sociation 52:112-12_'.
<br />BExt;50~. J~ 193S. Some ditficulties of interpretation
<br />enciittntered in the app(ic;ttion of the chi-square
<br />test. journal of the A+nerican Statistical rlssocia-
<br />tiou :33:36-•54•'_.
<br />Box. C. E. P. 19S0. Sampling and Baves~ inference in
<br />:cieutific• modelling and rob+tstness. Journal of
<br />the Rowal Statistic:u Socirh• 1-1:i::i8~>-430.
<br />. ~xn (:. C. Tta(,. 193. Bayesian inference in
<br />statistical :u,alwsis. Addison-\Vesle•:; Reading,
<br />Massachusetts. USA. ~ -
<br />Bt cla_\xn. S. T., 1:. P Bt~xxn:t». Axe \. H.:\t;-
<br />CC~TI\. 199 ~locie( selection: au integrated part
<br />o(intrrrncr. Bi+~n,rttic•s5:~:riU:i-(il•5.
<br />Bt-r,~na~t. I:. P...~xo D. R.:\~nt:xsox. 1995. Model
<br />selection ;uul inference: a prtctical intirrmation-
<br />thet~retic ahprnac•h. Shnrr r-\~rrlag. Vrw Yorl:.
<br />~irwr 1"rn'k. 1.~$iw.
<br />t:.t>I~'htil.~.. \1. L9y^_. Cttuhdrncr intrr~als. Rowal 5ht-
<br />tistical 5uc•ich' \rwws ;u,d \nh-s !S:~-•5.
<br />(:.trr.r~. li. I'.. ~~r, T..\. L.ut~t~. 1996. Baces and ru+-
<br />hirical H;n;•s me•tht+ds (nr datu ;urtlesis. (:hapm;ut
<br />~\ I fall. London, l nitetl him~dont.
<br />(:,\r;\"ha;, li. l'. 19 ti. Thr case a~aiust statistical sil~-
<br />nilic;uicr U-stingy. Ilan:o'd tdnc:+tiunal Rrvirwc
<br />t•ti:: i-~-: 1951.
<br />(:I..~t:F. (:. .\. 19(1:1. Ilwixrthrsis tr_stiu>; in rrl;ttion to
<br />statistical nu•thudnio~n. Rrw~irwv of F:ducatiun:d
<br />(:uru<\. I. 14)ti~ti. St+tishctl ltuwcrr :uralcsis (nr the l+r-
<br />haw inch x•icncrs. Src•ond edition. Ltwvrencr Erl-
<br />h:uun .\sstK•iatcs. Ilillsd;tlr, \r•.v jersey. U5;\.
<br />IS)si-{. I'hr e:u'th is round . )~ < .0.~1. Amrrir;u+
<br />Pswch++ltn~ist "5):5)9 -11111:3.
<br />D-UTU~. l'. ti. L95)ti. Reversal of the burden uf• prooF
<br />in fi.aLcrirs n+:ucn~rn+rnt. Scienc•r_'79:521 S2-'_.
<br />DICGR(>tPr. \I. I[. l9~(1. Optincd statistical decisions.
<br />MrCrawc-Fill, ~'rwc York. ~iewv York, USA.
<br />Dr;>n~c, \1'. E. 195. (h, prohahility :a a basis li,r
<br />action. Anucric:u+ Shttistician 39:1[6-1.52.
<br />Et.l.rsci~. :\. ~1. 199fi. An introduction to B;n'esi:u+
<br />inl~errnc•e lire ecolr>,gical research and environ-
<br />mental decision-makim~. Ea,lo;~ical Applications
<br />fi:10:i[i- LO-tfi.
<br />Cr:xAxr~. P. ll.. U. R. 5>IITH, :\~D C. WEER:\6liODY.
<br />~ CONCLUSIONS
<br />Editors of scientific journals, along with the
<br />referees they rely on, are really the arbiters of
<br />scientific practice. They need to understand
<br />how statistical methods can be used to reach
<br />sound conclusions from data that have been
<br />gathered. It is not sufficient to insist that au-
<br />• thors use statistical methods-the methods
<br />must be appropriate to the application. The
<br />' most common and flagrant misuse of statistics,
<br />in my view, is the testing of hypotheses, espe-
<br />cially the vast majorih~ of them known before-
<br />hand to be false.
<br />With the hundreds of articles already pub-
<br />lished that derv- the use of statistical hypothesis
<br />testing, I w•as somewhat hesitant about writing
<br />another. It ront:uns nothing ne~v. Blit still, read-
<br />ing The ~r»irnal of [Vilc~life Vla»a~erne~nt makes
<br />me realize that the message has not really
<br />reached the at.ulieuce of wildlife bioiogists. Our
<br />work is important. su we should use the best
<br />tools we have available. Rareh, how'rver. is that
<br />tool statistical ii~puthesis testing.
<br />ACKiVOWLEDGME~ITS
<br />\V. L. Thompson and C. :~. Rihic deserve
<br />thanks Coe or~ranizin~r the svmposimn that
<br />prompted this article. I appreciate the enconr-
<br />a,rment and cr»nments un the mamiscript pn>-
<br />~idrd bt~ D. R. Audrnon. J. O. Brrgrr. D. L.
<br />Larson: til. R. ~iester. \V. 1/. Nrv~~ton. T. L. Shaf-
<br />fer, S. L. Sberil(•. B. Tbuntpson. curl C. G \Vltite,
<br />who nrntetlteless reut:tin hlauteless fin- any mis-
<br />iutrrpretaticnts er+ntainrcl herein. B. R. Ltrliss as-
<br />sistr..d with the prr~par.ttion of the, manuscript.
<br />LITERATURE CITED
<br />:\BFl.ut~, ti. P 199. _~ rrtrus(trc•tiwr on the signi6-
<br />c;u+ce test h:u+ ou 1999 IIf there wvrrr no signif-
<br />ic•:utcr tests. thrw~ would he invrntrdl. Pages 11~-
<br />l-t l in L. L. [ [:u-low, S. A. ~Inlaik, and J. 11.
<br />Stri,~er, editors. 4\'Itat if there weer nu si~nifi-
<br />c;u+cr tests:' Lawvrrnc•r Erlb;unu Associates. ~luh-
<br />wvah, \rw Ic•ncw'. USr\.
<br />\~scr,~,ntr:. F. J. 195(1. Discussion on Dr. Dawid•s and
<br />Ur. luhnsons Paper. lourn:tl +,F the Royal Statis-
<br />tical Sucieh lS:°_~-'_i.
<br />
|