The Household, Income and Labour Dynamics in Australia (HILDA) dataset used for the analyses presented in this paper can be 
requested from the Melbourne Institute (https://melbourneinstitute.unimelb.edu.au/hilda/for-data-users/ordering-hilda-survey-data).

Three software codes accompany this paper:
(i)		“STATA Analysis of HILDA.do”: STATA code to derive variables from the HILDA dataset to be used for analysis, 
		and code of the GLLAMM procedure to estimate logistic mixed models either assuming normally distributed random effects
		or unspecified random effects distribution estimated using Gateaux derivative. 
(ii)	“VEM_Random Intercept logistic mixed model.r”: R code to perform VEM algorithm for random intercepts logistic model
(iii)	“VEM_Random Intercept and Random Slope logistic mixed model.r”: R code to perform VEM algorithm for random intercepts	
		and random slopes logistic model

Details about deriving the variables for analysis are described in the STATA code, however briefly:
	(i) 	marital status is derived from the HILDA variable 'mrcurr'; 
	(ii) 	youngest age of dependent children is derived from the HILDA variable 'hhd0_4' 
	(iii) 	Education at Baseline is derived from the HILDA variable 'edhigh' at wave 1
	(iv) 	Age at baseline is derived from the HILDA variable 'hgage' at wave 1
	(v) 	Outcome variable, employment status, is derived from the HILDA variable 'esbrd'
	(vi)	'xwaveid' is the unique subject ID in HILDA 
	(vii)	'wave' is the identifier of the study wave


  
