what code can be used to see if a key variable is missing systematically or not?

[Niall Deng]

Hey guys, I have a panel data and after I drop out the observations that have missing values for key varibales, I find out when I drop obs that miss A, observation in abc years are dropped way much more than those in def years, while for B the things go to other ways around. It seems like the data collector did a bad job for certain variables in certain years. I wonder what code can be used to easily see if a key variable is missing systematically or not, and any useful suggestions for handling such problems?

Thanks a lot!

[Niall Deng]

Variable	Missing	Total	Percent Missing
output	311,897	747,519	41.72
added_output	374,721	747,519	50.13
Variable	Missing	Total	Percent Missing
output	73,445	743,918	9.87
added_output	96,809	743,918	13.01
Variable	Missing	Total	Percent Missing
output	47,103	746,503	6.31
added_output	56,378	746,503	7.55
Variable	Missing	Total	Percent Missing
output	36,792	691,379	5.32
added_output	52,490	691,379	7.59
Variable	Missing	Total	Percent Missing
output	193,918	690,559	28.08
added_output	253,403	690,559	36.7
Variable	Missing	Total	Percent Missing
output	196,167	703,452	27.89
added_output	259,737	703,452	36.92
Variable	Missing	Total	Percent Missing
output	217,201	702,195	30.93
added_output	262,402	702,195	37.37
Variable	Missing	Total	Percent Missing
output	205,302	700,586	29.3
added_output	573,797	700,586	81.9
Variable	Missing	Total	Percent Missing
output	0	608,677	0
added_output	0	608,677	0
Variable	Missing	Total	Percent Missing
output	137,607	450,035	30.58
added_output	173,885	450,035	38.64

this is the output of my mdesc code, you can see the data is doing generally well in table 2349, but bad in others. what's worse the two variable are missing very differently in different years, like table 8, output miss 29.3% while added_output miss 81.9%!. I do need both varibales in my model, so what should I do?