Errata Itms with astrisks will still b in th Scond Printing Author wbsit URL: http://chs.unl.du/edpsych/rjsit/hom. P7. Th squar root of rfrrd to σ E (i.., σ E is rfrrd to not Th squar root of σ E (i.., P8. A narrow intrval indicats comparativly lss uncrtainty about a not A narrow intrval indicats comparativly lss crtainty about a P8. Thr ar two quations listd as.. Th scond. quation should b labld.. P8. Equation.3 should b ( α% CB : ˆ θ ± z ˆ ( α / σ( θ σ E is P64. To summariz our analyss, th nonlinar factor analysis providd support that a unidimnsional modl of th data is a rasonabl rprsntation [ ] th data. [of] is missing P73. Equation 4.. & n r = p ( X x, ϑν, N r i =.645 =.79 =.645 =.79.74369.74369 P79. Tabl 4.. PSD( ˆ θ PSD( ˆ θ P86. Lin : Should rad: A(X =.7648E-4 =.7648 P86. Lin 7: α + α should b α + α P9. Lin 6: IFNAME= C:\MATHRSCH.dat should b IFNAME= C:\MATHRSCH.PAR P9. (ust blow Equation 4.7. (or θ is th targt mtric. ξ is th P93. Equation 4. should rad κ = δ (or θ is th δ ζ δ δ on th targt mtric P. (ust abov Figur 5.. p( x= θα,, δ p( x = θα,, δ ξ is th δ Last updatd //5
P. Equation 5.5. p α ( θ δ α ( θ δ ' = α = α α ( θ δ α ( θ δ α ( θ δ ( ( + + + P4. R 774.95 397.3 397.3-64.99 = =.3 R = =.3 774.95 397.3 P43. Equation 6.8. Thr is no closing parnthsis or th opning parnthsis may nd to b rmovd. P6. Th last lin. Equation 6. Equation 6. P65. ( θ δ ( θ δ θ ϕ P65. Scond paragraph, last sntnc should rad Th probability of both of ths is givn by ( θ δ adding th (mutually xclusiv vnts of and, that is,. P65. Third paragraph, fourth and fifth sntncs should rad Thrfor, to obtain an x = th individual passs through x = (i.., ( θ δ, passs through x = (i..,, and thn ( θ δ + ( θ δ passs through th scond transition point (i..,. Thrfor, th probability of x = ( θ δ + ( θ δ is givn by. P65. Fourth paragraph, first sntnc should rad Whn ach of th thr trms (i.., x = : ( θ δ, x = : ( θ δ + ( θ δ, x = : is P65. Figur 7.: ( θ δ panl a: panl b: Thrfor, Figur 7. should appar as: x = x = x = θ δ + θ δ panl c: ( ( (a (b (c δ δ + ( θ δ + ( θ δ + ( θ δ θ Last updatd //5
P7. Tabl 7. 'NEXAMINEES=3' should b 'NEXAMINEES=94' P77. Th last paragraph. Thr is no A, B, and C in Figur 7.7. Figur 7.7A, Figur 7.7B, and Figur 7.7C corrspond to th top, middl, and bottom graphs, rspctivly. P97. th lin from th bottom..3 logit. logit (-9.6 = -.6. 4 th lin from th bottom. strongly agr strongly disagr P99. 9 th lin from th top. rspondnts. 45 rspondnts (8 popl/4 catgoris=45 P. First full paragraph, first sntnc: an itm δ k s should b an itm δ h s and in th last sntnc of this paragraph substituting δ k = δ -τ k should b substituting δ h = δ -τ h P. Figur 8. α S should b α. P5. 4 th lin from th top. th approximat rang of -.57 to.3 th approximat rang of -.88 to.3 (itm 6 s δ = -.88 { ' x } P3. Equation 8.8 should b: I x ' ' x x+ x x x+ p P P ( θ = = p P P P3. Th first quation aftr so that in th bottom half of this pag should b: α ( θ δ α ( θ δ ' ' ' p = [ P P ] = α = α ϕ ϕ P3. Th first quation should b: α ( θ δ α ( θ δ ' ' ' p = [ P P ] = α α ϕ ϕ α ( θ δ α ( θ δ = α ϕ ϕ P34. Equation 8.. xk xk Last updatd //5
P5. Th claus aftr th smicolon should rad: " in this cas th corrsponding HIGH catgory idntification dos not nd to b changd (i.., HIGH=(,3,4,3." P5. Last lin of th paragraph that continus from p. 5:..; w us th lnl valu blow. should rad..; w us th -lnl valu blow. P66. First full paragraph, lin 5:.. th diffrnc chi-squar is 88.4-59.73 = 9. should rad.. th diffrnc chi-squar is 88.4-59.3 = 9. P66. Scond to last sntnc in th first full paragraph: With a critical X of should b With a critical X of (i.., chi squar is not italicizd P68. Itm 4 s ORFs for th thta rang - to (lft panl and - to (right panl P79. 9 th lin from th bottom. According to Figur.3, th rightmost lin is rlatd to p =.95 instad of p =.9 α κ α κ P94. γ = γ γ = γ ζ ζ ( ( P96. Tabl. s Not should rad Nam=intrprsnl.inp Last updatd //5
P33. Equation.6 should rad κ = δ ζ δ P3. Tabl.3: Th initial valu of th K should b. not.. Thrfor, INITIAL VALUE FOR A=.8 INITIAL VALUE FOR K=. should b INITIAL VALUE FOR A=.8 INITIAL VALUE FOR K=. P33. Th TSW G DIF approach can b prformd using ithr MULTILOG or IRTLRDIF (Thissn, D. (. IRTLRDIF v.b: Softwar for th computation of th statistics involvd in itm rspons thory liklihood-ratio tsts for diffrntial itm functioning [Computr softwar and manual]. Chapl Hill: L.L. Thurston Psychomtric Laboratory, Univrsity of North Carolina. P338. Th G should rad 4.47 not 44.47. P338. Th stimatd full modl should b: z3 =.333 +.37 X +.774 RACE.7( X RACE P338. Th stimatd rducd modl ( should b: z3 =.5558 +.48 X.9577 RACE P339. Tabl.4 should rad: / Full Modl / : Th LOGISTIC Procdur Modl Fit Statistics Intrcpt Intrcpt and Critrion Only Covariats AIC 568.33 5.57 SC 573.947 47.98 - Log L 566.33 7.57 Tsting Global Null Hypothsis: BETA= Tst Chi-Squar DF Pr > ChiSq Liklihood Ratio 348.844 3 <. Scor 3.575 3 <. Wald 38.7854 3 <. Last updatd //5
Th LOGISTIC Procdur Analysis of Maximum Liklihood Estimats Standard Wald Paramtr DF Estimat Error Chi-Squar Pr > ChiSq Intrcpt -.333.366 3.533. X.37.86 5.356.7 RACE.774.4484.9596.854 XRACE -.7.93 36.96 <. : / rduc modl (/ full modl / : Modl Fit Statistics Intrcpt Intrcpt and Critrion Only Covariats AIC 568.33 64.93 SC 573.947 8.77 - Log L 566.33 58.93 Tsting Global Null Hypothsis: BETA= Tst Chi-Squar DF Pr > ChiSq Liklihood Ratio 37.475 <. Scor 9.768 <. Wald 6.78 <. Analysis of Maximum Liklihood Estimats Standard Wald Paramtr DF Estimat Error Chi-Squar Pr > ChiSq Intrcpt.5558.88 8.747.3 X.48.749 4.6 <. RACE -.9577. 6.366 <. Odds Ratio Estimats Point 95% Wald Effct Estimat Confidnc Limits X.49.34.65 RACE.4..79 P34. Th scond to th last sntnc in th paragraph should rad: Givn our coding of th RACE variabl if ˆ τ <, thn th itm favors th focal group. In contrast, if ˆ τ > thn th itm favors th rfrnc group. 6 P34. Tabl.5 should rad: / rduc modl ( / : Modl Fit Statistics Intrcpt Intrcpt and Critrion Only Covariats AIC 568.33 569.79 SC 573.947 58.94 - Log L 566.33 565.79 Last updatd //5
Tsting Global Null Hypothsis: BETA= Tst Chi-Squar DF Pr > ChiSq Liklihood Ratio.67.434 Scor.63.43 Wald.6.433 Analysis of Maximum Liklihood Estimats Standard Wald Paramtr DF Estimat Error Chi-Squar Pr > ChiSq Intrcpt.85.77 4.353 <. X -.488.69.6.433 P345. Endnot 6. Rducd modl (: should b z3 =.5558 +.48 X.9577 RACE Odds Ratio Estimats Point 95% Wald Effct Estimat Confidnc Limits X.995.983.7 and th corrsponding txt should rad: Thrfor, holding th obsrvd scor fixd and switching from th focal group to th rfrnc group rsults in a dcras in th log odds of obtaining a rspons of by.9577. In trms of odds w hav that th odds that a rfrnc group mmbr will produc a rspons of ar xp(-.9577=.4 to (not:.4 is th valu listd as th Point Estimat in th Odds Ratio Estimats sction. Altrnativly, holding th obsrvd scor fixd, on xpcts th odds of focal group mmbrs to corrctly rspond to th itm to b roughly 7 to (i.., /.4 = 7.83 rlativ to comparabl rfrnc group mmbrs. P356. 6 th lin from th top. if q =, thn δ =, and if q = N, thn δ = - P357. First Equation: ngativ sign in from of q : ln L(x δ q p δ N t ( t = + i i= P358. B.4 s subscripts ar not formattd proprly. It should appar as: σ (δ ˆ = L X= n p ( p X X X Last updatd //5
P37. Th last quation should b labld C.34. Th paragraph should rad: If w tak our total sampl of individuals and divid it into subgroups and w rdfin th standard dviation in Equation C.33 to b th standard dviation of a subgroup, σ i, with man µ i, thn its substitution into Equation C. givs Thurston's mntal ag modl; w'r assuming that ach subgroup is normally distributd. That is, Thurston (95;.g., s p. 44 dvlopd a modl basd on th cumulativ normal distribution to dtrmin th proportion of individuals of a spcifid ag group corrctly rsponding to an itm. P45. Equation E.4. χ χ = SC SC Last updatd //5