* Encoding: UTF-8.
***Paste MI y EPC.
*** MI = Modification index.
*** EPC = Expected parameter change.

DATA LIST FREE /MI EPC.
BEGIN DATA
10.542 .216
7.682 -.182
7.529 .178
3.380 -.117
3.099 -.117
2.520 .110
1.443 .076
1.130 -.061
.269 -.032
.015 -.008

END DATA.

***D = Mínimum value assigned for a misspecification, according to Saris et al. (2009).
***For factor loadings, D = .40.
***For others parameters, D = .10.
COMPUTE D = .40.

*** for alpha = .10, x = 2.706.
*** for alpha = .05, x = 3.841.
*** for alpha =  .01, x = 5.024.
COMPUTE X = 3.841459.

** POWERMIN = Minimum statistical power asumed as high**.
*** According to Saris et al., equal or higher than .75.
*** For Cohen (1998), equal or higher than a .80.
COMPUTE POWERMIN = .75.

** EPCMIN = EPC considered as relevant**.
*** For Saris et al., equal to .20.
COMPUTE EPCMIN = .20.

** Do not modify this code**.
COMPUTE ERROR = EPC/SQR(MI).
COMPUTE Lim.2.CI = EPC + 1.96*(ERROR).
COMPUTE Lim.1.CI = EPC - 1.96*(ERROR).
COMPUTE NCP = (MI/(EPC*EPC))*(D*D).
COMPUTE DF = 1.
COMPUTE POWER = 1-NCDF.CHISQ(X,DF,NCP).
COMPUTE RATIO = abs(EPC/D).
STRING DECISION (A15).

COMPUTE DECISION = "I".
    if MI >= X and POWER <= POWERMIN and RATIO >= .99999 DECISION = "M".
    if MI >= X and POWER >= POWERMIN and abs(EPC) > EPCMIN DECISION = "M".    
    if MI <= X and POWER >= POWERMIN DECISION = "nM".
    if MI <= X and POWER <= POWERMIN DECISION = "I".
**M: Misspecification
**nM: No misspecification
**I: inconclusive

FORMATS EPC Lim.1.CI Lim.1.CI NCP POWER RATIO (F8.3).
LIST EPC Lim.1.CI Lim.2.CI  NCP POWER RATIO DECISION.