Challenges relating to text narratives | Challenges relating to NVDRS/state VDRS data | |
---|---|---|
Challenge 1: Lack of or limited information on contextual factors relevant to deaths or populations being investigated | Limited or insufficient details on relevant demographic and circumstantial data in the narratives for case identification and characterization of deaths Blank or uninformative narratives (e.g., contents have little or no descriptions of circumstances around death) | Certain demographic and circumstance variables are not routinely collected or readily available for the researchers (e.g., medical records, data on child abuse) Missingness in certain demographic and circumstance variables (e.g., mental health history) Missing or incomplete data due to ongoing investigations or deaths occurring in states different from state of residence |
Challenge 2: Variation in reporting, coding and abstraction, and narrative information potential | Narratives are collected from informants and third parties (e.g., family/friends of the decedents) and limited to information known by coroner/medical examiner and law enforcement Contents of narratives depend on the interpretations and/or information deemed relevant by the abstractors Human errors in coding/abstraction process Level of details, missingness, and conflicting information from coroner/medical examiner and law enforcement narratives Variations in length, depth, completeness, and availability across county/state, characteristics of the decedent, and types of death could limit data-sharing | Differences in data collection, availability, coding and abstraction procedures across counties/states Variation in the degree of missingness depending on the nature of the data (e.g., toxicology and sensitive topics such as intimate partner violence are frequently missing) Inconsistencies between different data sources (e.g., data collected from family, friends, coroners/medical examiners, and law enforcement) Inconsistencies between the narratives and coded variables |