Content uploaded by Matthew E. Brashears
Author content
All content in this area was uploaded by Matthew E. Brashears on May 27, 2014
Content may be subject to copyright.
ONLINE SUPPLEMENT
to article in
AMERICAN SOCIOLOGICAL REVIEW, 2006, VOL. 71 (JUNE:353–375)
Social Isolation in America: Changes in Core Discussion Networks over Two Decades
Miller McPherson
University of Arizona and Duke University
Lynn Smith-Lovin
Duke University
Matthew E. Brashears
University of Arizona
APPENDIX TABLE OF CONTENTS
Table A1. Influence of Proportion Kin on Network Density and Heterogeneity
Table A2. Basic Parameters of Discussion Networks: 1985, 1987 and 2004
Table A3. Number of Respondents That Had Various Relationships with at Least One Discussion Partner
Table A4. Respondents with Spouses Only, Non-Spouse Kin, and Non-Kin in the Network
Table A5. Differences by Age, Education, Sex and Race in Network Size and Kin/Nonkin Composition
Table A6. Cooperativeness as Rated by the GSS Interviewer in 1985 and 2004
Table A7. Negative Binomial Regression of Size of Discussion Network
Table A8. Logistic Regression Coefficients for Having Someone to Talk to About Important Matters
ONLINE SUPPLEMENT
to article in
AMERICAN SOCIOLOGICAL REVIEW, 2006, VOL. 71 (JUNE:353–375)
McPherson, Smith-Lovin, & Brashears
“Social Isolation in America: Changes in Core Discussion Networks over Two Decades”
2
Table A1. Influence of Proportion Kin on Network Density and Heterogeneity (OLS Regression of Density and Heterogeneity on Proportion Kin)
Unstandardized Regression
Coefficient for Proportion Kin Predicted Value at Proportion Kin of R-squared
0 1
Dependent Variable 85 04 85 04 85 04 85 04
Density .36 .26 .42 .49 .78 .78 .18 .10
Age Heteregeneity (S.D. of alters’ ages) 5.48 7.53 7.30 6.56 13.54 13.88 .09 .15
Education Heterogeneity (S.D. of alters’
educations) .30 .20NS 1.61 1.38 1.92 1.58 .01 .001
Race Heterogeneity (IQV) -.08 -.12 .09 .15 .01 .03 .03 .04
Sex Heterogeneity (IQV) .27 .27 .54 .53 .79 .80 .06 .06
Note: All coefficients significant, p < .01, except where indicated.
ONLINE SUPPLEMENT
to article in
AMERICAN SOCIOLOGICAL REVIEW, 2006, VOL. 71 (JUNE:353–375)
McPherson, Smith-Lovin, & Brashears
“Social Isolation in America: Changes in Core Discussion Networks over Two Decades”
3
Table A2. Basic Parameters of Discussion Networks: 1985, 1987 and 2004
1985 (N=1531a) 1987 (N=1800) 2004 (N=1467b)
Network Size
0 10.0% 4.51% 24.6%
1 15.0% 14.27% 19.0%
2 16.2% 19.02% 19.2%
3 20.3% 49.06% 16.9%
4 14.8% 6.78% 8.8%
5 18.2% 3.20% 6.5%
6+ 5.4% 3.12% 4.9%
Mean 2.94 2.63 2.08
Mode 3 3 0
S.D. 1.79 1.28 1.81
Kin Network Size
0 26.4% 22.25% 42.4%
1 29.6% 35.70% 29.0%
2 21.6% 25.67% 14.5%
3 12.6% 16.39% 9.1%
4 6.3% 0.00% 3.5%
5c3.3% 0.00% 1.4%
Mean 1.44 1.36 1.12
Mode 1 1 0
S.D. 1.41 1.00 1.38
Non-kin Network Size
0 36.4% 36.79% 54.1%
1 22.2% 26.99% 21.3%
2 18.9% 20.03% 14.0%
3 13.0% 11.14% 5.7%
4 6.3% 3.50% 3.3%
5 3.3% 1.54% 1.5%
Mean 1.42 1.22 .88
Mode 1 0 0
S.D. 1.57 1.22 1.40
Proportion Kin
0 19.2% 18.57% 21.0%
.01-.33 15.4% 4.44% 10.9%
.34-.66 20.7% 25.71% 19.4%
.67-.99 14.5% 17.48% 11.6%
1 30.2% 33.80% 37.1%
Mean .53 .57 .60
S.D. .44 .38 .53
a Proportion kin is defined only for those who have at least one alter. Therefore, the N = 1395 for the proportion kin
analysis.
b N = 1070 for the analysis of proportion kin.
c Information on kinship was collected only on the first five alters cited. Therefore, the sum of the kin and non-kin alters is
not equal to the overall network size distribution.
ONLINE SUPPLEMENT
to article in
AMERICAN SOCIOLOGICAL REVIEW, 2006, VOL. 71 (JUNE:353–375)
McPherson, Smith-Lovin, & Brashears
“Social Isolation in America: Changes in Core Discussion Networks over Two Decades”
4
Table A3. Number of Respondents That Had Various Relationships with at Least One Discussion Partner (e.g. What
Percent of the Sample Mentioned A Spouse/Parent/etc. As A Person with Whom They Discussed
Important Matters?)
Type of Relationship to Respondentc1985 (N=1531) 1987a (N=1800) 2004b (N=1467)
Spouse 30.9% 43.21% 38.1%
Parent 23.6% 26.81% 21.1%
Sibling 21.1% 16.0% 14.11%
Child 17.9% 16.92% 10.19%
Other Family Member 18.2% 14.79% 11.8%
Co-worker 29.4% 28.44% 17.9%
Co-member of group 26.1% 21.48% 11.8%
Neighbor 18.5% 18.84% 7.8%
Friend 73.2% 77.98% 50.6%
Advisor 25.2% 45.03% 19.2%
Other 2.1% 4.50% 7.2%
a All differences between 1987 and 1985 are statistically significant at the p<.01 level except for Spouse, Parent,
Child, Friend, and Other, which are non-significant.
b All differences between 2004 and 1985 are statistically significant at the p<.01 level. Differences between 1987
and 2004 are statistically significant at the p<.01 level except for Siblings and Others, which are non-significant.
c Since more than one type of relationship can be mentioned for any given discussion partner (e.g., a co-worker can
also be a co-member of a group, an advisor and a friend), the percentages do not sum to 100.
Table A4. Respondents with Spouses Only, Non-Spouse Kin, and Non-Kin in the Network
1985 (N = 1531) 1987 (N=1800) 2004a (N = 1467)
Spouse ONLY Present 6.4% 5.5% 9.3%
At Least One Non-Spouse Kin 58.79% 60.19% 42.87%
At Least One Non-Kin 80.15% 83.79% 57.16%
a Significantly different from both 1987 and 1985 at the p < .01 level.
ONLINE SUPPLEMENT
to article in
AMERICAN SOCIOLOGICAL REVIEW, 2006, VOL. 71 (JUNE:353–375)
McPherson, Smith-Lovin, & Brashears
“Social Isolation in America: Changes in Core Discussion Networks over Two Decades”
5
Table A5. Differences by Age, Education, Sex and Race in Network Size and Kin/Nonkin Compositiona (Unstandardized OLS Regression Coefficients of
Network Variables on Respondents’ Demographic Characteristics)
Dependent Variables
Network Size # of Kin # of Non-Kin Proportion Kin
Independent Variables 85 87 04 85 87 04 85 87 04 85 87 04
A. Age
Age .02NS .02 NS .02 NS -.02 NS .02 .01 NS .03 -.01 NS .00 NS -.01 .01 .00 NS
Age2-.00 -.00 -.00 NS .00 NS -.00 -.00 NS -.00 .00 NS -.00 NS .00 -.00 NS -.00 NS
Constant 3.17 2.47 1.68 2.06 .97 .85 NS 1.17 1.51 .88 NS .69 .40 .48 NS
R .27 .14 .03 NS .08 .09 .04 NS .26 .08 .01 NS .18 .07 NS .05 NS
B. Educa tion
Sex
Educ (yrs) .19 .10 .14 .02 .02 .05 .15 .08 .08 -.03 -.01 -.02
Constant .61 1.33 .08 1.15 1.17 .45 -.47 .20 NS -.28 .87 .75 .79
R .35 .27 .23 .06 .05 .11 .35 .21 .20 NS .23 .12 .10
C.
Sex (f=1) -.04 NS .11 NS .20 NS .28 .17 .23 -.30 -.07 NS -.02 NS .07 .04 .01 NS
Constant 2.94 2.56 1.95 1.28 1.09 1.00 1.59 1.33 .89 .49 .50 .59
R .01 NS .04 NS .05 NS .10 .09 .03 .10 .03 NS .01 NS .10 .06 NS .01 NS
D. Race/ethnic (White is
reference category)
Black -.79 -.51 -.51 -.58 -.34 -.53 -.19 NS -.15 NS -.12 NS -.08 NS -.05 NS -.08 NS
Other -.43 NS -.06 NS -.64 -.42 -.03 NS -.49 .00 NS -.00 NS -.11 NS -.08 NS -.02 NS -.11
Constant 3.03 2.68 2.22 1.51 1.41 1.23 1.44 1.24 .91 .54 .58 .61
R .14 .14 .15 .14 .11 .17 .04 NS .04 NS .04 NS .07 NS .05 NS .10 NS
Note: All coefficients significant, p < .01, except where indicated.
a Marsden (1987) also analyzed differences in network size and kin composition by size of place, but this variable has not yet been coded for 2004 so comparable
analyses are not possible at this time. (The size of place variable is added to the data set after the data are collected, using the respondents’ addresses and current
Census tract information.)
ONLINE SUPPLEMENT
to article in
AMERICAN SOCIOLOGICAL REVIEW, 2006, VOL. 71 (JUNE:353–375)
McPherson, Smith-Lovin, & Brashears
“Social Isolation in America: Changes in Core Discussion Networks over Two Decades”
6
Table A6. Cooperativeness as Rated by the GSS Interviewer in 1985 and 2004
1985 (N = 1531) 1987 (N=1800) 2004 (N = 1467)
Level of Cooperativeness
Interested/Friendly 79.27% 77.05% 82.18%
Cooperative 16.74% 18.78% 15.13%
Restless/Impatient 3.58% 3.65% 2.65%
Hostile .41% .52% .04%
Note: X2 with 6 df = 32.07, p < 0.00
Table A7. Negative Binomial Regression of Size of Discussion Network
Model
Independent Variable I II III
Constant 1.069 1.129 .554
Wave2 (1 =1987) -.108 -.084 -.10
Wave3 (1=2004) -.349 -.336 -.404
Cooperative – -.187 -.121
Restless/Impatient – -.501 -.431
Hostile (Compared to Friendly/Interested) – -.785 -.618
Number Missing in Preceding Questions – -.197 -.162
Education (in yrs) – – .047
Female – – .069
Agea– – -.002
Currently Married – – .014NS
Black – – -.210
Other Race (Compared to White) – – -.177
Number of Adults in Household – – .018NS
Number of Children Ever Born – – .008NS
Number of Siblings – – .001NS
Alpha (Heterogeneity Coef.) 1.01*10-7 7.50*10-8 8.95*10-9
F 66.84 50.77 47.21
Note: All coefficients significant at p <.01, unless otherwise indicated.
a The squared term for age was not significant.
ONLINE SUPPLEMENT
to article in
AMERICAN SOCIOLOGICAL REVIEW, 2006, VOL. 71 (JUNE:353–375)
McPherson, Smith-Lovin, & Brashears
“Social Isolation in America: Changes in Core Discussion Networks over Two Decades”
7
Table A8. Logistic Regression Coefficients for Having Someone to Talk to About Important Matters (Dependent
Variable: No One to Talk to= 0, Someone to Talk to = 1)
Model
Independent Variable I II III IV
Constant 2.200 2.433 2.159 1.266
Wave2 (1 =1987) .851 .986 .897 1.273
Wave3 (1=2004) -1.080 -1.066 -1.392 -.247NS
Cooperative – -.364 -.159NS -.160NS
Restless/Impatient – -1.429 -1.293 -1.269
Hostile (Compared to Friendly/Interested) – -2.258 -1.968 -1.968
Number Missing in Previous Module – -.549 -.473 -.469
Education (in yrs) – – .089 .152
Education*Wave2 – – – -.038NS
Education*Wave3 – – – -.100
Femaleb– – .254 .253
Agea– – -.018 -.016
Currently Married – – .315 .250NS
Black – – -.903 -.863
Other Race – – -.284NS -.267NS
Number of Adults in Household – – – .089NS
Number of Children Ever Born – – – .018NS
Number of Siblings – – – -.013NS
F 118.56 50.78 37.28 27.12
Note: All coefficients significant at p < .01 unless otherwise indicated.
a The squared term for age was not significant.
b Significant at the p<.05 level.