Skip to main content

Table 5 Concordance rates between different clustering techniques and between indexes

From: A statistical procedure to create a neighborhood socioeconomic index for health inequalities analysis

 

HC (5)avs. quintilesb

HC (3)avs. tiertilesb

HC (3)avs. optimal thresholdsb

Optimal thresholds vs. tiertilesb

Carstairs

Townsend

HC (5)avs. quintilesb

Optimal thresholds vs. tiertilesb

HC (5)avs. quintilesb

Optimal thresholds vs. tiertilesb

Lille Métropole

41%

78%

98%

79%

38%

70%

42%

78%

Grand Lyon

48%

74%

93%

78%

47%

77%

40%

75%

Marseille urban unit

48%

69%

97%

67%

51%

67%

50%

69%

Global

63%

71%

97%

72%

57%

70%

55%

71%

  1. aHierarchical Clustering using Principal Components (in parenthesis, the number of categories chosen).
  2. bConcordance rate (percent of census block groups categorized into the same class using the two different clustering schemes).