INTRO:Planning list wise deletion of missing data from large data set (7772). Data is from online survey, participation self-select. Q's could be skipped (none mandatory). Repeated measures resulted from 'select all apply' answer option on several q's - each option became an individual variable in the dataset. Some q's had prerequisite condition (answer) so not relevant to all participants. Am assuming MAR or MCAR - data is de-identified I cannot contact participants.
PROBLEM: I planned my list wise deletion as per below (includes all variables in data set) - At stage 5 I find: 1) both uni variables (author uni; degree holder uni) have lots of missing data - much more than any other vars; 2) another var (media fl-up) didn't have other/none option so can't tell if missing because skipped or because no media fl-up.
SO SHOULD I: 1) a. remove the uni variables altogether - I am not interested in analysis by uni anyway and if I delete I am losing large number of smaller author group which is of key interest (stage 6). b. delete the missing to follow previous stages...and lose then N..
2) a. just ignore missing data for media fl-up as can't be identified b. delete the missing to follow previous stages...and lose then N..
OR do something else completely. any input appreciated.
Stage 1 - single measure IVs - relevant to all: identified variable with highest missing and deleted:
Variable N miss Ntotal % total N AFTER DELETION
Role title 623 7772 8.02% 7149
Study 36 7149 0.50% 7113
State 19 7113 0.27% 7094
Business 5 7094 0.07% 7089
Stage 2 - single measure DV relevant to all: none to delete
Variable N miss Ntotal % total N AFTER DELETION
(3) 0 7089 0.00% 7089
Stage 3 - repeated measure IV relevant to all:1 variable in this category - deleted missing
Variable N miss Ntotal % total N AFTER DELETION
Employment 35 7089 0.49% 7054
Stage 4 - repeated measure DV relevant to all: none to delete
Variable N miss Ntotal % total N AFTER DELETION
(5) 0 7054 0.00% 7054
Stage 5 - repeated measure IV relevant to some (total = total yes on prerequisite q)
Variable N miss Ntotal % total
Degree & Uni 924 4457 20.73%
Study & inst 8 1834 0.44%
Bus. & type 12 1904 0.63%
Stage 6 - repeated measure DV relevant to some (total = total yes on prerequisite q)
Variable N miss Ntotal % total
Authors & uni 103 292 35.27%
Auth & pub fl-up 6 292 2.05%
Auth & media fl-up 171 292 58.56%
Author & dash 9 292 3.08%