Alpha if Deleted and Loss in Criterion Validity 1. Appeared in British Journal of Mathematical and Statistical Psychology, 2008, 61,


 Blake Barrett
 3 years ago
 Views:
Transcription
1 Alpha f Deleted and Loss n Crteron Valdty Appeared n Brtsh Journal of Mathematcal and Statstcal Psychology, 2008, 6, Alpha f Item Deleted: A Note on Crteron Valdty Loss n Scale Revson f Maxmsng Coeffcent Alpha Tenko Raykov Mchgan State Unversty Author Note: I am grateful to B. Muthen for nstructve comments on constrant evaluaton, as well as to the Edtor and two anonymous Referees for valuable crtcsm on an earler draft of the paper, whch contrbuted consderably to ts mprovement. Correspondence on ths paper may be addressed to Tenko Raykov, Measurement and Quanttatve Methods, Mchgan State Unversty, East Lansng, MI 48824, USA; emal: RUNNING HEAD: ALPHA IF ITEM DELETED AND VALIDITY LOSS
2 Alpha f Deleted and Loss n Crteron Valdty 2 Abstract Ths note s concerned wth a valdtyrelated lmtaton of the wdely avalable and used ndex alpha f tem deleted n the process of constructon and development of multplecomponent measurng nstruments. Attenton s drawn to the fact that ths statstc can suggest dspensng wth such scale components, whose removal leads to loss n crteron valdty whle maxmsng the popular coeffcent alpha. As an alternatve, a latent varable modellng approach s dscussed that can be used for pont and nterval estmaton of composte crteron valdty (as well as relablty) after deleton of sngle components. The method can also be utlsed to test conventonal or mnmum level hypotheses about assocated populaton change n measurement qualty ndces. Keywords: coeffcent alpha, crteron valdty, nterval estmaton, latent varable modellng, multplecomponent measurng nstrument, relablty
3 Alpha f Deleted and Loss n Crteron Valdty 3 Alpha f Item Deleted : A Note on Loss of Crteron Valdty n Scale Development If Maxmsng Coeffcent Alpha Multplecomponent measurng nstruments are hghly popular n psychology and the behavoural scences. Before beng wdely used they typcally need to undergo a process of development through possbly repeated revsons that am to ensure hgh psychometrc qualty of fnally recommended scales, n partcular hgh relablty and valdty. A rather frequently used statstc for these purposes n emprcal research s Cronbach s coeffcent alpha (α; e.g., Cronbach, 95), and especally the ndex alpha f tem deleted that represents the ncrement or drop n the sample value of α f dspensng wth a scale component. Recently, however, Raykov (2007a) showed that n certan crcumstances that do not appear rare n behavoural research, ths ndex can suggest the deleton of such nstrument components whose removal leads to maxmal ncrement n α but entals consderable loss n composte relablty. Ths results from the fact that α n general ncorrectly evaluates scale relablty already at the populaton level (e.g., Novck & Lews, 967; Zmmerman, 972), and ponts out the possblty that whle seekng components to remove n order to maxmse coeffcent alpha, a psychologst can n fact serously compromse relablty of an nstrument beng developed. The present note deals wth an addtonal aspect of ths potentally serous lmtaton of the popular alpha coeffcent, and n partcular of the wdely utlsed statstc alpha f tem deleted. The remander ndcates that crteron valdty can smlarly decrease as a result of removng a component from a tentatve scale whle maxmsng coeffcent alpha, even f data were avalable from an entre studed populaton. As an alternatve to ths statstc, therefore, an extenson of the latent varable modellng procedure n Raykov (2007a) s recommended that yelds pont and nterval estmaton of crteron valdty, n addton to that of relablty, after dspensng wth sngle components. The method provdes ranges of plausble populaton values for these measurement qualty ndces followng any component s deleton, and can be used for testng conventonal as well as mnmum level hypotheses about them.
4 Alpha f Deleted and Loss n Crteron Valdty 4 Loss n Crteron Valdty When Deletng Components to Maxmse Coeffcent Alpha Ths dscusson s based on the assumpton that a set of congenerc measures s gven, denoted X, X 2,, X p (p > 2; Jöreskog, 97), that s, X = + ε = γ + β ξ + ε () T holds ( =,, p), where T,...,, T2 Tp and ε, ε 2,..., ε p are respectvely ther true and error scores, and ξ desgnates the common latent dmenson evaluated by the measures (e.g., ξ = T can be taken; Lord & Novck, 968). For dentfablty reasons, Var(ξ) = s also set, where Var(.) denotes varance n a studed populaton, and wth respect to ε t s only requred that ther covarance matrx Ψ be postve defnte (e.g., Zmmerman, 975; =,, p). Assume also that a crteron varable, C, s prespecfed. In the rest of ths artcle, the relablty of the composte Y = X + X X p wll be of nterest as well as that of closely related versons of t, along wth ther crteron valdty as reflected n the correlaton coeffcent Corr (Y, C) (e.g., Crocker & Algna, 986). 2 In the process of nstrument constructon and development, behavoural scentsts commonly follow the wdespread practce of repeatedly examnng the change n coeffcent alpha after sngle component removal, typcally referred to as alpha f tem deleted. Ths procedure s based on the sample value of the gan or loss n α occurrng f say the th component s dropped from a tentatve scale: Y, = αy αy, α, (2) where
5 Alpha f Deleted and Loss n Crteron Valdty 5 α Y = Cov( X, X p p Var( Y ) ) (3) s coeffcent alpha for the scale Y, and α Y, denotes ths coeffcent for the composte Y  = Y X,.e., represents alpha f ths tem X s deleted ( =,, p). The estmates of α Y, are furnshed by wdely crculated software, e.g., SPSS, SAS, STATISTICA, and are at present nearly routnely utlsed n the behavoural and socal scences for purposes of nstrument revson ( =,, p). Use of ths procedure s based on the tact, but n general ncorrect (see above), assumpton that α represents the change n relablty followng deleton of the th component ( =,, p). On ths presumpton, a wdely adhered to practce n emprcal research s to nspect the ndex alpha f tem deleted for each component n a tentatve scale, n an effort to dentfy a way of maxmally enhancng relablty va sngle tem deleton; then scholars commonly proceed wth the scale verson whch results from droppng the component assocated wth the hghest α Y, such that α Y > αy Y,, ( =,, p). As shown recently n Raykov (2007a), however, ths procedure cannot be generally trusted because of two mportant reasons. On the one hand, t depends crtcally on the sample estmate of the gan or drop n coeffcent alpha due to component removal, and n addton α n general ncorrectly evaluates scale relablty already at the populaton level, as mentoned earler. Consequently, a researcher relyng on the ndex alpha f tem deleted could decde to proceed wth such a revson of a tentatve composte, whch s assocated wth maxmal ncrease n α but n actual fact leads to lower relablty, even f data were avalable from an entre populaton. Ths lmtaton of the popular statstc alpha f tem deleted turns out to have further consequences to that ust ndcated. Specfcally, dspensng wth a component for whch α Y, > αy can lead also to loss n crteron valdty, a maor aspect of what may well be consdered the bottom lne n behavoural measurement. To see ths, from
6 Alpha f Deleted and Loss n Crteron Valdty 6 Equaton () follows Y k = k = ( β ) ξ + ε, (4) = and hence (e.g., Lord & Novck, 968) Corr Cov[ C,( k β ) ξ + k ε ] σ = = ξc = ξc ( Y, C) = = = ρ k k Y = ωy, (5) σ C k 2 k σ C Var( C) Var[( β ) ξ + ε ] ( β ) + θ = = = = k β σ say, where σ ξc, σ C, and ρ Y denote the latentcrteron covarance, crteron standard devaton and scale relablty, respectvely, whle θ = Var(ε ) ( =,, p). Now denote by ω and ρ correspondngly the crteron valdty and Y, Y, relablty of a tentatve scale from whch the th component s dropped ( =,, p). As shown n Raykov (2007a), there exst theoretcally and emprcally relevant settngs that do not appear rare n psychologcal research, where removal of a component assocated wth the maxmal ncrease n coeffcent alpha entals n fact a loss n relablty. 3 Let n such a settng the kth component possess accordngly the propertes that () α > α Y, k Y, () α Y, k s hghest for all k ( k p), and () ρ Y, k < ρ Y. From the nequalty n () and Equaton (5) (wth ts correspondng modfcaton for the sorevsed composte), t obvously follows ω σ ξc ξc Y, = Corr( Y X k, C) = ρy, k < ρy = Corr( Y, C) = ωy, (6) σ C σ C σ that s, after dspensng wth ts kth component the revsed scale has actually lower crteron valdty than ts mmedately precedng verson from whch t s obtaned. Equaton (6) lets one further observe that the amount by whch crteron valdty wll be
7 Alpha f Deleted and Loss n Crteron Valdty 7 compromsed n ths way, ω Y ωy,, depends on the assocated loss n relablty ndex and the correlaton between latent and crteron varables. Therefore, at least n the crcumstances outlned n Raykov (2007a; see Footnote 3), the resultng trmmed scale score, Y X k, wll have lower crteron valdty. Thus t s possble that a psychologst nvolved n nstrument development who follows the wdespread practce of removng a component from a tentatve scale, whch s assocated wth the hghest ncrease n coeffcent alpha, n actual fact arrves at a revsed scale that has consderably nferor crteron valdty (n addton to such relablty) relatve to ts verson before droppng that component. When ths practce s adhered to across several consecutve revsons, as s commonly the case n emprcal research, due to accumulaton of ths negatve effect t s obvous that the end verson may have substantally lower crteron valdty as well as relablty compared to an ntal scale. A Latent Varable Modellng Approach to Evaluaton of Measurement Qualty Followng Sngle Component Deleton In order to resolve these potentally serous defcences of the popular statstc alpha f tem deleted, the latent varable modellng approach n Raykov (2007a) can be extended to accomplsh pont and nterval estmaton of crteron valdty after deleton of any component from a tentatve scale. Ths procedure s not concerned wth the statstc alpha f tem deleted but s nstead entrely based on the coeffcent of crteron valdty, as well as that of relablty, for the composte resultng from deletng the th component of a gven scale ( =,.., p). To ths end, Equaton () s frst consdered as defnng a latent varable model (e.g., Muthén, 2002), and then 2p external parameters (new parameters, or auxlary parameters) are ntroduced (cf. Raykov, 2007b). The frst p of them, denoted π, π 2,, π p, are defned as the crteron valdty coeffcents of the verson resultng after deletng the th component (see Equaton (5)):
8 Alpha f Deleted and Loss n Crteron Valdty 8 π σ ξc = σ C k β =, k 2 k ( β + ) θ = = = Y, ω, (7) whle the second set of p external parameters, π p+, π p+2,, π 2p, are defned as the relablty coeffcents of the correspondng scale versons p 2 ( β ) = π p+ = k = k Y, 2 ( β ) + θ = = ρ, (8) ( =,, p). (Wth error covarances, the denomnators n the rghthand sde of Equatons (7) and (8) are extended by the sum of nonzero error covarance estmates; e.g., McDonald, 999.) It s emphassed that (7) and (8) are not model parameters but are functons of the latter, and hence can be estmated once those are so. When the maxmum lkelhood (ML) method s used for model fttng purposes, due to the nvarance property of ML, the rghthand sdes of (7) and (8) wrtten n terms of the partcpatng parameter estmates represent correspondngly the ML estmates of crteron valdty and scale relablty after removng the th component ( =,, p). Therefore, the latter estmates share all desrable largesample propertes of ML estmates consstency, unbasedness, normalty and effcency (e.g., Rao, 973). The estmates (7) and (8) do not address the queston of how close they are to the respectve populaton crteron valdty and relablty coeffcents after dspensng wth a gven component from a tentatve scale, whch are the actual coeffcents of nterest. To ths end, a standard error and confdence nterval for these quanttes s needed. Usng the delta method (e.g., Rao, 973), an approxmate standard error and confdence nterval was furnshed n Raykov (2007a) for scale relablty followng deleton of any
9 Alpha f Deleted and Loss n Crteron Valdty 9 component, and the same method can be used here to also render an approxmate standard error and confdence nterval for the crteron valdty of each scale verson obtaned n ths way. Denote frst the (2p  2) x vector of parameters n model () after deletng the th scale component by ψ = (ψ,,ψ , ψ +,, ψ p, ψ p+,, ψ p+, ψ p++,, ψ 2p ), where prmng stands for transposton and the notaton ψ = β,, ψ p = β p, ψ p = θ,, ψ 2p = θ p s used for ease of reference (see Equaton (); =,, p). The frstorder Taylor expanson of the crteron valdty coeffcent (7) around the populaton parameter ψ 0, = ( 0,,..., β0,, β0,,..., β0, p, θ0,,..., θ0,, θ0,,..., θ0, p ) β + + = ( 0,,..., ψ 0,, ψ 0,,..., ψ 0, p, ψ 0, p,..., ψ 0, p, ψ 0, p,..., ψ 0, 2 p ) ψ s 2 p ˆ p, ω p, ( ψ 0, ) + Dˆ ( ψˆ, ψ 0,, ) =,, p+ ˆ ω, (9) where denotes approxmately equal and Dˆ = ˆ ω p, ψ s the partal dervatve of ˆω p, wth respect to ts th argument, taken at the parameter estmate pont ( =,, , +,, p+, p++,, 2p; =,, p). (The explct expressons for these dervatves can be rendered followng wellknown rules for dfferentaton, but are actually not needed for the purposes of ths note, as ndcated below.) Hence, an approxmate standard error for crteron valdty followng removal of the th component s obtaned from Equaton (9) as: ˆ ω p ˆ, ω p, S. Eˆ.( ω p, ) = Cov( ψˆ ), (0) ψ ψ
10 Alpha f Deleted and Loss n Crteron Valdty 0 ˆ ω p where ( D D D D D D D ), = ˆ, ˆ,..., ˆ, ˆ,..., ˆ,..., ˆ, ˆ,..., ˆ ψ D 2 + p p+ p+ + 2 p s the row vector of above mentoned dervatves and Cov ψ ˆ ) s the covarance matrx of pertnent ( parameter estmators, evaluated at the model soluton ( =,, p; cf. Raykov, 2007a). Wth ths standard error, an approxmate 00(δ)%confdence nterval (0 < δ < ) for crteron valdty after droppng the th component results as follows by captalsng on the asymptotc normalty of the latent varable model parameter estmator (e.g., Muthén, 2002): (max(0, ˆω p,  z δ/2 S. Eˆ.( ω p, ) ), mn (, ˆω p, + z δ/2 S. Eˆ.( ω p, ))), () where z δ/2 denotes the δ/2th quantle of the standard normal dstrbuton whle max(.,.) and mn(.,.) stand for the larger and smaller numbers followng n parentheses, respectvely ( =,, p). The confdence nterval () provdes a range of plausble values, at a confdence level δ, for the populaton crteron valdty of the composte of all components but the th ( =,, p). From the dualty between hypothess testng and confdence nterval (e.g., Hays, 994) t follows that () could also be used to test, as well known, conventonal hypotheses at a sgnfcance level  δ about the crteron valdty (or, for the same matter, relablty) coeffcent after the th component s dropped ( =,, p). Moreover, () can be used to test mnmum level hypotheses about ths coeffcent. Such a hypothess states that after dspensng wth a component from a gven composte, the crteron valdty (or relablty) s equal to at least w 0 say (0 < w 0 < ), where w 0 s a substantvely desrable threshold for crteron valdty (or relablty) that a psychologst requres the composte to attan before beng recommended for wder use. Accordngly, the pertnent null hypothess s H 0 : ω  w 0 (or H 0 : ρ  w 0 ), wth correspondng alternatve hypothess H : ω  < w 0 (or H : ρ  < w 0 ) ( =,, p). Ths hypothess s tested by examnng whether the leftendpont of the correspondng confdence nterval s
11 Alpha f Deleted and Loss n Crteron Valdty entrely above the threshold, n whch case the null hypothess s consdered retanable. Otherwse the alternatve hypothess s accepted and further nstrument revson may be called for n order to accomplsh the desred mnmal level of valdty (or relablty). Emprcal mplementaton The descrbed approach can be mplemented n behavoural research wth the ncreasngly popular latent varable modellng program Mplus (Muthén & Muthén, 2006). Ths software ncorporates recent advances n numercal optmzaton, whch allow one to utlse readly the delta method applcaton outlned n the precedng secton for obtanng approxmate standard errors and confdence ntervals. Specfcally, fttng model () wth the added 2p external parameters n Equatons (7) and (8), upon a request for confdence nterval evaluaton, yelds pont as well as nterval estmates of these parameters,.e., for the crteron valdty (and relablty) coeffcents after removng any component from a tentatve scale. (The code accomplshng ths goal, wth annotatons, s provded n Appendx where t s appled wth data used n the llustraton secton.) It s emphassed that ths approach yelds nterval estmates of crteron valdty (or relablty) followng sngle component removal, as well as of an ntally consdered scale, whereas there s no counterpart nterval estmate avalable when one adheres to the wdely followed practce of usng the ndex alpha f tem deleted for scale revson purposes. The dscussed procedure s also drectly applcable n settngs wth mssng data that are frequently encountered n behavoural research dealng wth scale constructon and development. The method s then straghtforwardly employed va use of full nformaton maxmum lkelhood or multple mputaton f ther assumptons are plausble vz. data mssng at random and normalty (e.g., Muthén & Muthén, 2006; Lttle & Rubn, 2002). Last but not least, the proposed approach can be repeatedly used on scale versons resultng from precedng measure removal, n the search of yet further mprovement n ther crteron valdty (and relablty) followng deleton of any of ther own components. Fnal recommendatons regardng composte revson should be based,
12 Alpha f Deleted and Loss n Crteron Valdty 2 however, on results from a replcaton study on an ndependent sample from the same populaton, due to the possblty of captalzaton on chance. Illustraton on Data In order to demonstrate the possblty that the wdely used statstc alpha f tem deleted can suggest msleadng avenues of scale mprovement that are n fact assocated wth pronounced loss n crteron valdty, smulated multnormal data are employed n ths secton. These data wll also allow llustraton of the utlty of the dscussed approach to evaluaton of crteron valdty (and relablty) after sngle component deleton. To ths end, multvarate, zeromean normal data were generated for N = 500 cases and k = 5 components Y through Y 5 accordng to the model X = ξ + ε (2) X 2 = ξ + ε 2 X 3 = ξ + ε 3 X 4 = ξ + ε 4 X 5 = 6 ξ + ε 5, where ξ was standard normal and the error terms ε through ε 5 were ndependent zeromean normal varables wth varance.3 each; the crteron varable was generated as havng correlaton of.80 wth ξ. The resultng covarance matrx s presented n Table. Insert Table about here Conventonal scale analyss on the ntal composte contanng all 5 components Y through Y 5 (.e., of the scale score Y = Y + + Y 5 ) reveals an estmated alpha coeffcent of.702. The wdely used statstc alpha f tem deleted ndcates then that alpha wll be maxmsed f the last component, Y 5, s removed from ths composte. Specfcally, accordng to that statstc, ths removal would yeld a fourcomponent scale
13 Alpha f Deleted and Loss n Crteron Valdty 3 wth an alpha of.749 that s more than. hgher than the alpha resultng from droppng nstead any of the other four components (.e., Y to Y 4 ) from the ntal composte. It s stressed that alpha f tem deleted suggests here droppng the most relable component of all fve, n order to maxmse alpha. Indeed, Equatons (2) and mmedately followng dscusson mply that relablty of each of the frst four components s under.50, whle that of Y 5 s n excess of.95. To see the effect of deletng the last component on crteron valdty and relablty, the latent varable modelng approach of ths artcle s appled. Frst, fttng the congenerc model () (wth the 2p external parameters, whch do not affect model ft as they do not have any mplcatons on the covarance structure), one obtans acceptable goodness of ft ndces: chsquare = 3.73, degrees of freedom (df) = 9, pvalue (p) =.55, root mean square error of approxmaton (RMSEA) =.030 wth a 90%confdence nterval (0,.063). The crteron valdty and relablty of the fve versons of the ntal scale, whch result after each of ts components s dropped n turn, as well as of that startng scale are presented n Table 2 along wth correspondng standard errors and confdence ntervals. Insert Table 2 about here As seen from Table 2, removal of Y 5 as suggested by the statstc alpha f tem deleted n fact leads to a substantal decrement n crteron valdty from.773 (ntal scale) to.687 (composte of frst four components only), that s a drop by more than 0%. (Usng the data generaton parameters, ths crteron valdty loss s found to be equal to.096 n the populaton.) Smlarly, relablty drops from.938 to.742,.e., by more than 20%. (In the same way, ths relablty decrement s found to be.27 n the populaton.) These effects represent pronounced losses n measurement qualty, whch result f one were to follow the wdespread practce of deletng the sngle component whose removal maxmses coeffcent alpha. Note also that precson of estmaton, as udged by the wdth of the assocated confdence ntervals, also drops f one were to dspense wth the last component followng that popular procedure. Further, from Table 2 t s seen that deleton of any of the frst four components nstead does not have a notable effect on the
14 Alpha f Deleted and Loss n Crteron Valdty 4 pont estmate of crteron valdty or relablty, whle leadng to some relatvely mnmal loss of estmaton precson. Ths demonstraton exemplfes the pont that adherng to the wdely used statstc alpha f tem deleted for purposes of scale revson can be assocated wth a marked loss n crteron valdty and relablty, two measurement qualty ndces of specal relevance for psychology and the behavoural scences. Concluson For a number of decades, a wdespread practce has been followed by behavoural scentsts nvolved n nstrument development. Accordngly, the sample values of the popular coeffcent alpha before and after sngle component removal have receved crtcal attenton n an effort to fnd ways of revsng tentatve scales so as to maxmally enhance ther relablty. In partcular, the ndex alpha f tem deleted has been rather frequently nspected for ths purpose. The present note hghlghts a valdtyrelated lmtaton of ths statstc. The artcle shows that dspensng wth a scale component to maxmally ncrease coeffcent alpha, can n fact ental consderable loss n crteron valdty, a maor aspect of behavoural measurement qualty. In addton to a recent demonstraton n Raykov (2007a) that such a revson path can lead to loss n composte relablty, ths note further cautons psychologsts engaged n nstrument development that use of alpha f tem deleted can be serously msleadng n more than one mportant way. As an alternatve, the note dscusses a latent varable modellng procedure that provdes pont and nterval estmates of both crteron valdty and relablty followng deleton of each component n a tentatve scale. In addton, the outlned approach allows smultaneous examnaton of the factoral structure of a gven set of measures consdered as ts components. Moreover, the method s straghtforwardly applcable n cases wth mssng data usng maxmum lkelhood or multple mputaton, when ther assumptons are plausble (vz. multnormalty and data mssng at random), whch s qute often the case n emprcal contexts where nstrument development s conducted. The dscussed procedure, beng based on latent varable modellng that s grounded n an asymptotc theory (e.g., Muthén, 2002), yelds most trustworthy results wth large samples, and smlarly wth (approxmately) contnuous components. Further,
15 Alpha f Deleted and Loss n Crteron Valdty 5 beng concerned wth crteron valdty, the proposed method may yeld lmted nformaton about other relevant types of valdty of measurement (e.g., Crocker & Algna,986). In addton, ts results depend on the choce of a crteron varable, whch should be made n emprcal research based on detaled knowledge of a substantve doman of concern. Fnally, as presented n ths note, the procedure utlses the assumpton of congenerc measures, but t s stressed that t s readly extended to the case of more than a sngle underlyng source of latent varablty (see Footnote ).
16 Alpha f Deleted and Loss n Crteron Valdty 6 References Bollen, K. A. (989). Structural equatons wth latent varables. New York: Wley. Crocker, L., & Algna, J. (986). Introducton to classcal and modern test theory. Fort Worth, TX: Harcourt Brace Jovanovch. Cronbach, L. J. (95). Coeffcent alpha and the nternal structure of a test. Psychometrka, 6, Hays, W. L. (994). Statstcs. Fort Worth, TX: Harcourt Brace Jovanovch. Jöreskog, K. G. (97). Statstcal analyss of sets of congenerc tests. Psychometrka, 36, Lttle, R. J., & Rubn, D. B. (2002). Statstcal analyss wth mssng data. New York: Wley. Lord, F., & Novck, M. (968). Statstcal theores of mental test scores. Readngs, MA: AddsonWesley. McDonald, R. P. (999). Test theory. A unfed treatment. Mahwah, NJ: Erlbaum. Muthén, B. O. (2002). Beyond SEM: General latent varable modelng. Behavormetrka, 29, 87. Muthén, L. K., & Muthén, B. O. (2006). Mplus user s gude. Los Angeles, CA: Muthén & Muthén. Novck, M. R., & Lews, C. (967). Coeffcent alpha and the relablty of composte measurement. Psychometrka, 32, 3. Rao, C. R. (973). Lnear statstcal nference and ts applcatons. New York: Wley. Raykov, T. (2007a). Relablty f deleted, not alpha f deleted : Evaluaton of scale relablty followng component deleton. Brtsh Journal of Mathematcal and Statstcal Psychology, 60, Raykov, T. (2007b). Estmaton of revson effect on crteron valdty of multplecomponent measurng nstruments. Multvarate Behavoral Research, 42, Zmmerman, D. W. (972). Test relablty and the KuderRchardson formulas: Dervaton from probablty theory. Educatonal and Psychologcal Measurement, 32, Zmmerman, D. W. (975). Probablty measures, Hlbert spaces, and the axoms of classcal test theory. Psychometrka, 30,
17 Alpha f Deleted and Loss n Crteron Valdty 7 Table Covarance matrx of fve congenerc measures (N = 500) Varable Y Y 2 Y 3 Y 4 Y 5 C Y Y Y Y Y C Note. N = sample sze, C = crteron varable.
18 Alpha f Deleted and Loss n Crteron Valdty 8 Table 2 Pont and nterval estmates of crteron valdty and relablty of composte resultng after ndcated component s dropped from ntal scale wth all fve components, and for that ntal scale DM CV SE CI(CV) R SE CI(R) Y (.739,.807) (.924,.960) Y (.739,.807) (.923,.959) Y (.739,.807) (.924,.960) Y (.740,.808) (.925,.96) Y (.648,.727) (.707,.777) None (.739,.807) (.925,.952) Note. DM = dropped measure; CV = crteron valdty, R = relablty; SE = standard error, CI(CV) and CI(R) = 95%confdence nterval of crteron valdty and of relablty, respectvely. Entres n row None pertan to estmates and standard errors for crteron valdty and relablty of the ntal scale wth all fve components (.e., when none of the latter s removed).
19 Alpha f Deleted and Loss n Crteron Valdty 9 Footnotes If p = 2, addtonal dentfyng restrctons wll be needed, such as ndcator loadng equalty (true scoreequvalent measures) and/or error varance equalty (e.g., parallel measures; Lord & Novck, 968). Snce the locaton parameters, γ 2 γ k are not consequental for relablty n the settng underlyng ths γ,..., paper, for convenence they are all assumed equal to zero (e.g., Bollen, 989). The developments n ths note can be drectly generalsed to the case where more than a sngle latent dmenson s evaluated by a consdered set of measures, followng the correspondng approach n McDonald (999; omega coeffcent). 2 The procedure dscussed below s readly extended to the case when C s a latent varable wth at least two ndcators. As s common n latent varable modellng, C s also assumed unrelated to the error terms n the observed measures X,, X p (e.g., Bollen, 989). 3 As outlned n Raykov (2007a), at least the followng general setup belongs to these emprcal settngs, wth a sngle latent varable ξ and p = q + ndcators (p > 2; see Footnote ): X = β ξ + ε ( =,, q), X q+ = γ ξ + ε q+, where γ > β s suffcently large, Var(ε + ) Var(ε ) = θ ( =,, q; obvously, wthout lmtaton of generalty one can also presume that β > 0; n the last q+ equatons consderaton of component ntercepts s dspensed wth as they are nconsequental for relablty; see Footnote ). As shown n the last cted source, deleton of the last component n ths setup, whch leads to the hghest ncrement n coeffcent alpha, entals substantal loss of relablty. (Note that ths setup descrbes a case of q+ congenerc measures, of whch the frst q are parallel whle the last one s the most relable of all; see also Appendx 2 n that source.)
20 Alpha f Deleted and Loss n Crteron Valdty 20 Appendx Mplus Code for Evaluaton of Crteron Valdty and Relablty After Sngle Component Deleton TITLE: EVALUATION OF CRITERION VALIDITY/RELIABILITY AFTER COMPONENT DELETION DATA: FILE = <fle name>! PROVIDES NAME OF RAW DATA FILE. VARIABLE: NAMES = YY6;! ATTACHES LABELS TO OBSERVED VARIABLES MODEL: KSI BY Y* (P)! THIS AND NEXT 4 LINES DEFINE THE COMPONENTS Y2* (P2)! AND ATTACH TO THEM PARAMETER SYMBOLS TO BE Y3* (P3)! USED BELOW (SEE MODEL CONSTRAINT SECTION). Y4* (P4) Y5* (P5); Y* (P6);! THIS AND NEXT 4 LINES DEFINE THE ERROR VARIANCES Y2* (P7);! AND ATTACH TO THEM PARAMETER SYMBOLS TO BE USED Y3* (P8);! BELOW (SEE MODEL CONSTRAINT SECTION). Y4* (P9); Y5* (P0); C BY Y6*; C WITH KSI* (P);! C IS THE CRITERION VARIABLE FIXES LATENT VARIANCE AT, FOR MODEL IDENTIFICATION MODEL CONSTRAINT: NEW(PI_ PI_2 PI_3 PI_4 PI_5 PI_6 PI_7 PI_8 PI_9 PI_0 PI_ PI_2);! INTRODUCES THE AUXILIARY PARAMETERS π, π 2,, π 2 (SEE EQ. (7), (8)) PI_6=(P2+P3+P4+P5)**2/ ((P2+P3+P4+P5)**2+P7+P8+P9+P0);! = RELIABILITY W/OUT ST COMPONENT PI_7=(P+P3+P4+P5)**2/ ((P+P3+P4+P5)**2+P6+P8+P9+P0);! = RELIABILITY W/OUT 2ND COMPONENT PI_8=(P+P2+P4+P5)**2/ ((P+P2+P4+P5)**2+P6+P7+P9+P0);! = RELIABILITY W/OUT 3RD COMPONENT PI_9=(P+P2+P3+P5)**2/ ((P+P2+P3+P5)**2+P6+P7+P8+P0);! = RELIABILITY W/OUT 4TH COMPONENT PI_0=(P+P2+P3+P4)**2/ ((P+P2+P3+P4)**2+P6+P7+P8+P9);! = RELIABILITY W/OUT 5TH COMPONENT PI_2=(P+P2+P3+P4+P5)**2/ ((P+P2+P3+P4+P5)**2+P6+P7+P8+P9+P0);! = RELIABILITY WITH ALL COMP. PI_=P*SQRT(PI_6);! = CRITERION VALIDITY W/OUT ST COMPONENT. PI_2=P*SQRT(PI_7);! = CRITERION VALIDITY W/OUT 2ND COMPONENT. PI_3=P*SQRT(PI_8);! = CRITERION VALIDITY W/OUT 3RD COMPONENT.
PSYCHOLOGICAL RESEARCH (PYC 304C) Lecture 12
14 The Chsquared dstrbuton PSYCHOLOGICAL RESEARCH (PYC 304C) Lecture 1 If a normal varable X, havng mean µ and varance σ, s standardsed, the new varable Z has a mean 0 and varance 1. When ths standardsed
More informationThe covariance is the two variable analog to the variance. The formula for the covariance between two variables is
Regresson Lectures So far we have talked only about statstcs that descrbe one varable. What we are gong to be dscussng for much of the remander of the course s relatonshps between two or more varables.
More informationAn Alternative Way to Measure Private Equity Performance
An Alternatve Way to Measure Prvate Equty Performance Peter Todd Parlux Investment Technology LLC Summary Internal Rate of Return (IRR) s probably the most common way to measure the performance of prvate
More informationQuestions that we may have about the variables
Antono Olmos, 01 Multple Regresson Problem: we want to determne the effect of Desre for control, Famly support, Number of frends, and Score on the BDI test on Perceved Support of Latno women. Dependent
More informationInequality and The Accounting Period. Quentin Wodon and Shlomo Yitzhaki. World Bank and Hebrew University. September 2001.
Inequalty and The Accountng Perod Quentn Wodon and Shlomo Ytzha World Ban and Hebrew Unversty September Abstract Income nequalty typcally declnes wth the length of tme taen nto account for measurement.
More informationTHE METHOD OF LEAST SQUARES THE METHOD OF LEAST SQUARES
The goal: to measure (determne) an unknown quantty x (the value of a RV X) Realsaton: n results: y 1, y 2,..., y j,..., y n, (the measured values of Y 1, Y 2,..., Y j,..., Y n ) every result s encumbered
More informationInstitute of Informatics, Faculty of Business and Management, Brno University of Technology,Czech Republic
Lagrange Multplers as Quanttatve Indcators n Economcs Ivan Mezník Insttute of Informatcs, Faculty of Busness and Management, Brno Unversty of TechnologCzech Republc Abstract The quanttatve role of Lagrange
More informationBERNSTEIN POLYNOMIALS
OnLne Geometrc Modelng Notes BERNSTEIN POLYNOMIALS Kenneth I. Joy Vsualzaton and Graphcs Research Group Department of Computer Scence Unversty of Calforna, Davs Overvew Polynomals are ncredbly useful
More informationbenefit is 2, paid if the policyholder dies within the year, and probability of death within the year is ).
REVIEW OF RISK MANAGEMENT CONCEPTS LOSS DISTRIBUTIONS AND INSURANCE Loss and nsurance: When someone s subject to the rsk of ncurrng a fnancal loss, the loss s generally modeled usng a random varable or
More informationMultivariate EWMA Control Chart
Multvarate EWMA Control Chart Summary The Multvarate EWMA Control Chart procedure creates control charts for two or more numerc varables. Examnng the varables n a multvarate sense s extremely mportant
More informationModule 2 LOSSLESS IMAGE COMPRESSION SYSTEMS. Version 2 ECE IIT, Kharagpur
Module LOSSLESS IMAGE COMPRESSION SYSTEMS Lesson 3 Lossless Compresson: Huffman Codng Instructonal Objectves At the end of ths lesson, the students should be able to:. Defne and measure source entropy..
More informationStudy on CET4 Marks in China s Graded English Teaching
Study on CET4 Marks n Chna s Graded Englsh Teachng CHE We College of Foregn Studes, Shandong Insttute of Busness and Technology, P.R.Chna, 264005 Abstract: Ths paper deploys Logt model, and decomposes
More informationHYPOTHESIS TESTING OF PARAMETERS FOR ORDINARY LINEAR CIRCULAR REGRESSION
HYPOTHESIS TESTING OF PARAMETERS FOR ORDINARY LINEAR CIRCULAR REGRESSION Abdul Ghapor Hussn Centre for Foundaton Studes n Scence Unversty of Malaya 563 KUALA LUMPUR Emal: ghapor@umedumy Abstract Ths paper
More informationThe Development of Web Log Mining Based on ImproveKMeans Clustering Analysis
The Development of Web Log Mnng Based on ImproveKMeans Clusterng Analyss TngZhong Wang * College of Informaton Technology, Luoyang Normal Unversty, Luoyang, 471022, Chna wangtngzhong2@sna.cn Abstract.
More informationThe Analysis of Outliers in Statistical Data
THALES Project No. xxxx The Analyss of Outlers n Statstcal Data Research Team Chrysses Caron, Assocate Professor (P.I.) Vaslk Karot, Doctoral canddate Polychrons Economou, Chrstna Perrakou, Postgraduate
More information9.1 The Cumulative Sum Control Chart
Learnng Objectves 9.1 The Cumulatve Sum Control Chart 9.1.1 Basc Prncples: Cusum Control Chart for Montorng the Process Mean If s the target for the process mean, then the cumulatve sum control chart s
More informationNPAR TESTS. OneSample ChiSquare Test. Cell Specification. Observed Frequencies 1O i 6. Expected Frequencies 1EXP i 6
PAR TESTS If a WEIGHT varable s specfed, t s used to replcate a case as many tmes as ndcated by the weght value rounded to the nearest nteger. If the workspace requrements are exceeded and samplng has
More informationI. SCOPE, APPLICABILITY AND PARAMETERS Scope
D Executve Board Annex 9 Page A/R ethodologcal Tool alculaton of the number of sample plots for measurements wthn A/R D project actvtes (Verson 0) I. SOPE, PIABIITY AD PARAETERS Scope. Ths tool s applcable
More informationCausal, Explanatory Forecasting. Analysis. Regression Analysis. Simple Linear Regression. Which is Independent? Forecasting
Causal, Explanatory Forecastng Assumes causeandeffect relatonshp between system nputs and ts output Forecastng wth Regresson Analyss Rchard S. Barr Inputs System Cause + Effect Relatonshp The job of
More informationTHE DISTRIBUTION OF LOAN PORTFOLIO VALUE * Oldrich Alfons Vasicek
HE DISRIBUION OF LOAN PORFOLIO VALUE * Oldrch Alfons Vascek he amount of captal necessary to support a portfolo of debt securtes depends on the probablty dstrbuton of the portfolo loss. Consder a portfolo
More informationThe Analysis of Covariance. ERSH 8310 Keppel and Wickens Chapter 15
The Analyss of Covarance ERSH 830 Keppel and Wckens Chapter 5 Today s Class Intal Consderatons Covarance and Lnear Regresson The Lnear Regresson Equaton TheAnalyss of Covarance Assumptons Underlyng the
More informationChapter 7. RandomVariate Generation 7.1. Prof. Dr. Mesut Güneş Ch. 7 RandomVariate Generation
Chapter 7 RandomVarate Generaton 7. Contents Inversetransform Technque AcceptanceRejecton Technque Specal Propertes 7. Purpose & Overvew Develop understandng of generatng samples from a specfed dstrbuton
More informationCan Auto Liability Insurance Purchases Signal Risk Attitude?
Internatonal Journal of Busness and Economcs, 2011, Vol. 10, No. 2, 159164 Can Auto Lablty Insurance Purchases Sgnal Rsk Atttude? ChuShu L Department of Internatonal Busness, Asa Unversty, Tawan ShengChang
More informationForecasting the Direction and Strength of Stock Market Movement
Forecastng the Drecton and Strength of Stock Market Movement Jngwe Chen Mng Chen Nan Ye cjngwe@stanford.edu mchen5@stanford.edu nanye@stanford.edu Abstract  Stock market s one of the most complcated systems
More informationCommunication Networks II Contents
8 / 1  Communcaton Networs II (Görg)  www.comnets.unbremen.de Communcaton Networs II Contents 1 Fundamentals of probablty theory 2 Traffc n communcaton networs 3 Stochastc & Marovan Processes (SP
More informationSPEE Recommended Evaluation Practice #6 Definition of Decline Curve Parameters Background:
SPEE Recommended Evaluaton Practce #6 efnton of eclne Curve Parameters Background: The producton hstores of ol and gas wells can be analyzed to estmate reserves and future ol and gas producton rates and
More informationEconomic Interpretation of Regression. Theory and Applications
Economc Interpretaton of Regresson Theor and Applcatons Classcal and Baesan Econometrc Methods Applcaton of mathematcal statstcs to economc data for emprcal support Economc theor postulates a qualtatve
More informationSIMPLE LINEAR CORRELATION
SIMPLE LINEAR CORRELATION Smple lnear correlaton s a measure of the degree to whch two varables vary together, or a measure of the ntensty of the assocaton between two varables. Correlaton often s abused.
More informationErrorPropagation.nb 1. Error Propagation
ErrorPropagaton.nb Error Propagaton Suppose that we make observatons of a quantty x that s subject to random fluctuatons or measurement errors. Our best estmate of the true value for ths quantty s then
More informationCHAPTER 14 MORE ABOUT REGRESSION
CHAPTER 14 MORE ABOUT REGRESSION We learned n Chapter 5 that often a straght lne descrbes the pattern of a relatonshp between two quanttatve varables. For nstance, n Example 5.1 we explored the relatonshp
More informationChapter 14 Simple Linear Regression
Sldes Prepared JOHN S. LOUCKS St. Edward s Unverst Slde Chapter 4 Smple Lnear Regresson Smple Lnear Regresson Model Least Squares Method Coeffcent of Determnaton Model Assumptons Testng for Sgnfcance Usng
More informationA Computer Technique for Solving LP Problems with Bounded Variables
Dhaka Unv. J. Sc. 60(2): 163168, 2012 (July) A Computer Technque for Solvng LP Problems wth Bounded Varables S. M. Atqur Rahman Chowdhury * and Sanwar Uddn Ahmad Department of Mathematcs; Unversty of
More informationMAPP. MERIS level 3 cloud and water vapour products. Issue: 1. Revision: 0. Date: 9.12.1998. Function Name Organisation Signature Date
Ttel: Project: Doc. No.: MERIS level 3 cloud and water vapour products MAPP MAPPATBDClWVL3 Issue: 1 Revson: 0 Date: 9.12.1998 Functon Name Organsaton Sgnature Date Author: Bennartz FUB Preusker FUB Schüller
More informationHow Sets of Coherent Probabilities May Serve as Models for Degrees of Incoherence
1 st Internatonal Symposum on Imprecse Probabltes and Ther Applcatons, Ghent, Belgum, 29 June 2 July 1999 How Sets of Coherent Probabltes May Serve as Models for Degrees of Incoherence Mar J. Schervsh
More informationThe Probit Model. Alexander Spermann. SoSe 2009
The Probt Model Aleander Spermann Unversty of Freburg SoSe 009 Course outlne. Notaton and statstcal foundatons. Introducton to the Probt model 3. Applcaton 4. Coeffcents and margnal effects 5. Goodnessofft
More information8.5 UNITARY AND HERMITIAN MATRICES. The conjugate transpose of a complex matrix A, denoted by A*, is given by
6 CHAPTER 8 COMPLEX VECTOR SPACES 5. Fnd the kernel of the lnear transformaton gven n Exercse 5. In Exercses 55 and 56, fnd the mage of v, for the ndcated composton, where and are gven by the followng
More informationLatent Class Regression. Statistics for Psychosocial Research II: Structural Models December 4 and 6, 2006
Latent Class Regresson Statstcs for Psychosocal Research II: Structural Models December 4 and 6, 2006 Latent Class Regresson (LCR) What s t and when do we use t? Recall the standard latent class model
More informationIntroduction to Regression
Introducton to Regresson Regresson a means of predctng a dependent varable based one or more ndependent varables. Ths s done by fttng a lne or surface to the data ponts that mnmzes the total error. 
More informationFeature selection for intrusion detection. Slobodan Petrović NISlab, Gjøvik University College
Feature selecton for ntruson detecton Slobodan Petrovć NISlab, Gjøvk Unversty College Contents The feature selecton problem Intruson detecton Traffc features relevant for IDS The CFS measure The mrmr measure
More informationPortfolio Loss Distribution
Portfolo Loss Dstrbuton Rsky assets n loan ortfolo hghly llqud assets holdtomaturty n the bank s balance sheet Outstandngs The orton of the bank asset that has already been extended to borrowers. Commtment
More informationCHAPTER 7 THE TWOVARIABLE REGRESSION MODEL: HYPOTHESIS TESTING
CHAPTER 7 THE TWOVARIABLE REGRESSION MODEL: HYPOTHESIS TESTING QUESTIONS 7.1. (a) In the regresson contet, the method of least squares estmates the regresson parameters n such a way that the sum of the
More informationRecurrence. 1 Definitions and main statements
Recurrence 1 Defntons and man statements Let X n, n = 0, 1, 2,... be a MC wth the state space S = (1, 2,...), transton probabltes p j = P {X n+1 = j X n = }, and the transton matrx P = (p j ),j S def.
More informationAnalysis of Premium Liabilities for Australian Lines of Business
Summary of Analyss of Premum Labltes for Australan Lnes of Busness Emly Tao Honours Research Paper, The Unversty of Melbourne Emly Tao Acknowledgements I am grateful to the Australan Prudental Regulaton
More informationVariance estimation for the instrumental variables approach to measurement error in generalized linear models
he Stata Journal (2003) 3, Number 4, pp. 342 350 Varance estmaton for the nstrumental varables approach to measurement error n generalzed lnear models James W. Hardn Arnold School of Publc Health Unversty
More informationLuby s Alg. for Maximal Independent Sets using Pairwise Independence
Lecture Notes for Randomzed Algorthms Luby s Alg. for Maxmal Independent Sets usng Parwse Independence Last Updated by Erc Vgoda on February, 006 8. Maxmal Independent Sets For a graph G = (V, E), an ndependent
More information1. Measuring association using correlation and regression
How to measure assocaton I: Correlaton. 1. Measurng assocaton usng correlaton and regresson We often would lke to know how one varable, such as a mother's weght, s related to another varable, such as a
More informationWhat is Candidate Sampling
What s Canddate Samplng Say we have a multclass or mult label problem where each tranng example ( x, T ) conssts of a context x a small (mult)set of target classes T out of a large unverse L of possble
More informationOn the Optimal Control of a Cascade of HydroElectric Power Stations
On the Optmal Control of a Cascade of HydroElectrc Power Statons M.C.M. Guedes a, A.F. Rbero a, G.V. Smrnov b and S. Vlela c a Department of Mathematcs, School of Scences, Unversty of Porto, Portugal;
More informationCalculating the Trend Data
LASER INTERFEROMETER GRAVITATIONAL WAVE OBSERVATORY  LIGO  CALIFORNIA INSTITUTE OF TECHNOLOGY MASSACHUSETTS INSTITUTE OF TECHNOLOGY Techncal Note LIGOT990110B  D 3/29/07 Calculatng the Trend Data
More informationNasdaq Iceland Bond Indices 01 April 2015
Nasdaq Iceland Bond Indces 01 Aprl 2015 Fxed duraton Indces Introducton Nasdaq Iceland (the Exchange) began calculatng ts current bond ndces n the begnnng of 2005. They were a response to recent changes
More information1 Example 1: Axisaligned rectangles
COS 511: Theoretcal Machne Learnng Lecturer: Rob Schapre Lecture # 6 Scrbe: Aaron Schld February 21, 2013 Last class, we dscussed an analogue for Occam s Razor for nfnte hypothess spaces that, n conjuncton
More informationIntroduction: Analysis of Electronic Circuits
/30/008 ntroducton / ntroducton: Analyss of Electronc Crcuts Readng Assgnment: KVL and KCL text from EECS Just lke EECS, the majorty of problems (hw and exam) n EECS 3 wll be crcut analyss problems. Thus,
More informationExhaustive Regression. An Exploration of RegressionBased Data Mining Techniques Using Super Computation
Exhaustve Regresson An Exploraton of RegressonBased Data Mnng Technques Usng Super Computaton Antony Daves, Ph.D. Assocate Professor of Economcs Duquesne Unversty Pttsburgh, PA 58 Research Fellow The
More informationIMPROVEMENT OF CONVERGENCE CONDITION OF THE SQUAREROOT INTERVAL METHOD FOR MULTIPLE ZEROS 1
Nov Sad J. Math. Vol. 36, No. 2, 2006, 009 IMPROVEMENT OF CONVERGENCE CONDITION OF THE SQUAREROOT INTERVAL METHOD FOR MULTIPLE ZEROS Modrag S. Petkovć 2, Dušan M. Mloševć 3 Abstract. A new theorem concerned
More informationx f(x) 1 0.25 1 0.75 x 1 0 1 1 0.04 0.01 0.20 1 0.12 0.03 0.60
BIVARIATE DISTRIBUTIONS Let be a varable that assumes the values { 1,,..., n }. Then, a functon that epresses the relatve frequenc of these values s called a unvarate frequenc functon. It must be true
More informationSupport Vector Machines
Support Vector Machnes Max Wellng Department of Computer Scence Unversty of Toronto 10 Kng s College Road Toronto, M5S 3G5 Canada wellng@cs.toronto.edu Abstract Ths s a note to explan support vector machnes.
More informationL10: Linear discriminants analysis
L0: Lnear dscrmnants analyss Lnear dscrmnant analyss, two classes Lnear dscrmnant analyss, C classes LDA vs. PCA Lmtatons of LDA Varants of LDA Other dmensonalty reducton methods CSCE 666 Pattern Analyss
More informationGraph Theory and Cayley s Formula
Graph Theory and Cayley s Formula Chad Casarotto August 10, 2006 Contents 1 Introducton 1 2 Bascs and Defntons 1 Cayley s Formula 4 4 Prüfer Encodng A Forest of Trees 7 1 Introducton In ths paper, I wll
More informationSIX WAYS TO SOLVE A SIMPLE PROBLEM: FITTING A STRAIGHT LINE TO MEASUREMENT DATA
SIX WAYS TO SOLVE A SIMPLE PROBLEM: FITTING A STRAIGHT LINE TO MEASUREMENT DATA E. LAGENDIJK Department of Appled Physcs, Delft Unversty of Technology Lorentzweg 1, 68 CJ, The Netherlands Emal: e.lagendjk@tnw.tudelft.nl
More informationDEFINING %COMPLETE IN MICROSOFT PROJECT
CelersSystems DEFINING %COMPLETE IN MICROSOFT PROJECT PREPARED BY James E Aksel, PMP, PMISP, MVP For Addtonal Informaton about Earned Value Management Systems and reportng, please contact: CelersSystems,
More informationCalculation of Sampling Weights
Perre Foy Statstcs Canada 4 Calculaton of Samplng Weghts 4.1 OVERVIEW The basc sample desgn used n TIMSS Populatons 1 and 2 was a twostage stratfed cluster desgn. 1 The frst stage conssted of a sample
More informationRegression Models for a Binary Response Using EXCEL and JMP
SEMATECH 997 Statstcal Methods Symposum Austn Regresson Models for a Bnary Response Usng EXCEL and JMP Davd C. Trndade, Ph.D. STATTECH Consultng and Tranng n Appled Statstcs San Jose, CA Topcs Practcal
More informationAn Analysis of Factors Influencing the SelfRated Health of Elderly Chinese People
Open Journal of Socal Scences, 205, 3, 520 Publshed Onlne May 205 n ScRes. http://www.scrp.org/ournal/ss http://dx.do.org/0.4236/ss.205.35003 An Analyss of Factors Influencng the SelfRated Health of
More information8 Algorithm for Binary Searching in Trees
8 Algorthm for Bnary Searchng n Trees In ths secton we present our algorthm for bnary searchng n trees. A crucal observaton employed by the algorthm s that ths problem can be effcently solved when the
More informationANALYZING THE RELATIONSHIPS BETWEEN QUALITY, TIME, AND COST IN PROJECT MANAGEMENT DECISION MAKING
ANALYZING THE RELATIONSHIPS BETWEEN QUALITY, TIME, AND COST IN PROJECT MANAGEMENT DECISION MAKING Matthew J. Lberatore, Department of Management and Operatons, Vllanova Unversty, Vllanova, PA 19085, 6105194390,
More informationSTATISTICAL DATA ANALYSIS IN EXCEL
Mcroarray Center STATISTICAL DATA ANALYSIS IN EXCEL Lecture 6 Some Advanced Topcs Dr. Petr Nazarov 1401013 petr.nazarov@crpsante.lu Statstcal data analyss n Ecel. 6. Some advanced topcs Correcton for
More informationA Novel Methodology of Working Capital Management for Large. Public Constructions by Using Fuzzy Scurve Regression
Novel Methodology of Workng Captal Management for Large Publc Constructons by Usng Fuzzy Scurve Regresson ChengWu Chen, Morrs H. L. Wang and TngYa Hseh Department of Cvl Engneerng, Natonal Central Unversty,
More informationState function: eigenfunctions of hermitian operators> normalization, orthogonality completeness
Schroednger equaton Basc postulates of quantum mechancs. Operators: Hermtan operators, commutators State functon: egenfunctons of hermtan operators> normalzaton, orthogonalty completeness egenvalues and
More informationPassive Filters. References: Barbow (pp 265275), Hayes & Horowitz (pp 3260), Rizzoni (Chap. 6)
Passve Flters eferences: Barbow (pp 6575), Hayes & Horowtz (pp 360), zzon (Chap. 6) Frequencyselectve or flter crcuts pass to the output only those nput sgnals that are n a desred range of frequences (called
More information2.4 Bivariate distributions
page 28 2.4 Bvarate dstrbutons 2.4.1 Defntons Let X and Y be dscrete r.v.s defned on the same probablty space (S, F, P). Instead of treatng them separately, t s often necessary to thnk of them actng together
More information1 Approximation Algorithms
CME 305: Dscrete Mathematcs and Algorthms 1 Approxmaton Algorthms In lght of the apparent ntractablty of the problems we beleve not to le n P, t makes sense to pursue deas other than complete solutons
More informationTime Domain simulation of PD Propagation in XLPE Cables Considering Frequency Dependent Parameters
Internatonal Journal of Smart Grd and Clean Energy Tme Doman smulaton of PD Propagaton n XLPE Cables Consderng Frequency Dependent Parameters We Zhang a, Jan He b, Ln Tan b, Xuejun Lv b, HongJe L a *
More informationCalibration and Linear Regression Analysis: A SelfGuided Tutorial
Calbraton and Lnear Regresson Analyss: A SelfGuded Tutoral Part The Calbraton Curve, Correlaton Coeffcent and Confdence Lmts CHM314 Instrumental Analyss Department of Chemstry, Unversty of Toronto Dr.
More informationChapter XX More advanced approaches to the analysis of survey data. Gad Nathan Hebrew University Jerusalem, Israel. Abstract
Household Sample Surveys n Developng and Transton Countres Chapter More advanced approaches to the analyss of survey data Gad Nathan Hebrew Unversty Jerusalem, Israel Abstract In the present chapter, we
More informationQuality Adjustment of Secondhand Motor Vehicle Application of Hedonic Approach in Hong Kong s Consumer Price Index
Qualty Adustment of Secondhand Motor Vehcle Applcaton of Hedonc Approach n Hong Kong s Consumer Prce Index Prepared for the 14 th Meetng of the Ottawa Group on Prce Indces 20 22 May 2015, Tokyo, Japan
More informationEE201 Circuit Theory I 2015 Spring. Dr. Yılmaz KALKAN
EE201 Crcut Theory I 2015 Sprng Dr. Yılmaz KALKAN 1. Basc Concepts (Chapter 1 of Nlsson  3 Hrs.) Introducton, Current and Voltage, Power and Energy 2. Basc Laws (Chapter 2&3 of Nlsson  6 Hrs.) Voltage
More informationThe OC Curve of Attribute Acceptance Plans
The OC Curve of Attrbute Acceptance Plans The Operatng Characterstc (OC) curve descrbes the probablty of acceptng a lot as a functon of the lot s qualty. Fgure 1 shows a typcal OC Curve. 10 8 6 4 1 3 4
More informationTrafficlight a stress test for life insurance provisions
MEMORANDUM Date 006097 Authors Bengt von Bahr, Göran Ronge Traffclght a stress test for lfe nsurance provsons Fnansnspetonen P.O. Box 6750 SE113 85 Stocholm [Sveavägen 167] Tel +46 8 787 80 00 Fax
More informationFormula of Total Probability, Bayes Rule, and Applications
1 Formula of Total Probablty, Bayes Rule, and Applcatons Recall that for any event A, the par of events A and A has an ntersecton that s empty, whereas the unon A A represents the total populaton of nterest.
More informationLogistic Regression. Lecture 4: More classifiers and classes. Logistic regression. Adaboost. Optimization. Multiple class classification
Lecture 4: More classfers and classes C4B Machne Learnng Hlary 20 A. Zsserman Logstc regresson Loss functons revsted Adaboost Loss functons revsted Optmzaton Multple class classfcaton Logstc Regresson
More informationBrigid Mullany, Ph.D University of North Carolina, Charlotte
Evaluaton And Comparson Of The Dfferent Standards Used To Defne The Postonal Accuracy And Repeatablty Of Numercally Controlled Machnng Center Axes Brgd Mullany, Ph.D Unversty of North Carolna, Charlotte
More informationAn Investigation of the Performance of the Generalized SX 2 ItemFit Index for Polytomous IRT Models. Taehoon Kang Troy T. Chen
An Investgaton of the Performance of the eneralzed SX ItemFt Index for Polytomous IRT Models Taehoon Kang Troy T. Chen Abstract Orlando and Thssen (, 3) proposed an temft ndex, SX, for dchotomous
More informationCHAPTER 5 RELATIONSHIPS BETWEEN QUANTITATIVE VARIABLES
CHAPTER 5 RELATIONSHIPS BETWEEN QUANTITATIVE VARIABLES In ths chapter, we wll learn how to descrbe the relatonshp between two quanttatve varables. Remember (from Chapter 2) that the terms quanttatve varable
More informationMULTIPLE LINEAR REGRESSION IN MINITAB
MULTIPLE LINEAR REGRESSION IN MINITAB Ths document shows a complcated Mntab multple regresson. It ncludes descrptons of the Mntab commands, and the Mntab output s heavly annotated. Comments n { } are used
More informationStatistical Methods to Develop Rating Models
Statstcal Methods to Develop Ratng Models [Evelyn Hayden and Danel Porath, Österrechsche Natonalbank and Unversty of Appled Scences at Manz] Source: The Basel II Rsk Parameters Estmaton, Valdaton, and
More information1. Fundamentals of probability theory 2. Emergence of communication traffic 3. Stochastic & Markovian Processes (SP & MP)
6.3 /  Communcaton Networks II (Görg) SS20  www.comnets.unbremen.de Communcaton Networks II Contents. Fundamentals of probablty theory 2. Emergence of communcaton traffc 3. Stochastc & Markovan Processes
More informationThe Application of Fractional Brownian Motion in Option Pricing
Vol. 0, No. (05), pp. 738 http://dx.do.org/0.457/jmue.05.0..6 The Applcaton of Fractonal Brownan Moton n Opton Prcng Qngxn Zhou School of Basc Scence,arbn Unversty of Commerce,arbn zhouqngxn98@6.com
More informationv a 1 b 1 i, a 2 b 2 i,..., a n b n i.
SECTION 8.4 COMPLEX VECTOR SPACES AND INNER PRODUCTS 455 8.4 COMPLEX VECTOR SPACES AND INNER PRODUCTS All the vector spaces we have studed thus far n the text are real vector spaces snce the scalars are
More informationPRACTICE 1: MUTUAL FUNDS EVALUATION USING MATLAB.
PRACTICE 1: MUTUAL FUNDS EVALUATION USING MATLAB. INDEX 1. Load data usng the Edtor wndow and mfle 2. Learnng to save results from the Edtor wndow. 3. Computng the Sharpe Rato 4. Obtanng the Treynor Rato
More informationSensitivity Analysis in a Generic MultiAttribute Decision Support System
Senstvty Analyss n a Generc MultAttrbute Decson Support System Sxto RíosInsua, Antono Jménez and Alfonso Mateos Department of Artfcal Intellgence, Madrd Techncal Unversty Campus de Montegancedo s/n,
More informationRESEARCH DISCUSSION PAPER
Reserve Bank of Australa RESEARCH DISCUSSION PAPER Competton Between Payment Systems George Gardner and Andrew Stone RDP 200902 COMPETITION BETWEEN PAYMENT SYSTEMS George Gardner and Andrew Stone Research
More informationEfficient Project Portfolio as a tool for Enterprise Risk Management
Effcent Proect Portfolo as a tool for Enterprse Rsk Management Valentn O. Nkonov Ural State Techncal Unversty Growth Traectory Consultng Company January 5, 27 Effcent Proect Portfolo as a tool for Enterprse
More informationCS 2750 Machine Learning. Lecture 3. Density estimation. CS 2750 Machine Learning. Announcements
Lecture 3 Densty estmaton Mlos Hauskrecht mlos@cs.ptt.edu 5329 Sennott Square Next lecture: Matlab tutoral Announcements Rules for attendng the class: Regstered for credt Regstered for audt (only f there
More informationEstimation of Dispersion Parameters in GLMs with and without Random Effects
Mathematcal Statstcs Stockholm Unversty Estmaton of Dsperson Parameters n GLMs wth and wthout Random Effects Meng Ruoyan Examensarbete 2004:5 Postal address: Mathematcal Statstcs Dept. of Mathematcs Stockholm
More informationLecture 10: Linear Regression Approach, Assumptions and Diagnostics
Approach to Modelng I Lecture 1: Lnear Regresson Approach, Assumptons and Dagnostcs Sandy Eckel seckel@jhsph.edu 8 May 8 General approach for most statstcal modelng: Defne the populaton of nterest State
More informationAryabhata s Root Extraction Methods. Abhishek Parakh Louisiana State University Aug 31 st 2006
Aryabhata s Root Extracton Methods Abhshek Parakh Lousana State Unversty Aug 1 st 1 Introducton Ths artcle presents an analyss of the root extracton algorthms of Aryabhata gven n hs book Āryabhatīya [1,
More informationRisk Model of LongTerm Production Scheduling in Open Pit Gold Mining
Rsk Model of LongTerm Producton Schedulng n Open Pt Gold Mnng R Halatchev 1 and P Lever 2 ABSTRACT Open pt gold mnng s an mportant sector of the Australan mnng ndustry. It uses large amounts of nvestments,
More informationCharacterization of Assembly. Variation Analysis Methods. A Thesis. Presented to the. Department of Mechanical Engineering. Brigham Young University
Characterzaton of Assembly Varaton Analyss Methods A Thess Presented to the Department of Mechancal Engneerng Brgham Young Unversty In Partal Fulfllment of the Requrements for the Degree Master of Scence
More informationThe eigenvalue derivatives of linear damped systems
Control and Cybernetcs vol. 32 (2003) No. 4 The egenvalue dervatves of lnear damped systems by YeongJeu Sun Department of Electrcal Engneerng IShou Unversty Kaohsung, Tawan 840, R.O.C emal: yjsun@su.edu.tw
More informationHOUSEHOLDS DEBT BURDEN: AN ANALYSIS BASED ON MICROECONOMIC DATA*
HOUSEHOLDS DEBT BURDEN: AN ANALYSIS BASED ON MICROECONOMIC DATA* Luísa Farnha** 1. INTRODUCTION The rapd growth n Portuguese households ndebtedness n the past few years ncreased the concerns that debt
More information