New answers tagged

1 vote
Accepted

Comparing regression coefficients across non-nested SEM models (Lavaan)

can I directly compare the regression variables from cognition ~ CR1 / cognition ~ CR2 using the standardized coefficients It ...
Terrence's user avatar
  • 2,138
6 votes

SEM with 50% missing data (due to distribution of items over various survey ballots/waves)

Regarding this comment: Therefore, intuitively, using list-wise deletion does not seem very problematic to me, since most missingness should be MCAR (they result from randomization in the ...
Shawn Hemelstrand's user avatar
4 votes
Accepted

Stable violation of faithfulness

Conditioning on the child of a confounder does not (in general) block the path. The part in red above rests on the assumption that observing $B$ lets us infer the value of $A$. This is not the case if ...
Scriddie's user avatar
  • 2,289
10 votes

What sample size should be reported when using pairwise deletion approach with WLSMV?

You should report what happened. There isn't "a sample size"; there are multiple sample sizes. You might write something like "Data were collected from 645 respondents in the method ...
Jeremy Miles's user avatar
  • 17.9k
1 vote

Mediation: only c path and b path are significant?

First, I agree with Shawn's comment (both parts). And the FAQ for PROCESS also agrees, saying that PROCESS is for observed ...
Peter Flom's user avatar
  • 120k
1 vote

Structural equation modeling identification question

Preamble The topic of identification has been covered countless times on CV (e.g., here, and here), so I will keep my response short. Answer As you may know, a structural equation model (SEM) is ...
Preston Botter's user avatar
0 votes

Matrix Covariance Algebra

Hint For random vectors $u, v$ and fixed conformative matrices $A, B$, one has $$ \operatorname{Cov}(Au, Bv) = A\operatorname{Cov}(u, v)B^\top.$$ The proof is a straightforward check of the definition ...
Zhanxiong's user avatar
  • 18.9k
7 votes

How to analyze a dichotomous outcome with 50% missing data?

First, I don't see how variables at time 2 can be sensibly used to predict dropout at time 2. So, I don't think you need to even worry about the missing data. I'm not sure why you collected it. You ...
Peter Flom's user avatar
  • 120k
8 votes

How to analyze a dichotomous outcome with 50% missing data?

If you want to predict what happens at T2 from data at T1, you could run a three classes model with "dropout", "no dropout", and "T2 missing" as the classes. Note that ...
Christian Hennig's user avatar
11 votes
Accepted

How to analyze a dichotomous outcome with 50% missing data?

Your description implies that you committed the mortal sin of not pre-specifying the final model in the sense that you tried different models in a way not unlike stepwise variable selection does (...
Frank Harrell's user avatar
5 votes
Accepted

moderated mediation vs mediated moderation

Short answer: different parameters of the same model. Read this great paper by Preacher et al. (2007) for details: https://doi.org/10.1080/00273170701341316 You have to consider what is the adjective,...
Terrence's user avatar
  • 2,138

Top 50 recent answers are included