Generation Of A Socioeconomic Status Classifier From Kalamazoo County Postpartum Survey Responses
BACKGROUND: In Kalamazoo County, the death rate among black infants is 3.2 times that of white infants. Medical records review and phone surveys were conducted for Kalamazoo County women recruited from postpartum floors of the two delivery hospitals. Data were collected for prenatal, delivery, postpartum care, and material-social health determinants. Numerous items measured various aspects of SES.
GOAL: The goal of this analysis was to condense this rich set of SES information into a lesser number of dimensions with minimal information loss.
METHODS: Multiple Correspondence Analysis (MCA) followed by scree plots for randomly generated copies of SES responses were used to determine the data dimensionality. Ward cluster analysis was then performed on the MCA-generated dimensions, and frequency distributions for the clusters were obtained across demographics to describe them.
Results: Scree plots of the MCA output identified three dimensions describing SES. Ward cluster analysis recognized three distinct groups. Upon further quantitative analysis and community stakeholder feedback, it was determined that these three clusters represented meaningful SES grouping:
1. 44.7% (109) "Enough" - Adequate SES resources
2. 47.1% (115) "On the edge" - Low to medium income with safety net support
3. 7.8% (19) "Deep poverty" - Very low income and little support
DISCUSSION: Utilizing MCA with Ward cluster analysis provided a useful tool to generate a single and meaningful classification element. Further analysis elucidated distinct SES classes of Kalamazoo County mothers, creating an encompassing metric useful for analysis and interpretation of pertinent data.