TITLE: Neighborhood social characteristics and parental

advertisement
1
SUPPLEMENTARY INFORMATION
2
3
Title: Neighborhood built and social environment characteristics: a multilevel analysis of
4
associations with obesity among children and their parents
5
6
Summary:
7
Supplementary Table 1 describes inter-rater reliability for in-person neighborhood audits which
8
were conducted for a total of 4 330 street segments. When pairs of observers disagreed on their
9
assessment, three methods were used consecutively to select the consensus answer: in-person
10
reassessment, assessment using Google Street View, and random selection of consensus answer.
11
12
Supplementary Table 2 presents results from three separate principal components analysis, one
13
for each type of neighborhood data set used (i.e., census data, land use data, and observer data).
14
Rotated factor loadings and total variance in data explained by components obtained through
15
each principal component analysis are shown. Specifically, neighborhood poverty is defined by a
16
high proportion of residents with income below the low income cut-offs, single parent families,
17
unemployment, high mobility (more people who have lived ≤ 1 year at their current residence),
18
and a low proportion of residents who own their home. The low income cut-offs (LICOs) are
19
income thresholds below which a family will likely devote a larger share of its income on the
20
necessities of food, shelter and clothing than the average family. The approach is essentially to
21
estimate an income threshold at which families are expected to spend 20 percentage points more
22
than the average family on food, shelter and clothing.1 Neighborhood prestige is defined by a
23
high proportion of residents who are university educated and more expensive residences.
1
24
Neighborhood degree of urbanicity is defined by a high residential density, greater number of
25
parks and neighborhood area covered with parks, many street intersections and streets with
26
normal traffic at rush hour. Neighborhood traffic is defined by many streets (total length and
27
proportion) with high traffic at rush hour. Neighborhood physical disorder and deterioration is
28
defined by the presence of graffiti, litter, and deteriorated roadways and buildings. Lastly,
29
neighborhood pedestrian friendliness is defined by streets with lowered speed limits, the presence
30
of stop signs (all-direction at intersections and mid-street stop signs), zebra crossings, and
31
pedestrian crossing signs.
32
33
Reference:
34
1. Statistics Canada. Low income cut-offs; 2012 [cited 2013 Apr 12]. Available from:
35
http://www.statcan.gc.ca/pub/75f0002m/2012002/lico-sfr-eng.htm
36
2
37
38
39
40
41
42
43
44
45
46
47
48
49
50
Supplementary Table 1. Agreement between observers 1 and 2 (kappa coefficients) for
selected items of the in-person neighborhood assessment tool and methodsa used to obtain
consensus between observers 1 and 2 for a total of 4 330 street segments around the
residences of 512 families from the QUALITY study in 2005-2008.
Item
Kappa
Coefficient (95%
Confidence
interval)
Frequency
of Google
Street View
used to
obtain
consensus
(%)
3.6
Frequency
of random
selection of
consensus
answer
(%)
0.50 (0.45; 0.55)b
Frequency of
in-person
reassessments
used to
obtain
consensus
(%)
2.8
Graffiti visible on street
segment
Presence of enough litter to
fill up one average size
grocery bag
Condition of roadway on
street segment
Condition of buildings
visible from street segment
Presence of speed limit of ≤
30 km/hour on street
segment
Presence of all-direction
stop sign at intersection
Presence of mid-street
segment stop sign
Presence of a zebra crossing
at intersection
Presence of a pedestrian
crossing sign at intersection
0.33 (0.26; 0.40)
1.5
2.5
0
0.49 (0.47; 0.52) b
5.1
18.4
0.6
0.47 (0.44; 0.50) b
3.7
11.5
0.5
0.86 (0.83; 0.88)
0.6
2.6
0.07
0.88 (0.87; 0.89)
1.0
3.9
0.2
0.79 (0.67; 0.90)
0.09
0.3
0.02
0.83 (0.81; 0.86)
0.5
2.0
0.2
0.77 (0.73; 0.81)
0.6
2.0
0
0.05
a
When independent pairs of observers disagreed on their assessment, items were re-assessed by a
third independent observer on another occasion. If re-assessment was too unwieldy (e.g.
neighborhood far away from the research center, or only a few discordant items for a specific
neighborhood to justify revisiting it), consensus was obtained using Google Street View. If
Google Street View was not available for the area, the consensus answer was selected randomly
between the two discordant answers.
b
Weighted Kappa coefficient.
3
51
52
53
54
Supplementary Table 2. Rotated factor loadings and total variance in data explained by
components obtained through three separate principal component analyses (for census,
land use, and observer data) for the residences of 417 families from the QUALITY study in
2005-2008.
Neighborhood variables
Rotated
factor
loadings
2006
Component 1: Neighborhood poverty
Census
% residents with low income
0.84
data
% single parent families
0.79
% unemployment
0.77
% home ownership
-0.93
1 year mobility
0.65
Component 2: Neighborhood prestige
% residents with university degree
0.94
Average residential value
0.95
Total variance explained by components 1 and 2
72.6%
Land use
data
In-person
neighborho
od audit
data
Component 1: Neighborhood degree of urbanicity
Neighborhood residential density
Number of parks in neighborhood
Neighborhood area that is covered with parks
Number of 3 or more way intersections
Meters of streets with normal traffic at rush hour
Component 2: Neighborhood traffic
Total street length with high traffic at rush hour
% streets with high traffic at rush hour
Total variance explained by components 1 and 2
Component 1: Neighborhood physical disorder and
deterioration
% of streets where graffiti is visible
Presence of litter on at least one street
Presence of at least one roadway that is in bad condition
Presence of at least one street where over half of the buildings
are in bad condition
Component 2: Neighborhood pedestrian friendliness
% of streets where speed limit is ≤ 30 km/hour
% of streets with an all-direction stop sign at intersection
Presence of at least one street with a mid-street segment stop
sign
Presence of at least one street with a zebra crossing
% streets with a pedestrian crossing sign
Total variance explained by components 1 and 2
0.73
0.66
0.74
0.72
0.69
0.96
0.96
65.0%
0.72
0.71
0.64
0.71
0.60
0.46
0.54
0.61
0.73
42.3%
55
4
Download