Ordered Sample Scatterplots for Displaying Survey Data

• Activity Number: 520
• Type: Contributed
• Date/Time: Wednesday, August 6, 2014 : 10:30 AM to 12:20 PM
• Sponsor: Survey Research Methods Section
• Author(s): Edward Mulrow*+ and Susan Hinkins and Fritz Scheuren
• Companies: NORC at the University of Chicago and NORC at the University of Chicago and NORC at the University of Chicago
• Keywords: Survey ; Graphics ; Sample Scatterplots

Abstract

Korn and Graubard (1998) suggested various modifications to make scatterplots more informative for sample survey data. These included the notion of sampled scatterplots - plotting a subsample of the data such that the resulting subsample is approximately a simple random sample. Hinkins et al (2009) used inverse sampling to construct regression diagnostic scatterplots, and illustrated how multi-panel plots of many inverse samples could be used to account for the loss of information when subsampling. One criticism of this approach is that the multi-panel displays may be harder to view; a smaller set of plots, or ideally a single plot, may be better. We explore using the ideas of Flury and Tarpey (1993) to order a large collection of curves. Using this technique on smoothers through scatterplots of survey data subsamples will provide a way to order scatterplots and therefore reduce the number, e.g. a minimum, median and maximum plot, or some other alternative based on order statistics.