Skip to main content
. Author manuscript; available in PMC: 2013 Apr 3.
Published in final edited form as: Prof Geogr. 2012 Apr 3;64(2):157–177. doi: 10.1080/00330124.2011.583592

Table 4.

Goodness of fit statistics for models associating individual BMI with various combinations of built environment measures by corrected quasi-likelihood under the independence criterion (QICC)

BG scale CT scale Buffer scale
Male Female Male Female Male Female
Level 0 Only 7 controls, without any 3D variables QICC 51456.5 52088.7

Level 1 Level 0 + population density and intersection density
QICC 51354 52010.2 51221.5 51964.7 51101.9 51983.2
Improvement from Level 0 −102.5 −78.5 −235.0 −124.0 −354.6 −105.5

Level 2 Level 1 + area-based diversity measures

Level 2a Frank's index QICC 51293.1 52002 51149.7 51962.6 51062.3 51930
Improvement from Level 0 −163.4 −86.7 −306.8 −126.1 −394.2 −158.7
Level 2b Shannon's index QICC 51254.8 52013.1 51156.1 51965 51048.9 51895.8
Improvement from Level 0 −201.7 −75.6 −300.4 −123.7 −407.6 −192.9
Level 2c Simpson's index QICC 51266 52006.9 51140.7 51952.7 51051.5 51879.3
Improvement from Level 0 −190.5 −81.8 −315.8 −136.0 −405.0 −209.4
Level 2d Frank's 6 categories QICC 51213.2 51894.4 51129.4 51697.6 50934.8 51608.16
Improvement from Level 0 −243.3 −194.3 −327.1 −391.1 −521.7 −480.5

Level 3 Level 1 + destination-based diversity measures

Level 3a Distance to light rail station QICC 51230.8 51412.4 51101.1 51338.6 51006.2 51377.6
Improvement from Level 0 −225.7 −676.3 −355.4 −750.1 −450.3 −711.1
Level 3b Distances to LR & CBD QICC 51062.2 51184.4 50982.3 51171.9 50882.2 51191.1
Improvement from Level 0 −394.3 −904.3 −474.2 −916.8 −574.3 −897.6
Level 3c Distance to large grocer QICC 51351.8 52008.2 51210.3 51952.9 51084.7 51977.9
Improvement from Level 0 −104.7 −80.5 −246.2 −135.8 −371.8 −110.8
Level 3d Distances to LR, CBD, & large grocer QICC 51048.3 51172.2 50959.4 51167.9 50856.9 51186.6
Improvement from Level 0 −408.2 −916.5 −497.1 −920.8 −599.6 −902.1

Level 4 Level 1 + proxy-based diversity measures

Level 4a Percentage walk to work QICC 51296.4 51787.6 51167.6 51524.4 50983.1 51503.8
Improvement from Level 0 −160.1 −301.1 −288.9 −564.3 −473.4 −584.9
Level 4b Median year structure built QICC 51020.0 51382.4 50947.9 51381.1 50844.3 51327.9
Improvement from Level 0 −436.5 −706.3 −508.6 −707.6 −612.2 −760.8

Level 5 Combination of Levels 3b and 4b Level 1 + distances to LR & CBD + median year structure built
QICC 50960.5 51007.2 50908.4 51030.1 50819 51028.1
Improvement from Level 0 −496.0 −1081.5 −548.1 −1058.6 −637.5 −1060.6
Improvement from Level 3b 101.7 177.2 73.9 141.8 63.2 163
Improvement from Level 4b −59.5 −375.2 −39.5 −351.0 −25.3 −299.8

Level 6 Level 5 + six land uses, % walk to work, or both Level 1 + distances to LR & CBD + median year structure built + more

Level 6a Level 5 + six land uses QICC 50859.2 50802.6 50877.5 50744.7 50737.5 50743.6
Improvement from Level 0 −597.3 −1286.1 −579.0 −1344.0 −719.0 −1345.1
Improvement from Level 5 −101.3 −204.6 −30.9 −285.4 −81.5 −284.5
Level 6b Level 5 + % walk to work QICC 50958.4 50992.9 50910.4 50980.2 50806.8 50960.3
Improvement from Level 0 −498.1 −1095.8 −546.1 −1108.5 −649.7 −1128.4
Improvement from Level 5 −2.1 −14.3 2.0 −49.9 −12.2 −67.8
Level 6c Level 5 + six land uses + % walk to work QICC 50855.9 50795.4 50879.3 50712.4 50738.6 50679.2
Improvement from Level 0 −600.6 −1293.3 −577.2 −1376.3 −717.9 −1409.5
Improvement from Level 5 −104.6 −211.8 −29.1 −317.7 −80.4 −348.9

Note: Values in italics show improvement of fit in comparison with other models, where larger negative values are associated with larger improvement. Shaded values indicate the best fit model for each level of models by gender.