I am trying to include an important variable as an otucome/ dependent in my study for Structural equation modeling/PATH analysis(for example, asking plans to quit smoking cigar. If the respondent has already quit smoking, then this question results in an Inapplicable response. The item has more than 90% missing data due to --1 (Inapplicable) : This value is coded to identify items that were validly skipped on a route specific to a respondent. Does the PATH recommend any strategy to account for these 'missing data' through weights or I completely exclude them from analyses or use multiple imputation?
Advance thanks for the responses and guidance