How to interpret lavaan output?

[Marie-Sophie Ricoux]

HelloI'm new using lavaan and I have a lot of questions.

But firt, I struggle to lnow how to interpret the outputs given by the package.

I give you my model :

model <-'
Shannon ~ masse + BGI_win + BGI_nais + BGI_summ + poids_moyen + Nb_indiv + age_mere + masse_mere + Shannon_mere
masse ~ BGI_win + BGI_nais + BGI_summ + poids_moyen + Nb_indiv + age_mere + masse_mere + Shannon_mere
Shannon_mere ~ age_mere + masse_mere
masse_mere ~ age_mere
'

model_fitbis <- lavaan::cfa(model, data = donnees, test = "satorra.bentler", estimator = "MLM", se= "robust")

summary(model_fitbis, fit.measures = TRUE)

And there is the output :

lavaan 0.6-3 ended normally after 91 iterations

  Optimization method                           NLMINB
  Number of free parameters                         24

  Number of observations                            66

  Estimator                                         ML      Robust
  Model Fit Test Statistic                      19.878      24.905
  Degrees of freedom                                10          10
  P-value (Chi-square)                           0.030       0.006
  Scaling correction factor                                  0.798
    for the Satorra-Bentler correction

Model test baseline model:

  Minimum Function Test Statistic               64.839      79.546
  Degrees of freedom                                30          30
  P-value                                        0.000       0.000

User model versus baseline model:

  Comparative Fit Index (CFI)                    0.716       0.699
  Tucker-Lewis Index (TLI)                       0.149       0.098

  Robust Comparative Fit Index (CFI)                         0.705
  Robust Tucker-Lewis Index (TLI)                            0.116

Loglikelihood and Information Criteria:

  Loglikelihood user model (H0)               -349.232    -349.232
  Loglikelihood unrestricted model (H1)       -339.293    -339.293

  Number of free parameters                         24          24
  Akaike (AIC)                                 746.464     746.464
  Bayesian (BIC)                               799.016     799.016
  Sample-size adjusted Bayesian (BIC)          723.459     723.459

Root Mean Square Error of Approximation:

  RMSEA                                          0.122       0.150
  90 Percent Confidence Interval          0.036  0.201       0.068  0.235
  P-value RMSEA <= 0.05                          0.071       0.028

  Robust RMSEA                                               0.134
  90 Percent Confidence Interval                             0.069  0.202

Standardized Root Mean Square Residual:

  SRMR                                           0.069       0.069

Parameter Estimates:

  Information                                 Expected
  Information saturated (h1) model          Structured
  Standard Errors                           Robust.sem

Regressions:
                   Estimate  Std.Err  z-value  P(>|z|)
  Shannon ~                                           
    masse            -0.035    0.031   -1.150    0.250
    BGI_win           0.000    0.002    0.171    0.864
    BGI_nais         -0.002    0.002   -0.827    0.408
    BGI_summ         -0.001    0.001   -0.674    0.500
    poids_moyen       0.096    0.121    0.791    0.429
    Nb_indiv          0.003    0.001    3.934    0.000
    age_mere          0.080    0.025    3.194    0.001
    masse_mere        0.060    0.036    1.686    0.092
    Shannon_mere     -0.118    0.136   -0.864    0.387
  masse ~                                             
    BGI_win           0.018    0.006    2.958    0.003
    BGI_nais         -0.013    0.007   -1.702    0.089
    BGI_summ         -0.004    0.004   -0.977    0.329
    poids_moyen       0.377    0.445    0.849    0.396
    Nb_indiv         -0.004    0.005   -0.770    0.441
    age_mere          0.098    0.094    1.050    0.294
    masse_mere        0.043    0.123    0.345    0.730
    Shannon_mere     -1.235    0.546   -2.264    0.024
  Shannon_mere ~                                      
    age_mere          0.071    0.023    3.150    0.002
    masse_mere       -0.003    0.030   -0.109    0.913
  masse_mere ~                                        
    age_mere         -0.029    0.098   -0.299    0.765

Variances:
                   Estimate  Std.Err  z-value  P(>|z|)
   .Shannon           0.233    0.026    8.946    0.000
   .masse             3.540    0.564    6.272    0.000
   .Shannon_mere      0.194    0.020    9.591    0.000
   .masse_mere        2.895    0.407    7.115    0.000

Can someone help me and tell me what can I say with this output? (For example if my model is relevant or what variable influence the Shannon index?)

Thank you a lot For your help! And sorry for my english, I'm french.

[Deha deha]

Hi Marie-Sophie,

I am sorry to say that but your SEM model does not fit the data well. Roughly, RMSEA must be less than 0.08 but yours is 0.134, CFI and TLI must be above 0.90 but yours is 0.70 and 0.12. Here is a reference for goodness of fit indices: https://www.cscu.cornell.edu/news/Handouts/SEM_fit.pdf

You must re-adjust your model based on theory and research, try to add some correlations or reduce some associations or use a new estimator based on your data.. etc. Good luck!
Orhan



13 Mayıs 2019 Pazartesi 17:25:30 UTC+3 tarihinde Marie-Sophie Ricoux yazdı:

[Orhan Kaplan]

By the way, if your model does not fit the data well, there is no point to interpret it. If some goodness of fit indices were satisfactory for model-data fit, I think you could still interpret the variables and associations. However, with these indices, I would not recommend. Adjust the model and try to find a good fit first.

 

13 Mayıs 2019 Pazartesi 17:25:30 UTC+3 tarihinde Marie-Sophie Ricoux yazdı:

HelloI'm new using lavaan and I have a lot of questions.

[Marie-Sophie Ricoux]

Ok thanks but if it was good fiting well, how can I interpret the rest?

[Marie-Sophie Ricoux]

Thank you for your answer ! I'll look at the link you gave me.
Thank you again