ARCHES 2 (prepublication version)

ARCHES 2

Pre-publication version

0) Summary

We have previously shown a core model capturing how ARCHES participants perceive the dementia risk system. With an expanded number of models, we are now able to test the hypothesis that key demographic qualities of group model building participants affect the ways they perceive and build these models. We show that neighbor hood risk (measured by mean ADI of group) significantly affected the dementia risk system models produced by groups, with important implications to understanding and intervening in this system.

1) Demographics of groups

ARCHES round 2 had 36 groups. (ADI is National. ADI_level is ADI cut into bottom 25%, middle 50%, and top 25% groupings levels). I’ve left round and mdl.file.name for now, but presume that would be cleaned off for a manuscript.

Group n_people mean_age percent_female mean_education_years mean_ADI_N round model facilitator ADI_level
1 8 58.4 0.9 16.6 56.0 1 ./Arches-CLD-JFT-MS-17-03-23-Final.mdl JF low
2 11 59.6 1.0 15.7 59.0 1 ./Arches-GMB-Vensim-18-3-23-Final.mdl JF low
3 6 68.5 0.9 15.4 84.8 1 ./Arches-CLD.3.19.2023_MM-AW-final.mdl MM high
4 6 67.8 1.0 15.8 74.2 1 ./CLD-19-03-23-G2_JF_RAM_6_Final.mdl JF mid
5 8 66.4 0.9 15.2 76.0 1 ./ARCHES GMB Vensim Number 01 03.24.2023_Final.mdl JF mid
6 7 69.0 0.9 15.2 62.2 1 ./ARCHES GMB Vensim Number 02 03.24.2023_MS&AW-Final.mdl MS low
7 7 68.6 0.9 16.0 89.7 1 ./ARCHES GMB Vensim Number 01-AW-WZ-03.25.2023_Final.mdl AW high
8 6 60.2 1.0 14.2 74.5 1 ./ARCHES-GMB-JFT-NR-03-25-23-Final.mdl JF mid
9 5 62.4 0.8 13.2 77.0 2 ./Arches-GMB-Standardized-JFT-Ram-05-03-24-Final.mdl JF mid
10 7 69.7 1.0 14.7 62.7 2 ./ARCHES-GMB-AW-CC-05-03-24-Final.mdl AW low
11 5 63.4 0.9 14.8 73.4 2 ./ARCHES-GMB-Standardised-AW-MM-05-0424-final.mdl AW mid
12 4 57.8 0.5 18.0 49.8 2 ./Standardized ARCHES-05-05-24-Meena,JF,Chen-final.mdl JF low
13 6 64.0 1.0 15.5 73.3 2 ./ARCHES-GMB-5.5.24-AW-Ram-standardized-final.mdl AW mid
14 8 68.5 0.9 14.9 72.0 2 ./May 10 JF_chen.mdl JF mid
15 8 61.4 1.0 13.9 69.6 2 ./ARCHES051024.mdl AW low
16 7 59.6 0.9 14.6 72.7 2 ./GMB_Vesim Diagram_05112024_Shuya (1).mdl JF mid
17 7 65.3 1.0 17.3 79.4 2 ./ARCHES_workshop_5.11.24_AWgroup.mdl AW mid
18 6 66.3 1.0 14.0 83.2 2 ./Arches-GMB-YZ-JF-5-16-24-Final.mdl JF mid
19 5 64.4 1.0 16.8 86.6 2 ./Arches-GMB-05-16-24-AW&YZ-Final.mdl AW high
20 8 62.4 0.5 13.2 76.4 2 ./Arches-GMB-Standardized-MS-MM-05.16.24-final.mdl MM mid
21 5 60.0 1.0 14.2 74.0 2 ./Arches-GMB-5-17-24JF-Final.mdl JF mid
22 8 72.8 0.9 14.8 81.4 2 ./Arches-GMB-Standardized-AW-T-05-17-24-Final-v2.mdl AW mid
23 6 74.8 1.0 17.2 71.3 2 ./Arches-GMB-Mario-05-1724-Final.mdl MM mid
24 6 69.8 1.0 16.3 87.2 2 ./Arches-GMB-5-18-24-JFT-MS-Final.mdl JF high
25 7 68.4 1.0 15.6 91.9 2 ./ARCHES_GMB-5.18.24-AW&TB-Final.mdl AW high
26 6 67.3 1.0 15.8 80.8 2 ./Arches-GMB-05-18Mario-Yiou-Final.mdl MM mid
27 3 66.7 0.5 14.7 63.3 3 ./ARCHES CLD 10_19_24_JFT_Mia_Updated_Final_11.13.24.mdl JF low
28 8 58.0 1.0 14.0 64.5 3 ./ARCHES CLD 10_19_24_AW_Chen_updated_V4_Final_11.14.24.mdl AW low
29 4 56.5 0.5 12.5 88.8 3 ./ARCHE CLD 10_19_24_Mario_JW_updated_Final_11.14.24.mdl MM high
30 6 62.8 1.0 14.7 69.5 3 ./ARCHES CLD 10_20_2024_JFT_Mia_Final_11.14.24.mdl JF low
31 5 58.6 1.0 14.8 82.2 3 ./ARCHES CLD 10_20_2024_AW_Chen_updated_final_11.14.24.mdl AW mid
32 5 56.4 0.5 11.6 92.6 3 ./ARCHES CLD 10_20_24_MM_JW_V5_updated_Final_11.15..24.mdl MM high
33 8 66.8 1.0 16.1 84.3 3 ./ARCHES CLD 10_25_2024_JFT_Mia_Updated_Final_11.15.2024.mdl JF high
34 8 72.6 1.0 14.4 84.2 3 ./ARCHES CLD 10_25_24_Alexis_Vishnu_Updated_Final_11_15.24.mdl AW high
35 8 71.5 0.8 14.6 75.6 3 ./ARCHE CLD 10_25_24_Mario_Inema-JFT_Updated_final_11.15.24.mdl MM mid
36 6 68.3 1.0 14.7 77.0 3 ./ARCHES CLD 10_25_24_RS_TB_Updated_Final_11.15.2024.mdl RS mid

2) Models

We now have 36 models from these 36 groups. Here’s a summary of each and interactive examples of three of them. As before, light blue indicates “increase” and light pink “decrease”. Darker colors are Action Ideas.

Here’s electronic supplementary material of the interactive network for each model in

Round 1 (models 1-8)

Round 2 (models 9-26)

Round 3 (models 27-36)

Group number_edges number_unique_nodes
1 73 45
2 116 81
3 94 73
4 108 67
5 122 65
6 116 68
7 98 56
8 137 74
9 118 66
10 67 41
11 68 43
12 115 57
13 141 51
14 108 63
15 100 46
16 72 46
17 98 32
18 106 61
19 91 52
20 101 50
21 126 76
22 125 59
23 59 45
24 145 76
25 197 60
26 72 35
27 69 44
28 148 57
29 96 56
30 141 74
31 138 47
32 107 42
33 148 62
34 169 53
35 101 39
36 136 63

3) Core model

As in the previous publication, we built a model from the union set (i.e. including every edge from every group), and then focused it to a core model by A) considering only edges mentioned by at least two models, B) considering only loops/cycles (so no ‘externals’ or ‘dead end’ nodes) and C) only considering those that have some connection to Dementia Risk. This gives us finally 89 nodes/vertices, connected by 417 edges/vectors.

Here’s an interactive version of this, with heavier line weight indicating connections reported by many models.

Question - How do you want to display this for static submission? I think we discussed maybe a zoomed out version with a zoomed-in pane?

4) Indirect (i.e. intervening nodes)

When considering the similarity among the models different group propose (as in, e.g. calculating line width above), we have so far only considered as similar groups that directly/explicitly named a certain connection between two nodes (a direct edge), e.g. (Group 1: “A increases B” = Group 2 “A increases B). However, thinking of each edge separately does not fully characterize this connected system, which I would argue importantly includes the effects emergent from multiple connections. To approximate this, we may consider indirect edges from models that express the same causality, but with intervening nodes, e.g. (Group 1:”A increases B” = Group 2 “A increases C increases B”).

The table below shows the 89 edges (i.e. combination of “cause”, “increases / decreases” and “effect), total number of group models that that reported the edge, and which group models reported them as direct or indirect edges. You may have to scroll to the right to see all of the columns.

One example: Group 20 (as well as groups 26,28,32 and 35) directly specified that (increasing) health literacy will decrease dementia risk. This causal connection was also indirectly indicated by group 1 (and almost all other groups). The shortest path in Group 1’s path was Health literacy increases Exercise decreases Chronic disease` (which would increase) Dementia risk, although this is not the only path (e.g. Group 1 also identified paths leading from Health literacy to Dementia risk that pass through Healthy diet and paths that pass through Family caregiving.

cause polarity effect total models models (direct) models (indirect)
Health literacy decreases Dementia risk 30 20,26,28,32,35 1,2,3,4,5,6,8,9,10,11,12,13,15,17,18,19,21,22,24,25,29,30,33,34,36
Stress increases Dementia risk 27 4,6,13,14,15,22,33,34,36 3,5,7,8,9,12,16,17,19,20,21,24,25,26,28,30,32,35
Chronic disease increases Dementia risk 25 1,3,4,5,6,9,10,12,13,14,16,18,21,22,25,27,30,33,34,35 2,20,24,28,32
Healthy diet decreases Dementia risk 25 5,10,12,32,36 1,2,3,4,6,7,8,9,16,17,18,19,21,22,24,25,28,29,30,33
Social isolation increases Dementia risk 24 1,7,13,17,25,26,29,33,34 2,3,5,6,8,9,12,14,20,21,22,24,28,30,35
Access to quality healthcare decreases Dementia risk 24 3,4,14,20,23 1,6,8,9,12,13,16,17,19,21,22,24,27,29,30,32,33,35,36
Healthy diet decreases Chronic disease 23 1,4,6,7,8,9,10,18,21,22,25,28,30,31,33,36 3,5,12,16,24,32,35
Access to quality jobs increases Income 22 2,3,4,8,9,10,11,14,15,16,17,18,21,22,24,25,27,28,29,33,34,36
Stress decreases Mental health 22 3,4,5,6,7,8,12,16,17,20,25,26,29,30,32,35 9,19,21,24,34,36
Income increases Healthy diet 22 3,6,8,10,11,17,19,24,28,33,36 4,9,14,16,18,21,25,29,32,34,35
Dementia risk increases Stress 22 8,14,15,22,25,28,34,35 5,11,12,13,16,17,20,21,23,24,26,30,32,33
Exercise decreases Dementia risk 22 5,12,15,26,28,29,36 1,2,4,7,9,14,16,17,19,20,21,24,25,27,32
Income increases Access to quality healthcare 22 3,13,15,21,23,25,35 4,8,9,11,14,16,17,19,22,24,26,27,32,33,36
Income decreases Stress 22 11,17,19,28,29,34,36 6,8,9,13,14,15,16,21,23,24,25,26,32,33,35
Access to quality jobs decreases Stress 22 3,11,14,16,17,20 5,7,8,12,13,15,21,22,24,25,28,29,30,33,34,36
Health literacy decreases Stress 22 6,8,9,26 5,11,12,13,15,17,20,21,22,23,24,25,28,30,32,33,34,35
Access to quality jobs increases Access to quality healthcare 22 29,33,36 1,3,4,7,8,9,11,12,14,15,16,17,19,21,22,24,25,27,30
Quality health insurance increases Access to quality healthcare 21 1,3,7,8,9,13,14,15,16,17,19,21,22,24,26,27,29,33,36 25,30
Access to quality education increases Access to quality jobs 21 5,6,13,15,16,18,20,21,24,25,27,28,30,33,36 2,3,4,9,11,19
Dementia risk increases Social isolation 21 1,8,10,19,20,22,26,29,33 9,12,13,17,18,21,24,25,28,30,34,36
Substance use increases Dementia risk 21 5,15,17,28,32,36 2,3,6,8,9,12,13,14,18,20,21,22,24,25,34
Dementia risk decreases Healthy diet 21 10,11,34 1,4,8,9,12,14,16,17,18,21,24,25,28,30,32,33,35,36
Income increases Quality health insurance 20 3,6,13,14,15,16,17,18,19,23,24,25,26 8,9,10,22,27,33,36
Stress increases Substance use 20 5,7,8,12,13,14,20,21,25,28,30,32,34 3,9,15,17,22,26,29
Health literacy increases Healthy diet 20 1,4,5,10,11,18,29,30,32,33 8,9,12,17,21,24,25,28,34,36
Access to quality healthcare increases Health literacy 20 8,9,14,19,24,32,35 1,6,12,13,15,17,20,22,23,25,30,31,33
Access to quality healthcare increases Mental health 20 20,21,27,29 4,6,8,9,11,12,16,17,19,24,25,30,31,32,35,36
Access to quality jobs increases Quality health insurance 19 1,5,8,9,10,14,16,17,18,27,30,33,36 3,15,20,22,24,25
Mental health decreases Dementia risk 19 8,9,12,17,20,21,24,25,26,30,32,35,36 3,7,16,18,27,34
Health literacy increases Exercise 19 1,4,5,11,24,30 8,9,12,15,17,20,25,26,28,29,31,32,36
Access to quality healthcare decreases Stress 19 11,14,22,25 6,8,9,12,13,16,17,20,21,23,24,30,32,33,35
Dementia risk increases Chronic disease 19 10,12,25,35 1,4,5,8,9,13,14,18,21,22,28,32,33,34,36
Health literacy increases Access to quality healthcare 19 13,17,35,36 3,7,8,9,12,15,18,19,21,22,25,31,32,33,34
Food desert decreases Healthy diet 18 1,4,6,7,8,9,11,12,14,16,18,19,25,28,30,36 5,33
Stress increases Chronic disease 18 5,6,8,12,13,14,25,28,32,33 4,9,16,21,30,34,35,36
Dementia risk decreases Mental health 18 5,8,18,25,32,35,36 4,12,16,17,20,21,24,26,29,30,34
Exercise increases Mental health 18 7,9,19,20,25,27,32 4,5,12,16,17,21,24,26,29,31,36
Chronic disease increases Stress 18 6,7,12,30,33,34 5,13,14,16,20,21,22,24,25,28,32,35
Stress increases Social isolation 18 12,13,17,22 6,8,9,19,20,21,24,25,26,28,30,33,34,36
Dementia risk decreases Access to quality jobs 18 10,20,25 1,5,8,12,13,15,16,17,18,21,24,28,30,33,34
Social isolation increases Stress 18 8,25,34 6,9,12,13,14,17,20,21,22,24,26,28,30,33,35
Substance use increases Stress 18 22,25,34 5,6,8,9,12,13,14,15,17,20,21,24,28,32,36
Crime increases Stress 17 6,7,10,13,21,24,25,30,34,35 5,8,9,11,14,15,22
Substance use decreases Mental health 17 3,5,6,20,25,31,36 8,9,12,17,18,21,24,29,32,34
Mental health decreases Substance use 16 3,8,9,25,26,29,31 7,12,17,18,20,21,30,32,34
Social isolation decreases Mental health 16 5,17,20,25,30,34,35 3,6,8,9,12,21,24,26,29
Access to quality education decreases Crime 16 10,14,19,21,25,36 2,3,5,9,13,15,18,27,28,30
Exercise decreases Chronic disease 16 1,4,7,8,14,25 5,9,12,16,21,24,28,31,32,36
Stress decreases Sleep 16 4,9,15,17,30 3,8,14,21,22,24,25,28,32,34,35
Exercise decreases Stress 16 7,14,15,25 5,9,11,12,16,19,20,21,24,26,28,32
Healthy diet increases Mental health 16 7,24,30,31 4,5,6,8,12,16,18,21,25,29,32,36
Healthy diet decreases Stress 16 22,24,25,28 5,6,7,8,11,12,16,17,21,30,32,33
Crime increases Social isolation 16 7,18,24 6,8,9,10,13,19,21,22,25,28,30,34,36
Dementia risk decreases Income 16 22,34,35 2,8,10,15,16,17,18,21,24,25,28,32,33
Quality health insurance decreases Stress 16 17,20,22 5,6,8,9,13,14,16,21,24,25,26,30,33
Substance use increases Chronic disease 16 6,25,28 2,5,8,9,12,13,14,18,21,22,31,32,36
Community support increases Health literacy 15 1,3,9,12,13,14,19,32 2,6,20,26,28,30,31
Mental health decreases Stress 15 7,20,21,24,25,26,34,35 8,9,12,16,17,30,32
Dementia risk decreases Exercise 15 17,20,29 5,8,14,15,16,24,25,26,28,30,32,36
Substance use decreases Access to quality jobs 15 17,28,36 5,8,9,12,13,15,18,20,21,24,25,34
Dementia risk decreases Health literacy 14 20,23,26,35 13,14,15,17,22,25,28,30,32,33
Community support increases Mental health 14 4,16,21 3,6,9,12,19,20,24,26,30,31,32
Dementia risk decreases Sleep 14 10,28,35 8,14,15,17,21,22,24,25,30,32,34
Exercise increases Physical health 13 2,3,7,11,15,19,20,30,36 1,9,14,31
Income decreases Crime 13 6,8,19,25,34,35,36 9,15,18,21,27,28
Community support decreases Social isolation 13 12,14,18,19,20,24 6,9,13,21,26,28,30
Sleep decreases Stress 13 14,15,22,25,28 7,8,9,17,24,32,34,35
Access to quality jobs decreases Poverty 13 9,12,30,33 2,4,11,14,15,18,21,24,27
Access to quality jobs decreases Crime 13 2,30,36 5,8,15,18,19,21,25,27,28,34
Family quality relations decreases Substance use 13 25,31,36 3,13,14,15,17,18,21,29,30,34
Health literacy increases Self advocacy for health services 12 3,6,12,17,22,30,32,34,35,36 21,31
Substance use increases Crime 12 5,8,9,15,18,24,25,31,34,36 21,28
Brain activity decreases Dementia risk 12 3,5,11,12,13,15,28,29 9,17,24,36
Family caregiving decreases Access to quality jobs 12 1,4,5,8,9,21,25,34 12,16,24,28
Family caregiving increases Stress 12 5,8,12,14,24,25,28,34 6,16,21,35
Healthy diet increases Physical health 12 3,7,8,11,19,31 1,2,9,18,30,36
Depression increases Dementia risk 12 14,18,27,28,33 6,9,12,21,22,24,36
Family quality relations decreases Social isolation 12 10,18,24,29,34 1,13,17,21,25,30,36
Access to quality healthcare increases Physical health 12 11,13,20,31 3,8,9,14,15,19,30,36
Appropriate medication decreases Dementia risk 12 9,23,24,26 5,8,16,18,20,21,27,29
Depression increases Social isolation 12 6,10,19,22 9,12,18,21,24,28,33,36
Income decreases Homelessness 12 2,16,17,22 4,9,13,14,19,24,27,28
Dementia risk decreases Brain activity 12 23,26,36 5,12,13,15,17,21,24,28,29
Exercise increases Brain activity 12 17,31,36 5,9,12,15,21,24,26,28,29
Mental health increases Access to quality jobs 12 9,17,24 8,12,16,18,20,21,25,30,34
Mental health increases Exercise 12 31,32,36 8,9,16,17,20,24,25,26,30
Taxes increases Access to quality education 12 10,13,27 8,12,14,15,19,21,24,25,28
Income decreases Poverty 11 4,9,11,14,15,18,21,24,27,33 2
Public transportation increases Access to quality jobs 11 8,13,17,19,20,25,34,36 10,15,30
Public transportation increases Access to quality healthcare 11 3,4,7,8,13,15,19 17,25,34,36
Self advocacy for health services increases Access to quality healthcare 11 3,22,31,33,34,35 12,16,17,21,32
Family quality relations decreases Stress 11 13,17,25,34,36 14,15,21,23,24,30
Quality health insurance increases Regular healthcare visits 11 2,20,22,25 9,10,14,15,16,24,30
Self advocacy for health services increases Health literacy 11 12,30,34,35 6,17,22,28,31,32,33
Dementia risk decreases Family quality relations 11 10,21,29 4,14,15,18,24,25,30,34
Family quality relations increases Mental health 11 17,21,25 3,18,24,29,30,31,34,36
Mental health increases Sleep 11 3,30,35 8,9,17,21,24,25,32,34
Stress increases Depression 10 6,9,12,14,21,22,24,33 3,28
Dementia risk increases Family caregiving 10 2,8,12,21,24,25,34 1,5,16
Crime increases Food desert 10 4,5,8,9,18 6,14,25,28,36
Social isolation increases Depression 10 6,9,14,22,33 3,12,21,24,28
Chronic disease decreases Physical health 10 7,9,20,30 1,2,13,14,18,31
Dementia risk decreases Physical health 10 11,18,20,36 1,2,9,14,15,30
Mental health decreases Crime 10 8,34,35,36 9,18,21,25,30,31
Access to quality healthcare increases Regular healthcare visits 10 9,14,23 15,16,20,22,24,25,31
Community support increases Brain activity 10 23,24,26 3,9,12,13,21,28,31
Crime decreases Family quality relations 10 4,18,30 14,15,21,24,25,31,34
Depression increases Stress 10 9,21,24 6,12,14,22,28,33,36
Exercise increases Self care 10 19,21,26 2,9,12,15,17,25,28
Preventive health testing decreases Dementia risk 10 14,17,25 2,11,12,13,16,20,27
Stress decreases Physical health 10 7,11,30 8,9,14,15,19,20,36
Crime decreases Safe neighborhood 9 6,8,11,12,15,22,31,34,36
Regular healthcare visits decreases Chronic disease 9 2,10,16,25,30 9,13,14,31
Exercise increases Sleep 9 15,17,25,32 9,14,21,24,28
Sleep increases Mental health 9 7,8,34,35 9,17,24,25,32
Self care decreases Stress 9 15,19,26 9,12,17,21,25,28
Side effects of medication increases Dementia risk 9 2,13,36 15,17,21,22,24,34
Stress decreases Self care 9 6,12,25 9,15,17,21,26,28
Substance use increases Homelessness 9 2,13,22 9,14,17,19,24,28
Substance use decreases Self care 9 2,12,25 6,9,15,17,21,28
Crime decreases Black business 8 7,8,13,15,24,25,31 2
Access to quality healthcare increases Preventive health testing 8 12,13,15,16,27 14,17,25
Poverty increases Crime 8 9,13,15,24,30 18,21,27
Depression increases Substance use 8 18,21,28,36 9,12,14,22
Misdiagnosis increases Dementia risk 8 2,5,25,30 6,33,34,36
Self care increases Exercise 8 9,19,25,28 12,15,17,26
Social isolation decreases Community support 8 6,12,13,20 3,24,25,26
Appropriate medication decreases Substance use 8 8,21,30 9,18,20,26,29
Dementia risk increases Health literacy 8 1,2,21 8,9,12,24,34
Exercise decreases Depression 8 3,27,28 9,12,14,21,24
Health literacy increases Family caregiving 8 1,4,35 2,7,10,15,28
Health literacy increases Regular healthcare visits 8 10,23,24 9,20,22,25,31
Poverty increases Stress 8 12,13,21 9,15,24,30,33
Regular healthcare visits increases Physical health 8 19,20,31 2,9,13,14,30
Appropriate medication decreases Chronic disease 7 5,8,9,16,18,27 21
Education level increases Access to quality jobs 7 1,2,3,7,9,11 12
Income increases Private transportation 7 2,10,19,25,28,34 4
Black business increases Access to quality jobs 7 8,12,13,15,24 7,25
Depression decreases Mental health 7 6,9,12,18,24 21,36
Homelessness increases Stress 7 9,14,17,28 13,16,24
Income increases Quality neighborhood 7 11,15,23,25 18,28,36
Over medication increases Dementia risk 7 17,26,34,36 2,32,33
Physical health decreases Dementia risk 7 2,9,19,20 8,13,36
Preventive health testing decreases Chronic disease 7 2,16,25,27 12,13,14
Private transportation increases Access to quality healthcare 7 1,7,11,19 25,34,36
Self care increases Healthy diet 7 9,19,21,25 12,17,28
Chronic disease increases Mortality 7 27,31,33 18,24,28,30
Dementia risk increases Over medication 7 15,26,34 17,30,33,36
Over medication increases Substance use 7 2,17,30 15,26,32,34
Poverty increases Depression 7 18,24,33 9,12,21,27
Self care increases Mental health 7 17,19,25 9,12,21,26
Taxes increases Public resources 6 3,14,22,24,28 21
Access to quality healthcare increases Appropriate medication 6 8,23,24,27 6,9
Dementia risk increases Regular healthcare visits 6 9,15,30,33 2,24
Knowledge of family health history increases Health literacy 6 11,17,31,34 9,15
Misdiagnosis increases Stress 6 6,25,33,34 5,30
Private transportation increases Access to quality jobs 6 19,25,28,36 10,34
Public resources increases Access to quality education 6 2,9,14,28 19,24
Taxes increases Public transportation 6 8,10,13,19 15,28
Physical health decreases Stress 6 7,11,13 8,15,20
Public transportation increases Access to quality education 6 13,19,25 8,15,20
Quality neighborhood increases Access to quality education 6 11,13,15 18,25,28
Self care increases Sleep 6 15,17,25 9,21,28
Sleep decreases Depression 6 3,24,28 9,14,22
Social isolation decreases Self esteem 6 21,22,34 9,25,33
Crime decreases Quality neighborhood 5 13,15,18,25,36
Access to quality healthcare decreases Misdiagnosis 5 6,30,33,36 25
Family quality relations increases Family caregiving 5 10,15,23,34 1
Healthy diet decreases Obesity 5 3,24,33,36 32
Obesity increases Chronic disease 5 3,24,27,33 32
Over medication increases Side effects of medication 5 15,17,34,36 2
Family quality relations increases Knowledge of family health history 5 15,31,34 17,29
Mortality decreases Family quality relations 5 24,31,34 15,30
Mortality increases Social isolation 5 30,33,34 24,28
Poverty increases Homelessness 5 9,24,27 2,13
Public transportation increases Regular healthcare visits 5 2,15,20 10,25
Safe neighborhood decreases Stress 5 11,15,34 3,6
Self esteem decreases Stress 5 21,23,34 9,25
Access to quality education increases Education level 4 2,3,9,11
AI Awareness about lifestyle increases Healthy diet 4 7,8,12 5
Misdiagnosis increases Mortality 4 25,33,34 30
Misdiagnosis increases Over medication 4 33,34,36 30
Physical health increases Mental health 4 8,20,36 7
Black business increases Tax revenue 3 7,15,25
Health literacy increases AI Awareness about lifestyle 3 5,8,12
Quality neighborhood increases Taxes 3 11,15,25
Tax revenue increases Quality neighborhood 3 11,18,25

5) Fitting characteristics of groups to similarity of the models they propose

Using this method (indirect & direct edges) of quantifying the similarity of one model to another, we may use methods borrowed from ecology to 1) construct a matrix of (dis-)similarity of each model to each other model, 2) use non-metric multidimensional scaling (NMMDS) to represent this multi-dimensional set of distances in a two-dimensional space, and 3) fit various characteristics of model-building groups the distances of models from one another in this NMMDS ordination space. (Note: the fit algorithm compares the fit of the observed characteristic to 999 permuted data sets, and so pvalues may vary slightly in each run)

Here are the fits for group means of age, gender, education, and ADI National, (in addition to – as tests for the robustness of our method – group size n_people and facilitator).

In the table below Pr(>r) is the fit pvalue, with asterisks indicating significance thresholds. r2 is the amount of variation explained by the factor. You can ignore NMDS1 and NMDS2 for now – this gives the coordinates in the NMMDS ordination where the centroid of a factor (or the middle of a vector) would be.

In summary:

We see that group means of ADI National have significant (pval<0.05) and moderately strong (r-squared ~0.30) fit with where a group’s model falls in the ordination space.


***VECTORS

                        NMDS1    NMDS2     r2 Pr(>r)   
n_people              0.02529 -0.99968 0.0290  0.630   
mean_age             -0.31583  0.94882 0.0871  0.234   
percent_female        0.41287  0.91079 0.0813  0.237   
mean_education_years -0.99906  0.04345 0.0465  0.442   
mean_ADI_N            0.42724  0.90414 0.2864  0.004 **
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Permutation: free
Number of permutations: 999

***FACTORS:

Centroids:
                NMDS1   NMDS2
facilitatorAW -0.0161  0.1700
facilitatorJF  0.0596 -0.1346
facilitatorMM -0.0971  0.0452
facilitatorMS  0.0085 -0.2807
facilitatorRS -0.0292 -0.0571

Goodness of fit:
                r2 Pr(>r)
facilitator 0.0776  0.697
Permutation: free
Number of permutations: 999

If we break ADI National (a continuous variable) into quantile levels (lowest 25%, middle 50%, highest 25%), we can see they each occupy distinct areas of the ordination space, although there is significant overlap (as expected, given that ADI as a continuous variable only explains about 30% of the variation in models — that is, that there is still agreement on a core model across ADI levels).

Here, each point is one group’s model, and distances between each pair of points are defined by the dis-similarity in the edges (indirect and direct) in those models. So, points that are relatively close together are groups that had similar models, points that are widely spaced apart had more dis-similar models. I’ve colored and drawn outlines around groups that fell into each of the three ADI levels.

We may also characterize which components are relatively more common in ADI categories, (i.e. what the differences are that are driving the dissimilarity).

_____

Every edge that occurred in >75% of one or more ADI level

This shows every edge that had high frequency (was reported by 75% or more of the groups) in at least one of the ADI levels. Edges at the right were reported with high frequency across all levels. Edges with long great bars are ones with a large difference between the frequency with which they are reported by groups of the different ADI levels.

_____

Edges with the greatest difference (at least 40%)

This shows edges that were maximally different (at least 40% difference) among the high and low ADI levels. For instance, here, we can see that groups in the high ADI level were much more likely to report that Stress decreases Mental health, with groups in the midADI level in between. In contrast, more than half of the groups in the lowADI level reported that Substance abuse decreases Self care, which very few of the midADI and lowADI groups reported.