Hello,
I have an unbalanced dataset of 6381 firm observations, from 23 countries and 152 industries measured over 6 years.
To my knowledge, three levels can be identified in an hierarchical model, namely firm year observations (1), industry (2) and country (3).
As I'm trying to examine whether cross-national differences occur between my DV (corporate social performance) and IV (board gender diversity moderated by three cultural values of Hofstede (moderating variables), all continuous variables, I'm trying to control for cluster effects on the industry level as it is not directly related to my research question.
My IV (BGD) and DV (CSP) are both measured at the firm level and the cultural value at the country values. Further, all control variables are also measured at the firm level, leaving no variables at the industry level.
However, the stata formula used produces the following error:
Code:
mixed CSP c.BGD##c.DMAS c.BGD##c.DPD c.BGD##c.DID ROE ROA TBQ LEV BINDEPD BSIZE BSTRUC CEODUA logTA i.Year || COUNTRY:BGD, vce(cluster INDUSTRY)
note: BGD omitted because of collinearity
note: BGD omitted because of collinearity
highest-level groups are not nested within INDUSTRY
What am I doing wrong if I want to control for the clusterseffects within the industry hierarchy?
Related Posts with Multi Level Modelling with clustered standard errors
Documentation for graphic schemesDear All, Is there a formal documentation for creating new graphics schemes? The Stata documentatio…
Too many weights errors …
showing mean of a variable over several variable and categoryIn short: I want to plot the mean of variable Y over variable X_1 to X_9. Each variable is a dummy …
Gravity model with GEPPML: General equilibrium analysis with PPMLDear All, I am applying the method of General equilibrium analysis with Poisson pseudo-maximum-like…
How to add Arellano-Bond and Sargan tests results to esttab optionsDear Statalisters, I am using Stata version 14.2. to run a two-step difference GMM using the follow…
Subscribe to:
Post Comments (Atom)
0 Response to Multi Level Modelling with clustered standard errors
Post a Comment