

Parallel process, piecewise growth model 

Message/Author 

jmquinn posted on Tuesday, May 22, 2018  10:16 am



Good afternoon, I apologize if this has been discussed previously, but I did my best to search the forum before posting. I'm attempting to fit the following 6wave parallel process growth model: ipf1 spf1  W1_pf@0 W2_pf@1; spf1@0; ipf2 spf2  W3_pf@0 W4_pf@1 W5_pf*2 W6_pf*3; ior sor qor  w1_or@0 w2_or@1 w3_or@2 w4_or@3 w5_or@4 w6_or@5; I am aware that spf1 cannot covary with other growth parameters by default since its variance is fixed to 0 for model identification. However: Am I able to include spf1 in ON statements? If so, are there any conventions on how to calculate its R^2? I am interested in the effects of spf1 ON ior/sor/qor, since there is theory on why these relations would be in this direction. Thanks for your help! 


Q1: Yes. Q2: R2=1, but should be ignored  you are merely allowing spf1 to have different values as a function of covariates. 

Jamie Quinn posted on Wednesday, May 23, 2018  1:27 pm



Thank you, Bengt! (I am responding on behalf of jmquinn above) Because spf1 is included in an ON statement, should I use the STDYX output when interpreting this coefficient, or does one strictly use the unstandardized output when running latent growth models? I specifically ask in this case because the STDYX estimate of spf1 ON sor is 1.65 (se = .710, p = .020). How do I interpret a standardized coefficient that is greater than 1? Your FAQ page pointed me to the papers by Joreskog (1999) and Deegan (1978), but in my interpretation (of Deegan particularly), this seems more important for standardized loadings on oblique factors and for coefficients in a path analyses. Does this mean that spf1 and sor have a high degree of multicollinearity, and what does that substantively mean if they are latent factors? Thank you for time and for helping me to understand this phenomenon. 


We need to see the full output  send to Mplus Support along with your license number. 

Back to top 

