Table1 Table A1 - Panel 1

Table 2

Table 2:
Summary of Nonlinear Estimates and Comparisons Between Linear and Nonlinear Models for Ever Used Family Planning
 
 
Variables Social Interactions Represented by:
Village/Cluster Averages Network Averages
KDHS 1989 KDHS 1993 KDICP 1994/5 KDICP 1994/5
 
average for dependent variable 0.194 0.347 0.226 0.226
 
Panel A. Nonlinear Model Estimates
1 equilibrium level ye 0.178 0.328 0.222 0.186
2 direct effect, no social interaction, at ye 0.211 0.285 0.170 0.152
3 direct effect, with social interaction, at ye 0.147 0.115 0.102 0.114
4 total effect 0.191 0.201 0.149 0.138
5 effect of intensifying social interaction on total effect -0.055 -0.029 -0.042 -0.036
6 % bias in est. direct program. effect if social interactions ignored 43.5% 147.8% 77.7% 33.3%
7 total effect relative to direct effect (= 100% x social multiplier) 130% 175% 146% 121%
8 effect of intensifying social interaction on multiplier effect 0.104 0.357 0.165 0.144
 
Panel B. (Linear Model Estimates -Nonlinear Estimates)/Nonlinear Estimates in %
1 equilibrium level ye 9 5.8 1.8 6.9
2 direct effect, no social interaction, at ye -7.1 -1.4 -7.4 21
3 direct effect, with social interaction, at ye -20.1 -5.2 -2.9 27.4
4 total effect -16.1 -4.2 0.7 33.7
5 effect of increasing the impact of social interaction on total effect -210.9 -527.2 -281 -247.2
 
Panel C. Total Effects and in Nonlinear Model for High- and Low Education Subpopulation
1 total effect Low education
High education
0.164
0.408
0.189
0.277
0.137
0.237
0.130
0.229
2 total effect relative to direct effect
Low education
High education
126%
163%
170%
203%
142%
173%

140%
170%

 

Table1 Table A1 - Panel 1

Empirical Assessments of Social Networks, Fertility and Family Planning Programs: Nonlinearities and their Implications
Hans-Peter Kohler, Jere R. Behrman, Susan Cotts Watkins
© 2000 Max-Planck-Gesellschaft ISSN 1435-9871
http://www.demographic-research.org/Volumes/Vol3/7