Missing Data Pattern Analysis

Missing data pattern analysis tabulates which combinations of variables are missing together, the missing rate of each variable, and how strongly missingness co-occurs across variables.

결측 유형 분석은 어떤 변수 조합이 함께 비어 있는지, 변수별 결측률은 얼마인지, 결측이 서로 얼마나 함께 나타나는지를 표로 정리하는 방법이다.

How do I describe the amount and structure of missing data in my dataset before analysis?

분석 전에 자료의 결측이 얼마나, 어떤 모양으로 있는지 어떻게 정리하나요?

A missing-data pattern table lists each distinct combination of observed and missing variables with its frequency, alongside the per-variable missing percentage. Co-missingness counts and correlations between missingness indicators show whether variables tend to drop out together, and a monotone pattern indicates dropout-type missingness.

When to use Missing Data Pattern Analysis

• You need to report missing-data extent in a methods section
• You are choosing between deletion and imputation
• You suspect dropout or skip patterns in a questionnaire

When to use something else

• You need a formal test of randomness; use Little's MCAR test instead
• You want to fill in the missing values; use multiple imputation instead

Assumptions and requirements

• Missing values are coded as empty cells rather than placeholder numbers
• Columns are listed in a meaningful order for reading patterns

What the analysis reports

• Pattern table with counts and percentages
• Missing percentage per variable
• Number and share of complete cases
• Co-missingness counts and phi coefficients
• Monotone pattern flag

How to report it

Of [N] cases, [n] ([value]%) had complete data; [value]% of cells were missing across [k] patterns, with the highest rate for [variable] ([value]%).

Related methods

• Little's MCAR Test
• Multiple Imputation (MICE) with Pooled Regression

Run Missing Data Pattern Analysis on Quorant