1
$\begingroup$

I am getting confused on how to approach structural equation modeling. I have read some articles and textbooks about SEM using the lavaan and semPlot package though I'm getting mixed up with interpretation.

By looking at my lavaan output, it seems that my variables are significant predictors.

Regressions:
                              Estimate  Std.Err  z-value  P(>|z|)   Std.lv
  RBR_Index_Rep_used ~                                                    
    DEV_c_tld_nw_1              -0.346    0.021  -16.161    0.000   -0.346
    DEV_ci_Uniq_1                0.289    0.037    7.781    0.000    0.289
    DEV_c_Wll_t__1               0.375    0.046    8.211    0.000    0.375
  PPS_Index_Rep_used_unique ~                                             
    DEV_c_tld_nw_1               0.770    0.043   18.083    0.000    0.770
    DEV_ci_Uniq_1               -0.224    0.074   -3.034    0.002   -0.224
    DEV_c_Wll_t__1               1.465    0.091   16.125    0.000    1.465
  CEI_pps_Rep_used_unique ~                                               
    DEV_c_tld_nw_1               0.309    0.048    6.435    0.000    0.309
    DEV_ci_Uniq_1                0.073    0.083    0.873    0.383    0.073
    DEV_c_Wll_t__1               1.823    0.102   17.794    0.000    1.823

I created a semPlot to better visualize the SEM. However, I am confused at this point.

#Plot SEM diagram
semPaths(fit,'std', style="lisrel",layout="tree2", edge.label.cex = 1, 
         label.prop=0.9, rotation = 4, nCharNodes=0,
         equalizeManifests = FALSE, optimizeLatRes = TRUE, node.width = 4, 
         edge.width = 0.6, shapeMan = "rectangle", shapeLat = "ellipse", 
         shapeInt = "triangle", sizeMan = 5, sizeInt = 2, sizeLat = 4, 
         curve=2, unCol = "#070b8c")

enter image description here

Does the negative (red) path coefficients mean that there is a bad relationship/prediction between the variables, contridicting the lavaan output? Where in my lavaan output are these path coefficients? What are these dotted curved lines vs non-dotted curved lines? What the numbers next to the represent?

Improve this question
$\endgroup$
1
  • $\begingroup$ From what I can tell, the red indicates a negative effect, while green indicates positive; and the shade indicates the strength of this effect. This is why the two associations that your output estimates as negative yield red lines - the second of these (the middle line from left to right) is just too faint for you to recognize it as red. $\endgroup$
    – user281754
    Commented Feb 4, 2021 at 8:47

1 Answer 1

Reset to default
1
$\begingroup$

The red lines show negative effects (i.e. as x increases, y decreases). The green lines show contango effects (as x increases, y increases). To see the faint line add in your syntax:

fade = FALSE

Improve this answer
$\endgroup$
1
  • $\begingroup$ Is a "contago effect" simply monotonicity? $\endgroup$
    – Galen
    Commented Feb 11, 2022 at 4:16

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.