logistic regression in lavaan

[Karolina Ścigała]

Hi,

I have a question about the link = "logit" function. I can see it in the manual, but when I try running t in R, I get this message:

mod <- sem(population.model, data=data, ordered = "cheat1", link = "logit", estimator = "MML")

Error in lav_model_gradient_mml(lavmodel = lavmodel, GLIST = GLIST, THETA = THETA[[g]],  : 
  logit link not implemented yet; use probit"



My problem is that I am not sure how to interpret effect sizes when using probit regression, and therefore I would prefer using logistic regression. Is there any way to do this? If not, is there a straightforward way to obtain and interpret effect sizes from probit regression? 

Looking forward to hear from you. 

Best,
Karolina

[jpma...@gmail.com]

Hi Karolina,

You can get approximate logits from probits as  Beta_logit=1.7xBeta_probit. Then, OR =exp(Beta_logit).

See if this works ok for you.

Best,

João Marôco

--
You received this message because you are subscribed to the Google Groups "lavaan" group.
To unsubscribe from this group and stop receiving emails from it, send an email to lavaan+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/lavaan/26a476b9-9ba0-48b2-93a2-e5f10917fc56%40googlegroups.com.

[Karolina Ścigała]

Hi João,

Thank you so much! I think this will solve my problem. I have two follow-up questions:

1. I can also apply this formula to the 95% CIs for the betas, right?

By this I mean:

- first calculate the 95% CI of the beta logit (= 1.7x 95% CI of the beta probit)

- then exponentiate it to get the CIs for the odds ratios

Or will this cause some problems?

2. Can I apply the same formula to standardized beta coefficients? I think it shouldn't be a problem, but just to make sure.

Again, thanks very much

Best,

Karolina

On Thursday, 5 December 2019 09:30:46 UTC+1, jpma...@gmail.com wrote:

Hi Karolina,

You can get approximate logits from probits as  Beta_logit=1.7xBeta_probit. Then, OR =exp(Beta_logit).

See if this works ok for you.

Best,

João Marôco

 

From: lav...@googlegroups.com <lav...@googlegroups.com> On Behalf Of Karolina Scigala
Sent: 4 de dezembro de 2019 17:06
To: lavaan <lav...@googlegroups.com>
Subject: logistic regression in lavaan

 

Hi,

 

I have a question about the link = "logit" function. I can see it in the manual, but when I try running t in R, I get this message:

 

 

mod <- sem(population.model, data=data, ordered = "cheat1", link = "logit", estimator = "MML")

Error in lav_model_gradient_mml(lavmodel = lavmodel, GLIST = GLIST, THETA = THETA[[g]],  : 

  logit link not implemented yet; use probit"



My problem is that I am not sure how to interpret effect sizes when using probit regression, and therefore I would prefer using logistic regression. Is there any way to do this? If not, is there a straightforward way to obtain and interpret effect sizes from probit regression? 

Looking forward to hear from you. 

Best,
Karolina

--
You received this message because you are subscribed to the Google Groups "lavaan" group.

To unsubscribe from this group and stop receiving emails from it, send an email to lav...@googlegroups.com.

[jpma...@gmail.com]

Hi Karolina,

The formula for the CI for OR is

 

But if you don’t have the cell counts (a,b,c,d,e), than you can approximate from the CI for the betas.

To unsubscribe from this group and stop receiving emails from it, send an email to lavaan+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/lavaan/9118c3a7-6832-4038-9a83-f49e2b5fcca8%40googlegroups.com.

[Karolina Ścigała]

It is working well! Thank you so much.

Best,

Karolina

To view this discussion on the web visit https://groups.google.com/d/msgid/lavaan/9118c3a7-6832-4038-9a83-f49e2b5fcca8%40googlegroups.com.