jacob-long / dpm Goto Github PK

Hi (sorry but to preface I have not written a github issue before): I have some imputed dataframes in a list and I'd like to run dpm over them with:

dpm.pidideo.news.out <- lapply(a.out.list, function(x){ dpm(newsideoattn2 ~ lag(pidideo), data = x)})

But I get the error:
Error: Can't convert NULL to a string
Backtrace:

1. base::lapply(...)
2. global::FUN(X[[i]], ...)
3. dpm::dpm(newsideoattn2 ~ lag(pidideo), data = x)
4. rlang::as_name(enexpr(id))
5. rlang::as_string(x)
6. rlang:::abort_coercion(x, "a string")

Do you know why this might be or how I can use dpm with multiply imputed dataframes generally? This same setup works with plm.
Thank you for your time.

Sam

Edit:
Nevermind I figured out that making the dataframes into panel_data and assigning id and waves resolved the issue. I'm not sure how to delete this comment. Thank you for your time and work in producing such useful packages.

Hi Prof. Long,
I'm looking to pool dpm results from many multiply imputed datasets. I have created a list of type "S4(dpm;;dpm)" objects using the lapply function in conjunction with the dpm call. The issue is that I cannot use summary(pool("list of S4 objects")) like I normally can with a list of lavaan outputs. When I use mice::pool on the list of dpm model outputs, I receive the error:

"Error: $ operator not defined for this S4 class"

Do you know how I might go about pooling my results using Rubins rules with a list of dpm model outputs?
By the way, thanks again for this excellent package. It's saved me from many opportunities to introduce human errors using lavaan syntax.

Sam

Hi, thanks for creating the great package. I was wondering if there is any mechanism via dpm to do post-estimation tests for serial correlation, equivalent to the Arellano–Bond Test of Serial Correlation used in GMM products (e.g. mtest for plm::pgmm

As of now you get stuck with duplicate model element errors. Have to make sure this makes sense within the model.

Hi Jacob,
Thanks for this excellent package. I was wondering though if dmp allows for the inclusion of wave or time dummy variables, as in a typical two-way fixed effect specification. I tried including the time variable specified in the panel_data argument as an independent variable, but I got the error: 'Error in lav_data_full(data = data, group = group, cluster = cluster, : lavaan ERROR: some variables have no values (only missings) or no variance'.
Many thanks in advance,
Matias