Interviewer effects in the Chacobo database
How does interviewer explain where interviewees fall in ordination space?
Plant-space
## r pvals
## gender 0.001890013 0.592
## ethnicity 0.050040369 0.009
## interviewer 0.367763296 0.001
## age 0.013932917 0.159
Use-space (SUB.CATEGORIA)
## r pvals
## gender 0.002450793 0.489
## ethnicity 0.050320810 0.011
## interviewer 0.413031531 0.001
## age 0.018707152 0.059
Plant-use space (Nombre.cientifico+SUB.CATEGORIA)
## r pvals
## gender 0.0005452161 0.851
## ethnicity 0.0480889886 0.025
## interviewer 0.4052127014 0.001
## age 0.0412818140 0.002
So we’ve shown significant differences among interviewers in which plants and uses they people they interview report. Does number of plants / number of uses elicited differ significantly among interviewers?
And do interviewers who themselves reported more plants or uses tend to elicit more plants and uses? No: there’s no trend of this.
## Source: local data frame [6 x 8]
##
## interview age gender ethnicity interviewer
## <chr> <dbl> <fctr> <fctr> <fctr>
## 1 Abigail Morán Toledo MOV NaN Mujer Chácobo MOV
## 2 Adelia Jimenez Martínes ESR 82 Mujer Colona ESR
## 3 Agustin Rodriguez Arauz JSM 78 Hombre Chácobo JSM
## 4 Alberto Chávez Yaco GOS 31 Hombre Chácobo GOS
## 5 Alberto Rutani Cepa ESR 48 Hombre Chácobo ESR
## 6 Alberto Yacu Roca GCM 18 Hombre Chácobo GCM
## Variables not shown: numberSpecies <int>, numberUses <int>, Informant.name
## <chr>.
However, there are many reasons why interviewers might interview distinct sets of the populations. What about those who were interviewed twice? How did their respones change based on who was interviewing them?
Plants
Uses
Plant-uses
So do informants who elicit answers similar to their own still find plants, uses, and plantuses NOT reported in their own interview?
Three who tend to elicit answers similar to their own still also elicit many new plants, uses, and especially plant-use combinations, and the likelihood of any given mention they elicit being novel is relatively high.
BCM
## new_plants np_mentions same_plants sp_mentions new_uses nu_mentions
## 1 84 1193 57 2866 27 4059
## same_uses su_mentions new_puses npu_mentions same_puses spu_mentions
## 1 22 3895 216 2030 58 2029
## novel_sp novel_use novel_plantuse
## 1 0.2939148 0.5103093 0.5001232
MOV
## new_plants np_mentions same_plants sp_mentions new_uses nu_mentions
## 1 102 710 46 1719 34 2429
## same_uses su_mentions new_puses npu_mentions same_puses spu_mentions
## 1 16 2181 282 1372 52 1057
## novel_sp novel_use novel_plantuse
## 1 0.2923014 0.526898 0.5648415
SCO
## new_plants np_mentions same_plants sp_mentions new_uses nu_mentions
## 1 106 1206 48 1697 29 2903
## same_uses su_mentions new_puses npu_mentions same_puses spu_mentions
## 1 20 2779 233 1541 52 1362
## novel_sp novel_use novel_plantuse
## 1 0.4154323 0.5109117 0.5308302
Compare with GCM and MSM - it’s fairly similar
GCM
## new_plants np_mentions same_plants sp_mentions new_uses nu_mentions
## 1 93 686 49 1420 33 2106
## same_uses su_mentions new_puses npu_mentions same_puses spu_mentions
## 1 25 2026 252 1095 60 1011
## novel_sp novel_use novel_plantuse
## 1 0.325736 0.5096805 0.519943
MSM
## new_plants np_mentions same_plants sp_mentions new_uses nu_mentions
## 1 104 722 49 1061 29 1783
## same_uses su_mentions new_puses npu_mentions same_puses spu_mentions
## 1 23 1570 266 1225 48 558
## novel_sp novel_use novel_plantuse
## 1 0.4049355 0.5317626 0.6870443