Sociology, Department of

 

First Advisor

Kirk Dombrowski

Second Advisor

Lisa Kort-Butler

Date of this Version

5-2017

Document Type

Article

Citation

Habecker, Patrick. 2017. "Who Do You Know: Improving and Exploring the Network Scale-Up Method." PhD dissertation, Department of Sociology, University of Nebraska, Lincoln, NE.

Comments

A DISSERTATION Presented to the Faculty of The Graduate College at the University of Nebraska In Partial Fulfillment of Requirements For the Degree of Doctor of Philosophy, Major: Sociology, Under the Supervision of Professors Kirk Dombrowski and Lisa Kort-Butler. Lincoln, Nebraska: May, 2017

Copyright © 2017 Patrick Habecker

Abstract

The purpose of this dissertation was to examine ways to improve and explore the network scale-up method (NSUM). This dissertation improved the NSUM by proposing a new mean of sums (MoS) estimation process, improving recursive back-estimation techniques, exploring how NSUM design changes effected estimates of personal network size, what predicts having larger personal networks, and the cognitive process used by participants taking a NSUM survey. Data was collected from an address-based survey (n=617) of Nebraskans conducted in 2014 and a series of cognitive interviews (n=19) conducted in 2016.

The MoS estimator better predicted the size of a target group than the traditional estimator. Further, recursive back-estimation was shown to retain more scaling variables when used with the MoS than the traditional estimator. However, the MoS estimator did produce larger average estimates of personal network size. The application of recursive back-estimation reduced the average of both the MoS and traditional estimates of personal network size to comparable levels. Differences in the treatment of item nonresponse among NSUM scaling questions had little to no impact on the average estimate of personal network size.

Eighteen different estimates of personal network size were calculated based upon different assumptions and methodological choices for regression models. In all eighteen models rural Nebraskans had larger networks than their urban counterparts, and those who made less than $25,000 had smaller networks than those who made between $50,000 and $99,999. In some models education, religious attendance, and age were associated with expected network size, but these associations were erratic. This shows that NSUM methodological decisions NSUM can have effects on both estimates of network size and statistical inference.

Finally, cognitive interviews revealed a series of issues around participants’ ability to accurately answer NSUM questions including memory search, definition retention, and differences between the known-population technique and the summation method. A series of suggestions for practical implementation and further testing of these issues are discussed. This dissertation demonstrates new ways to adapt the NSUM without having to use the generalized NSUM and explores how participants’ process NSUM style questions when developing their answers.

Advisors: Kirk Dombrowski and Lisa Kort-Butler

Share

COinS