MCAR, MAR and MNAR describe missingness assumptions, not recovered outcomes
Read what missing observations may depend on without using a label to invent follow-up results or select a statistical repair.
No Vape · Updated

Locate the missing quantity before the acronym
Find whether the report concerns absent follow-up outcomes, unanswered items or another missing variable. Also identify the information actually observed. A sentence saying data were missing does not tell you why. Different quantities in the same project can need different accounts, so avoid assigning one unexplained acronym to every blank.
AHRQ’s 2018 registry methods chapter describes the three mechanisms and emphasizes that their consequences depend on the underlying relationships. We use its introductory definitions and caution only. We do not supply a missing-data method, diagnose the mechanism in a named study or claim that a label alone determines whether an estimate is biased.
Make the conditioning in MAR explicit
The ordinary phrase at random can sound like missingness happens for no reason. That is not what MAR necessarily says. Write the observed information the authors rely on, then ask what further relationship with the unobserved value they assume absent. Keep that as a stated assumption rather than something a software command has proved.
In a fictional contact exercise, follow-up response differs by a scheduling characteristic already recorded. One proposed account says that within each recorded scheduling group, the unseen answer does not further affect response. Another says the unseen answer still matters within those groups. The first illustrates the conditional structure of MAR; the second leaves a possible MNAR account. Neither is a claim about an actual quitting study.
Do not let a reason code settle the whole mechanism
A reported reason for nonresponse can be informative without establishing every relevant relationship. People with the same recorded reason may differ in unobserved ways. Likewise, the proportion missing by itself does not tell you which mechanism holds. Read the authors’ rationale and what remains untestable from the observed information.
A reader can ask what evidence or contextual knowledge supports the assumption and whether reported sensitivity analyses explore a meaningful alternative. Those checks do not recover the unknown answers. Avoid choosing whichever label makes the headline easiest to defend or declaring that all analyses sharing a label must have identical limitations.
Preserve an assumption note instead of recalculating
Write the missing variable, observed information invoked, assumption stated and alternative the report considers. If the acronym is unexplained, keep it as an unresolved methods question. A consequential health interpretation belongs with qualified expertise and the complete evidence, rather than a reader substituting a preferred set of missing outcomes.
This conceptual map is distinct from identifying an imputation method or the denominator of a percentage. A study can use both an assumption and a method, but naming one does not explain the other. Personal blank days remain unknown unless actual observations can be recovered; research terminology is not permission to convert them into successful or unsuccessful days.
Sources and scope
Sources inform the guidance. The worksheets and examples are No Vape editorial suggestions, not tested treatments.
Written with AI assistance and checked against the cited sources. No clinical review is claimed. This guide covers practical planning, not individual medical treatment. How we research and write