ࡱ> 87 \p Sean Wallis Ba==t@48X@"1Arial1Arial1Arial1Arial1Symbol1Arial1Arial1Arial1Arial1Arial1Arial1Arial1 Arial1Arial1Arial1Arial1Arial1.Times New Roman""#,##0;\-""#,##0""#,##0;[Red]\-""#,##0""#,##0.00;\-""#,##0.00#""#,##0.00;[Red]\-""#,##0.005*0_-""* #,##0_-;\-""* #,##0_-;_-""* "-"_-;_-@_-,)'_-* #,##0_-;\-* #,##0_-;_-* "-"_-;_-@_-=,8_-""* #,##0.00_-;\-""* #,##0.00_-;_-""* "-"??_-;_-@_-4+/_-* #,##0.00_-;\-* #,##0.00_-;_-* "-"??_-;_-@_-"Yes";"Yes";"No""True";"True";"False""On";"On";"Off"],[$ -2]\ #,##0.00_);[Red]\([$ -2]\ #,##0.00\) 0.0000 0.00000                + ) , *          $@ @  $@  @ @  (@  @  l@ @+  h@ @+  (@ @  h@ +  h@ +  (@   ,@ @   ,@          $@     ,@@  $ @  ,@  $  $   H3     L3   H+    l@@+  l @+  l@ +  l +  l@ +  ,@@  , @  ,@  ,  (@  , `single intervalA'NW test with error,ZR3  @@   lLpNp' PopulationSampleobservedzlowerhigher scale factorcritical valuesize!Wilson score interval calculation<http://www.ucl.ac.uk/english-usage/statspapers/z-squared.pdfsIt employs the Wilson score interval to compute the interval, but adjusts it by employing a modified sample size N.qrCommentsThis calculator obtains a scaled confidence interval for a population based on a subsample where the sample is a credible proportion of a finite population.a ffreqobs 1obs 2 Subsamplee'zaw-w+totalprobWilsony-y+width ObservationNewcombe-Wilsonp1 - p2W-W+test differenceintervalModified observationy+(err) = Wallis (2013), z-squared: The origin and use of c. JQL 20:4. 012 38< See alsop(err) p(err) x f MThe upper bound of the error interval is combined with the lower bound of p1.JLThe error mean, p(err), is transferred to the observation according to the worst-case scenario (i.e. observed errors are indeed errors).Often, it is impractical to review every case to check whether it has been misclassified or not, but it is possible to review a sample of those cases.fWe employ the Wilson score interval to estimate the interval range for a given a for all observations.OP@See http://corplingstats.wordpress.com/2014/04/10/imperfect-dataIf misclassified cases in observation 1 should be in observation 2, then the same operation is applied to the upper bound of p2.}1 set this cell to zero if errors do not transfer scale factor- p1 > p2 so don't worry about other interval Assumes p1 > p2%NB. If p1 < p2, swap obs 1 and obs 2.INewcombe-Wilson test, with a modified subsample following manual review 2-tailed1-tailedlThis calculator applies the finite population correction to a manually-reviewed subsample of a corpus query.BWilson score interval with finite population correction adjustmentf.p.c.Since the subsample is a sample of a finite query, the confidence interval of the subsample may be reduced by a finite population correction. n' = n/nnn = (N-n)/(N-1)   adjusted n z/n'z/Nn = (N-n)/(N-1)  adjusted n l Before computing the interval we calculate n = (N-n)/(N-1) (where n is the sample size, N population size).+01258;CDYZFinite population correctionoWe then divide n by n to obtain a new, increased n' which is substituted into the Wilson formula in place of n.03mnJ  F dDt@cqbx ϰ0Yxr< "-0yzl34S00P"0g Dk4l0,0`KT0l3>n0tIr0@@hxhhxhS0t00^n00  00Ԣ0   70  tj0 HPercent0] Ν02B_T0 Q0_T0#)!T00!0ߐ|n|<bT0|(3`T0`T0@X|0ShS~0BT0xSSu0(1T0|0SZ0S0HZS\0ZHTd7> TL hcc  !&  dMbP?_*+%MbCutePDF WriterS 4dXXA4PRIV0''''P4(u^mSMTJCutePDF WriterResolution600dpiPageSizeLetterPageRegionLeadingEdgeInputSlotOnlyOne" dXX??U} }    ;T0   , , , ,         ;D l0      &>     B    E   @pp@! Q0Bd_? DD1#c3V?DDDK@O%X?5D DDDD AD'=Wz׍? D D D  !~ "b@# !$%{@ DD!a|@`@ DD!/g(q? DD !%v,3{@ DD!Q;d@ DD!/`-@  +g,@A  dMbP?_*+%MbCutePDF WriterS 4dXXA4PRIV0''''P4(u^mSMTJCutePDF WriterResolution600dpiPageSizeLetterPageRegionLeadingEdgeInputSlotOnlyOne" dXX??U} $ } } m +;T0;,,    D l00|T|LSw; &:& (8 ~  @ ' &\? D A(" ,@ DD  ( ! )"(? DD! )mͱV~א? DD * +-! *? DD (? ! )Rj? D D ! )kz? D D  ( *** ,C0 -? DDDA 6 ( *** .A% /|>f@ DD  ?D ( F! *\./9?  D D! *m~8?  D D * +E! *? D D ! *^h#?  D D  F! *lEq͗9? D D ! *4RR8? D D  ( 1 *7;}—?D D D 1 *qĉ?D D D  * +1 *X4w?D D D 1 *xF ?D D D  1 *ߵ? D D D 1 *z,)? D D D ( K0)Ih?5D D D D DAD K0o`d?5D D D D DAD * +K0D„?5D D D D D AD K0t7py+?5D D D D D AD   K 0f)$g? 5D D D D D AD K 0cLe? 5D D D D D AD  ( !*7 ge? D D!* 3? D D* +!*(><`? D D!*F^? D D  ! *C¤? D D ! *kEx? D D ( !*7m7˚? D D!*~fד? D D* +!*ܾ? D D!*"JIT? D D  ! *LW{z? D D ! *w;9=ɔ? D D  ( !*XSf? D D!*ځ&b?  D D* +!* i"?  D D!*yD?  D D  ! *H 0f? D D ! *dG˅c? D D ( !0(rZMi? DD !0e? DD * +!0 ͥ?  DD !0? DD   ! 0!?a>i? D D ! 0~Άf? D D (( ; <  2*! 3ZS}`lP?  DD  3ZS}`lP? D (  , *K/g? D D A~  * 7(  ~  *, *UQ@g? D D A( ("( (' #!*jDz?  D D   #! *d^`q? D D  (( $-*q"oDDA ( %,*2}—o?DDA  %, *op? D D A (&[((U EDD  s-,DD  s+ ns"Bs+  (&[ (0U ED D  s-,D D  s+ ns"Bs+ & =@D *_~| j_|   xdV tW!T0"#$%&(* !1 "2 #@ $0 %/ &4 ((9 *3(  p  6NMM?`K,]`S  "??3`  `  )3d23 M NM4 3Q:  ObservationQ ; Q ;Q3_ 3:   3:d4E4 3QQ ;QQ3_ O NM  3:d4J[(rZMi?4 3QQ ;QQ3_ O NM  3:d4J[XSf?4 3Q: ,Modified observationQ ; QQ3_4E4 3QQ ; QQ3_ O   MM<4J[4 3QQ ; QQ3_ O   MM<4J[K/g?4D$% M 3O&Q4$% M 3O&Q4FA2 3Ohu3 b#M43*{Gz?N#M! 43" 0 3O/ o% M3O&Q423 M NM44444ee(rZMi?XSf?K/g?e?ځ&b?UQ@g?e >@ Oh+'0@H\p  Sean Wallis Sean WallisMicrosoft Excel@1@"1՜.+,D՜.+,( PXd lt| UCL  single intervalNW test with error  Worksheets 6> _PID_GUIDAN{80BE0A10-5217-47EC-A800-1AB14F5AC311}  !"#$%&()*+,-.0123456Root Entry FWorkbookGLSummaryInformation('DocumentSummaryInformation8/