i have this variable which is fixed broadband subscription but the data only started in 2000s. my study is time series using VAR method so i need a larger timeframe. the problem is, the said variables have so many missing datas from 1980-2000. what should i do about it? is there any test? do i need to drop the variable even tho it has datas from 2001-2023? thank you so much for the helpful answers!
Yeah, but thread carefully in this case, if the data for the variable doesn't exist because subscriptions really weren't a thing for instance during that time period then imputing your data doesn't make any sense.
But yes, there is a test, Little's missing completely at random test (MCAR), but if you uncover systematic missing-ness you can't really start imputing this amount of data as you can't control for well, any trends really, but also spurious problems too. Best option would then to either drop the variable or reduce/narrow the timeframe (which would be a problem as you needed the larger timeframe)
There are no tests that work without assumptions and those assumptions will often not hold in practice.
“all of these tests rely on some assumption that cannot be tested from the data. The paper notes that tests of the missingness mechanism are frequently misinterpreted and it clarifies the appropriate interpretation of such tests. […] Any justification for a particular approach to handling missing data must be come from substantive knowledge of the missingness process, not from the data.”
Rhoads, C. (2012) Problems with Tests of the Missingness Mechanism in Quantitative Policy Studies. Statistics, Politics and Policy, 3(1), Article 6.
- Badges
- Science topic