Using Statistical Process Control to Understand Variation in Computer Assisted Personal Interviewing Data

Robyn Sirkis1* Matt Jans2

1U.S. Census Bureau, 4600 Silver Hill Road, Washington, DC 20233

2U.S. Census Bureau, 4600 Silver Hill Road, Washington, DC 20233

QC 34

Abstract

This paper discusses research using statistical process control with paradata obtained during data collection for the National Health Interview Survey (NHIS). Statistical process control (SPC) involves using statistical techniques to measure and analyze variation in operational processes.

 The goal with this approach is not to simply monitor, but to improve the quality of the process over time. For this paper, we group interviewers into statistical clusters based on census tract level housing unit and respondent demographic characteristics and produce control charts which examine the variation of the process over time for each cluster.

We compare the means of interviewer performance indicators within each cluster to determine if they are significantly different from the overall mean of the process for the cluster, and examine some of the potential causes of process variation using selected control charts.

We address advanced SPC techniques such as multivariate charting. Indicators of data quality used in the paper are item nonresponse and interview duration. The charts are intended to demonstrate how survey managers can use paradata to monitor the data collection process using SPC principles and techniques.  

Key Words:

Paradata, Statistical Process Control, NHIS