`test_performance` for `lavaan`

[rempsyc]

It would be desirable to eventually support lavaan fit objects (CFA/SEM) in performance::test_performance. Reprex below:

library(performance)
library(lavaan)

# Testing performance for lm models works
m1 <- lm(Sepal.Length ~ Petal.Length, data = iris)
m2 <- lm(Sepal.Length ~ Petal.Width, data = iris)
test_performance(m1, m2)
#> Name | Model |      BF
#> ----------------------
#> m1   |    lm |        
#> m2   |    lm | < 0.001
#> Each model is compared to m1.

# But not for lavaan models
HS.model <- ' visual  =~ x1 + x2 + x3
              textual =~ x4 + x5 + x6
              speed   =~ x7 + x8 + x9 '
fit <- cfa(HS.model, data = HolzingerSwineford1939)
HS.model2 <- ' visual  =~ x1 + x2
               textual =~ x4 + x5
               speed   =~ x7 + x8 + x9 '
fit2 <- cfa(HS.model2, data = HolzingerSwineford1939)
test_performance(fit, fit2)
#> Error: evaluation nested too deeply: infinite recursion / options(expressions=)?

# Yet it is possible to compare performance for lavaan models
compare_performance(fit, fit2)
#> # Comparison of Model Performance Indices
#> 
#> Name |  Model |   Chi2 | Chi2_df | p (Chi2) | Baseline | Baseline_df | p (Baseline) |   GFI |  AGFI |   NFI |  NNFI |   CFI | RMSEA |    RMSEA  CI | p (RMSEA) |   RMR |  SRMR |   RFI |  PNFI |   IFI |   RNI | Loglikelihood |      AIC | AIC weights |      BIC | BIC weights | BIC_adjusted
#> -----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#> fit  | lavaan | 85.306 |  24.000 |   < .001 |  918.852 |      36.000 |       < .001 | 0.943 | 0.894 | 0.907 | 0.896 | 0.931 | 0.092 | [0.07, 0.11] |    < .001 | 0.082 | 0.065 | 0.861 | 0.605 | 0.931 | 0.931 |     -3737.745 | 7517.490 |     < 0.001 | 7595.339 |     < 0.001 |     7528.739
#> fit2 | lavaan | 55.186 |  11.000 |   < .001 |  545.030 |      21.000 |       < .001 | 0.952 | 0.877 | 0.899 | 0.839 | 0.916 | 0.116 | [0.09, 0.15] |    < .001 | 0.077 | 0.063 | 0.807 | 0.471 | 0.917 | 0.916 |     -2991.866 | 6017.731 |        1.00 | 6080.752 |        1.00 |     6026.838

^{Created on 2022-05-25 by the reprex package (v2.0.1)}

(Interestingly, while the error above mentions evaluation nested too deeply, in my console I am getting a different error: Error: C stack usage 15928784 is too close to the limit.)

Given that it is possible to compare performance of lavaan models, would also testing performance of lavaan models be something implementable?

[ConnorEsterwood]

+1 to this for sure. Just ran into this limitation.

[DominiqueMakowski]

should be straightforward to add, just need someone to do the PR :)