Positional error and time-activity patterns in near-highway proximity studies: an exposure misclassification analysis

Table 2 Distance bin misclassification by geocoding methodology

Distance group (m)	# of Residences by proximity to highway			Measures of agreement
	Tiger geocode	Orthophoto corrected	Confirmed match	Sensitivity	Specificity
0–50	30	48	25	52.08%	99.17%
51–150	208	182	169	92.86%	91.61%
151–250	130	145	121	83.45%	98.21%
251–450	72	83	65	78.31%	98.76%
451– 999	30	10	8	80.00%	96.55%
>=1000	177	179	175	97.77%	99.57%
	Streetmap USA	Orthophoto corrected	Confirmed match	Sensitivity	Specificity
0–50	26	48	24	50.00%	99.67%
51–150	208	182	171	93.96%	92.04%
151–250	112	145	98	67.59%	97.21%
251–450	97	83	61	73.49%	93.62%
451–999	27	10	7	70.00%	96.86%
>=1000	177	179	176	98.32%	99.79%
	Parcel geocode	Orthophoto corrected	Confirmed match	Sensitivity	Specificity
0–50	41	48	40	83.33%	99.83%
51–150	192	182	180	98.90%	97.42%
151–250	141	145	134	92.41%	98.61%
251–450	86	83	80	96.39%	98.94%
451–999	10	10	8	80.00%	99.69%
>=1000	177	179	176	98.32%	99.79%

The analysis omits those addresses not successfully geocoded to TIGER, StreetMap USA and Parcel datasets (n = 647).
Confirmed match represents the number of residences classified in the distance group by each geocoding method and orthophoto corrected location assignment.
Sensitivity is the percentage of confirmed positive residences for each distance bin (confirmed match divided by orthophoto corrected).
Specificity is the percentage of confirmed negative residential locations for each distance bin (confirmed negative divided by the orthophoto corrected).

ISSN: 1476-069X