?_4($06? %08>BLU_ek  )15<@$+0;AHP\ FOv WZ^c9 A D J M P V [ ^      & ) 0 6 = A J Q Y _ d p y    # . 7 A G N R [ _ i I N S s ~ " + 1 9 B G U Z ^  #>j #+U_gpt{ $,7GQZbhntz #(/6=BGLUag5<BGQ[agkry#+1:Hcltz  (16fkqy%.6=CGMPTYv{"both""d@efaultp = @ (as interaction.plot), "swiXtch, 0  orienta) of x.f6or and" 0""full""ordered"2"ucQ@"test"# Th e funu i@s curr[lyined as# re(19@91,(Pos,(Wiley)(between(ci0(@ m?(0s(first(lm)(Hlsd.@opal(positive)(taken-axis.0 F0levels are!tbelow t margiH means  effects.. FB }, ls faMvalLueBC...),> #> att"(catalyst)> .>  s(>ALSOARGUMENTSAysisBarsB'1Cat)Chaer @stringlength 2 h nam teauQ6fit d ?Qp!TA+@).ConDESCRIPTIONDE@TAILSD2Diagnosti2ci2to_}powns<m@removeEXAMPLESEmersonEr rorExra y11sFisher'sFlsGet A "beside u(.GroupHalf-NalHoag,ICs PNR!Responsewi1-   P LDea9Squls, MR0s@mLars)LpbarM"jm"P.a@McellerMiple@`S Z0 by &`Q-Q 0O"@1pst1O!AL bPRINTER")) i pm sREFERENCESREQUIR ED>pDQSEE See+AStpard !"Q cg!}#Thi@sTukey /rMel)[[UxNIXL5piUSAGEVALUEVarianceWARNINGWxinY ieldac5will` creaakshowmrequ>:eO["_$_Dac[acyaddadjall-nd/approdraw  di*ylsoaheyymoiiHnguZd.u!Xsas..fO>avg@ebackghnFd(s)sedbekk3dboigb#chcalca%cerch   %4F;widthcircleclassco ic llis!HolummbcY ommpute samKsize:9:nf? tainwcor.grelK', ctovB,C2e%movH@;h!EP@tx@/VBcrrW%Ǩ! pwhi%(pr) b5&obj& P7Rom*&if omi>)"bbb4YHo@wu0 P} ,rr(1O9HdegreesG|h la"7m guq1@1p|g/b3 ,KediQ7 X &!)dev s f.&cdigqdotlna ICguspecifi (9[(23'7yMip|A t49t @d1&IeleD@eninviron 38 imexaexperi7:J;?YǪ JwhoIiUse(sZ+099XXPsQ:1f y*& (. :G3#SZ ungse-cf`f&B )S%f|domcvO[gXA))≼;).: g!tviewg g@s8 X0/,%bt% %,| mon 'U;n"r v, o rP.Fheigh\ rehp z96lItpl;sfG Mtead ,ьܟwdk, jr3;le ,=0.05,, n*X rsxz.04te(M=r-0 W!/ U:%c*(EuJg n ' w 1rusr.tcBOrW"."NA)VO8C !-,j,(1if@xl0@;Library @ pda# ! AHؙ? a)p% E`Spiq!ap6EopgUp}pz3, !`tGpĆoӌ aa hn`.covarsT0factoprediU0fA@burst 0YRc~\R% f ke s xUy "~$& `ZUVD b pSer `xNzV- (b.0rTC b~'*r38.dft%g~2Brr>kmrt:pҚ0m1.labelU"RG `[`!8E=QlpqPlp!.n"""`@ .ploI2in Model_R.res ponseR `U&0Z2 eC4"tLi-46;t &,W40ʏtV6tM4<{H )ؑ2K4nested.treA0!e3g8^jp&1.sLizt!aISQ,%Bska#bW_3M16u). T>`)*@ $P@!~ G&d= %:'c% NF or  Lo)(  %! ;u(object, @uu-DSc;? L_ ! `3D>* e$  Qj;%;!y= +/Iq%@& .) V* e$hzzx^> O n zP#JumpId("splus.hlp", "all.name@s_t")0 .* h $d> . . *S! <- y~x+a*b,data)turn c("a","b#1fit!`"x"6;x;?FBy"K^ 3)h a 1 Ea J f58. *upu B(, a %ll^3G0 %Z`^ !, $=1=0.2^=width Q p# );c+ @(x) g P outliers, if any* $ ii?,  %Xias 0!]=F* 8 `$Wj Y ;W#!WWboxUw:a,c* 8 $c~a J ' h<m\/63Z*(,8i0,xlim=range )*c(1,01.1)J2max@1 { 1]u{ 3J *, { ڀ %v6B& %L (rooi,@ d=1: "p P_2 .k )s b6 ) ))e0  &* $kk*%bk6** %$8 >w !`e=" a1= p{if(flip\(i  (pocAi{ e (#p)}3% 1 E B 5-B*! Z! , n`% 9B(h3 5C=sample_*`),=dfS sd=s@td.dev@'4=qt($1 U/):(`n)~ ʀO$$WY<5uX= per 2 (single OK balanced)ORspп0 t distribution a!O|by J] igno* S$ik)@%) Ww !u (critical.value*(sd/f)} * S$ 7.x ,j%u?y 3AP1P$5Q*d#[,, P`%M5@%0Z@nfYJ>: m  "* t$GJn&JI is n1it 4co rr"R-1. = EP B0 (e} i citlySly1um up dxl: Q?#6D;,extr [wmake priate guess S'L5 [/* t`$_A'n @& N.!xZ @3"?@1?@t@C5/ @t@* $^P!~ c , ?@@% 4 /0t@@%(+`! [ b(data)^B@.A5@ : !e( lk 8Iin matrix m, perhaps 3by * @XA$  .ADB&- Ireinterpreted g. is useful postproces sing files ".""9 (e.g..SAS input) rather than NA. Column8 !any :s y0a; replacjseNA ow-toform |ly.* XAnBP$2DBC dQ $JumpId("splus.hlp", "read.table_t"4)-P'--as..fram0a, C* nB>C$dCC* "!ad <-N(tolo, her=T)$ b *0*0)$entry AfactorE@03*>CD1P8 D9DXGt5<C9DP*xqtYDD&( SgW{ch. 8C) sof kQIgo9DD& RNp(fit, =T, =F)rA=10=0.2, 0=jitj3:0xD)Fl #)p+#_uso  d u4standardize 1-inf@luence(iJ#th?mD}sidewayt]b =.1EEz)1] * DSF$/)FF%$ H* SFF$S=7FF. ,Xpӌރr* F G$`K/FXG%&lr(lm(Yield~Conc+Temp*Cat,clyst< GG1 GGK@e xs>'XGҀG*TbNT T;, GG`%]k3GH) ԡ (x=My=1E#x$sxsyrho-e|0,4̮show#:8(xVhomosked#.hsqrt(qcq(1-leve l,2))3GQJu 7(x,y or Jxsx,syN,ÓQB /correlation between x-1#H n 6HarcDd0'P0itive, c[ maaxe(([y =TRTUE̊exist([* [U(_$sy* H{J$ ""@9QJJ% rm#'H !@ (s * {JJ$8JK@+ g #!turn coonatu [1~x2=3=.5," # 95%9ntou$bivariB ; F32JK1W" KK%NRAGKKP*~ ~-8.labels 5R, K+L%7@KkL% <g(objec@S+LL? L_,)DBDM>* kLL$<6L$M%s ll! * LNM$K#$MM: DFRiqAJt: : &* NMM$bRM%N@&0 Oå"w72 arby~a+b,S-T 3*MXN18 XNNP5:%NN*"t&, XNN`%&:\VNO&(& +.!  k=get.l0%H;sSC&NOQ p$`,|  s (gerMS!  D age0$hc,a0,c("trt","hc")3P,Q1\  ,QbQS6%PbQP* Jpb5, ,QQ%$\QbQQ& b l5ps@E 4 =2  4 -QRHp ^I~ܖ@'iR[ ruXP9S,height,poPveni @I* QR$ qQIRS&%!?gh(h> "my!=.ps" RS*4;Ǯ?F0IX: ls.off ?sp aw0 itZg buttons ^ send( Z  3SS1". SSU54SS*+ hg+,, S!T%g2-SNT%J&?!TT5 :>,bbrO+O!* NTT$ 3TT%) (>har monic* T 0$cPT SIMTU* $! (x){/sum(1/}30 1.  ~5@UP*{{P!~ { ~, %@ (  F- (8ݰc(x.factor, , `:+=.plot="lsd")"PfitGCI0p=@lsmean0 data,0s),=F, apc(c ,"ci!test none")w @ X!X!- 4 ;!Pp#5T0@ _p!"IxH22@. should x be sorted by  =F)*Œ8<*XX]s a!iB 0Kato -drawing  . re some T, such as $ edge can c@hange Qaai 2!" (=defau lt), ! 0or 1la 2)@ contrion of ximate al aris&o I^ #gcA@NOTE:݁XKl@gree irder ! but n@ot HYY! y <@A [3 mE> F7dny]lin(' #''int 0)![ `s VF svoa corre@sponds![I="b1overridden. A q+s  ed* $ ńU`% !sd0ei,th a* ,$~~Qۆ ܀?R*JumpId("splus.hlp"ieB%,_tkt: PĦ%g!P6E#P/@5, S @@:* ,@$dۆ2 027 V?'mimicK@FA(ConcaVD+yougetJ fancy ?[dC" Zc B? <- aov(Yield~g+Cat*T emp,c)lyst)P.sPFV112<% `$[onlyA>^}}WPggg&=[@$0@==20,]t"rseٿ3܈1#M܈7&*LSlpq, ܈?%UU8yr' #2_ (pr#=u nix(" 0nv( }rleCjzPL ="-2rO?XH ^)0)/'k'ewr 0rs( usualPostScriptAetibleLM{* $U||X*u u)#Heck s@queueU)send "xyz"d'rLr( save pa0per!U @381M8m͏5!m@*+ B*, 80%:{m' !k (y=" u=  = .cov )'Z))~u ~%3$6H"y& Y/(musBtP^ y9 C-mr:E(foui %: !R+< #! @two eY ies. i 1*(%) a6 *x* drofpceoT (J%workV|U!& -to -zeroIsts*J u bwistandarM*!a @-l; P6 -)E * $bb&R ev2cx;!C n 2edO(se1:  !C+ty cells * $ VM jrP!i&c'n7(* =$ 7͏*8 "5S0Hy@,x1,x2mWdoekth\ > y~Xx1+&9FM8pHx1 x2d3-=1 ;;@͏;*6ww? .'7, g`%}}?UU;''(c@| O= Sgi 3N~4sn9n.N= way = ;(additi@ve) * ȑ$Q8 >V s;* ȑC$ K%: DJ1T&r#{* C0$H@'<x "xJr giy<===1'=|Model?S|P*` <  ^*<, =%k@h `e|&  (full 'kvKy J< = " ",... b~ $K*GX!(eanHa!t7 a R q ax*P}j{8~{|V cul0("|"?$ (s: q(r!=default)hs9ci" nw4(""B0X* ٕ$ Q*8$* ٕT$d0*+F \`!diqT>k mT(T+,  .plot @.* TU$  Adv+' )! +,> <- Yield ~ Zc + Cat,catalyst)$$update(reduced,~.^2)Jrgin .lm(`full,!PAp3'U1uM54M* $hP!~ ! , y% !"M ) A$_ (x, y, @:"p="l", lty=1:length(unique(group))Ftex t=F, L@yo P  ! p#@*000/px,yCoordins of bskJw'_or is ). be as x y-2="p!"b" uM7AVinB(see pa0r())q !if true, ߒ`; AfalseP[.i!y,f!,e!T) equivat to++p)[8W1q S WaPlong)6Z O(* $WVM%.#) a consisting!Pmwin Z"!* w$!"M. DPJumpId("splus.hl1s_t")O2s (point)R) QEbsEZ%* wX$.* "2#attach(iron)Cactual ,RespP2=Clevel`p ict((~5Phwk37B,$)P 32X*1$*_5A_*E%%, *ƀ% %2_%+25%m$ 3H ^g%$$SPamploaov objectPassed dirly#~(Ws n umbers\abyN p vP"#* `$}c&h l6c0s~Ifbxa , ts ;Z.` *0a half-normal [9t7* 0$7*c0@ ܀TT0!qq_"$K)p.-"I~P%)* Z`$fH0((#library(a$(1:10,letters[ ]lsfit(y~a*b,* 3,Z1u (45:(*t34*R$ \(&( kQunddo,pmttr2e xima0tion69 measure.N#md  * R`$ ( 7yted(  Cw= proj( 9)!`C@="y 6=c(1,ncol f 2 = x/w@@*$c ,cT6 5 8:ly{!\ S!h6*7$}FL }r y++==ں0# " 0 v9v!<     Q+0` 't7t9d (1=:ercep2=whoZ, 3=sub-, et|c.` eren 1Ii z4 * 7$ }-8p >g5aD/A!  s= p1 couX ca&s j!* W$^mB-1 2Q`C.default Pjl codBe5tail,*W$(Τ_KM%~putqs @, makv Qss liNOTED!.SH* w$!Mw x4j:pw!))Pkw/* w)$jW& ӯ /)U%tCc("A", %2k A*BbS7)1'$ e3C# *ͱFhRtr@0*7$C7 z% n@1ps/KorV * 7$zX(SpK(y&namHes(4nsw6=iC )`[-1]offset=0.05 2&="e="stEr=70/2  *$*XT v)-J&F,,&# 5 67+M{!n &*@$Ho 7wÐ𩼟1P  G+owab (scalarr <, nr\ZgU>B2*LaT2@z':E &f"&b>H7t|׆_EithN (") muzN 2z' P.,esR Ft0%&Jl "OLkB,* $sBc>s% >"Ao#eatfix3,bar="ci",s l 0,0,?3831g  3h5Gh* $h_ ^, 3%tpWh?& j^(...{=unix("echo ?$TERM"), =unix("echo $ SHOME =24, las=1)$Z P!~! :p#P8to@ pass  s XUNIX inal  .@name of S or Splus#sH Y(if not@ xtermrotate y axin* ?N$ 5&$% L !R+}0a* N$^^d, %L^) ^ ^^(3 1| A5)A* $R!~ c k , m%14A%,j(val ue,seah?m >@ L~`SEE/@Nsetandard ~(s)".* 3$[O %` W offk)2 p one extra ^* 3$i 6%0J75.111,.123!1{|!\;/\P*^.size% , !%(I6\&` lڒbp( x, : `fi"tCwZ 0xNR9U$itio in D4>* $ Rmpvw%mRIn cas$L@, is. Fmobjec8by * ;`$kP' list(x1,x2,x3))qP)pm(y~a+b,data)3:;1  5H*E, :%4,Pf0 .< (Hxpo&, mod = "", cap DSE adj0A&= .0e(x.factTorQb$U@<= C,=0s= Z2$s0  q54ical.%!2) * qt(1-level/2,rdfA0X100* ,"%\ nLSD"p=""P:w  0&/AS %t}7&"$ f'FK&y!% ifier(@"*"  unequszm(omitted @ 0-left  right L1J0@)  5 ends (can be 0("@ 2JN h<@ TE)C1O F, Dlat fmf1 [ 777I8*" Js5 п 6 ۂ2-sidjtai"4 t-dri,bu< KHMw ~ &_&yShhlse.barq!}n lsd0]j2cbind(4y-). usesS met(hodB!.@defaul ling lm af set2tup 796E4S*tuRrl_90deinQ @ ', /Zf7nce r Qmiples]"HconKucc.i 4t* 1  $ YtS9~  %t rYq e/j5e%^% * % $lK p : D8[iq  ,  f * %  $D<p ހ '^x^@F[(Conc, Ca )x3 1?LP,;% L* J.pl ? L, x% Z** $)42J bYȨ @ `ja{nest.int <- (get.)G(droT!q# ne|o check 1% # how;' ?a@ a paHste"~N X$08e^)*7# whoiZ Ŭ i@f(missJ(twe\s|"@)(is.na(pmatch-P%K * ))8 $[[H@]]rep(1q < R0 # Ǔ U@F[!dulpl (_Q @], rowme= pbT=bnA0WW]<UPF Zٚ '"P*"~tapply(@Within,62aum bser(B!Qcova )" n)A((i{-00(`M![i]1 x1S k2P+qP} (Be tween T%)81 $ўB:+P€*f6V$vp7@, 0%u<2 %$u(>NE3e5 :zf]9*/* $*e% ** $TA%$& W&n x}3>W1[?WKCb<P$*/ }sFwbW&&&~@g {ch. 13EEQ skQ8;wv)0(1977 .9E,F kDQ(Ad on-We0sley7xc,@&tBB@>p=all.predi}.e.[1:2],3s&,@$=predict(fit, = "terms"), :&@Q pP!~! qto are. y must be in , as well as ir in order.ss of s8* ,@A$  @A& ! Power = 1 - slope . SE comes e regression is at best advisory. Zsider formation y^p = exp(p*log(y)), or log(y) if p=0. In practice, nearest half integer.* AB$  7ANB. ,$!ӌ r.* BxB$  dNBKC, &! # Bekk data {(1991).fit <- smooth~lab*mat, )r(bekk.fit)s.plot(bekk.fit).lfit <- log(smooth)~lab*mat, )r(bekk.lfit)s.plot(bekk.lfit)3?xB~C1~CCF5QKCC* $P!~  , ~CC%  cCBD' ! (x.factor, , =O= )(x.factor, , =O= )CD` }P!~! "x4@.8Ka.N, = [ways.* BDE$  QDhE8 >! " 7 * EE$  s+hEFb V!*JumpId("splus.hlp", "interaction.plot_t") t: ,  .* E/F$  d[_FF( ! @> (Conc, , VJndel.plot(Conc, , VD30/FF1 FFH5AFF* $P!~  , FG%  8FVG% <! ( x, = 0.2 )b<GG> LxP!~! x@minimum ion of * VGG$  @'G"H% N! of x's !slight ment* GLH$  d=!"HH% B! ( x, (y), state )3'LHH1!HHAK57HH* $nP!~  , HI%  A%H^I% J! (x.factor,  ="factor")k\II> LP!~! "wof  or   usually or  * ^II$  6!I)J% B! of vs of "* ISJ$  $)JJ ҀH!!JumpId("splus.hlp", "unclass_t") *JumpId("splus.hlp", "interaction.plot_t") unclass, .* SJK$  d-JAK% ! (a)1K1"lHelvTimes New RomanSymbolArialCourier New   uEEM. 8""|M. . M  u$  | ?/&;)i2455all.covarsall.factorsall.predictorsbox.fig branchburstcharci.plotci.width circle$Contents(cor.coef,cov.coef0df.resid4dotna8effect.fit<effect.plot@ellipseDenscriptHfactor.labelsLget.levelsPget.listTghostviewXharmonic.mean\int.plot`lpqdlprhlsd.barllsd.plotplsmeantmandel.plotxmargin.plot|margin.plot.lmmlinesmodel.responsemplotmpointsmqqnormnestednested.treenorm.ellipseplotterprecisionpssample.sizese.barse.coefsplit.plotstd.devtrans.plottukey.plotuncollideunfactorMP\~hc29iates, predictors (factors & cov s) C:\TEMP\~hc35all factors, covariates, predictors (factors & covariates) all.predictors Qdetermine names of all factors, covariates, predictors (factors & covariates) names *Ji.ib_._ H?lz3Help for pdaBrowseButtons()_$_/&;)Lz#H#Table of Contents determine names of all factors, covariates, predictors (factors & covariates)udraw boxplot figure at specified positionEutility functions for ci.plotcorrelation, covariance, or standard errors of model coefficientsresidual degrees of freedomreinterpret "." as NA in dataframeGet fit effects and plot beside residuals.draw circles and ellipses8determine labels for all factors used in model fit"get elements of data with names in factors|UNIX interfaces to printersharmonic mean of numeric vectorMInteraction Plots of Response with Interval Bars. UNIX interfaces to printers Least Squares MeansMargin Plots of Response for Pairs of FactorsMargin Plots of Response for Pairs of Factors in Linear ModelMultiple Plots (Points, Lines) by GroupuMultiple Q-Q Normal and Half-Normal Plots by Group$nested weighted means for experimental units draw nested treegraphics device selection based on environment variables round value to precisioncompute sample size by groupStandard Error and Fisher's LSD Bars for Interaction Plots ~function to do ????standard deviation of objectDiagnostic plot to find power transform to remove interaction.Tukey/Mandel Interaction Plots of Response for Pairs of Factorsspread out values to avoid collision in plottingunclass factor (but keep numeric level)=/&;)L4!!Ć tGE  & $ ̂6ҵ%Eӌׂ0muz3,uT"M~'|hn. >km.08V6?t:Hؙ? 6E3.dĦ%gLi gUpiqM<{he variables named in the fit object (taken from fit if omitted). factors Character string of length 2 with names of x.factor and trace.factor as found in fit and data (default is first 2 factors in fit). string of length 2 with names of x.factor and trace.factor as found in fit and data (default is first 2 factors in fit). size (sample size(s) at each bar position. size(s) at each bar position. rdf residual degrees of freedom. degrees of freedom. sd standard deviation. deviation. level significance level. level. critical.value )2-sided tail value of t-distribution. tail value of t-distribution. ... Foptional specification of other parameters to be passed to se.bar. specification of other parameters to be passed to se.bar. Standard error bars (se.bar) or an LSD bar (lsd.bar) are plotted at position(s) cbind(xpos,ypos). Lsd.bar uses the S method, with the lsd.bar.default calling lsd.bar.lm after setting up the fit, data and factor. Lsd.bar.lm in turn calls se.bar after de error bars (se.bar) or an LSD bar (lsd.bar) are plotted at position(s) cbind(xpos,ypos). Lsd.bar uses the S method, with the lsd.bar.default calling lsd.bar.lm after setting up the fit, data and factor. Lsd.bar.lm in turn calls se.bar after de termining the bar height using size, rdf, sd and level. Confidence intervals or other multiple bars can be constructed ci.bar (see ci.plot). the bar height using size, rdf, sd and level.