Supplementary Material S4. Conditional Inference Trees. Conditional Inference (CI) classification trees for predicting perceptions of climate change and its ...
Supplementary Material S4 Conditional Inference Trees Conditional Inference (CI) classification trees for predicting perceptions of climate change and its impacts on forest ecosystems across the Canadian forest sector. Conditional Inference (CI) classification trees split the dataset into different groups based on certain values of the explanatory variables. At each intermediate mode, dataset is split into two groups based on the values of the explanatory variable indicated in the branches under the node. Stacked bar plot at each terminal node indicates the proportion of respondents that disagreed (dark gray), agreed (light gray) or were unsure (gray) about the statement. Total sample size depends on the statement, and can be calculated adding the sample size of the terminal nodes (indicated by brackets). Each tree only shows statistically significant variables at p < 0.05. The classification accuracy for each conditional inference tree is also provided. Education level can have 4 levels: Non Univ: non universitary studies; BSc: bachelor of sciences, MSc: Master of Sciences, PhD: doctorate. Provinces are BC : British Columbia; AB: ALberta; ON : Ontario; QC : Québec; NB: New Brunswick. Stakeholders are: F.Gov: Federal Govewrnment; P.Gov: Provincial Government; Indus: Industry; Priv.: other private organizations; Acad: academia (professors, researchers); Stud: graduate students.
1
1.2 Climate change impacts are exaggerated (Classif.accuracy = 73.2%)
Politics p < 0.001
≤3
>3
n = 639
n = 333
1
1
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
0
0 Disagree
Unsure
Agree
Disagree
Unsure
Agree
1.5 I do not understand the impacts of climate change (Classif.accuracy = 76.7%) Education p < 0.001 ≤ BSc
> BSc
Stakeholder p < 0.001 {P.Gov, Indus} {F.Gov, Priv., Acad, Stud} Education p = 0.004 ≤ Non Univ. n = 225
> Non Univ.
n=8
n = 109
n = 624
1
1
1
1
0.8
0.8
0.8
0.8
0.6
0.6
0.6
0.6
0.4
0.4
0.4
0.4
0.2
0.2
0.2
0.2
0
0
0
0
Disagree
Agree
Disagree
Agree
Disagree
Agree
Disagree
Agree
1.6 There is ample time to adapt to climate change (Classif.accuracy = 71.1%)
Politics p < 0.001
≤3
>3
Gender p < 0.001
Male
Gender p = 0.004
Female
n = 412
Male n = 225
Female
n = 266
n = 64
1
1
1
1
0.8
0.8
0.8
0.8
0.6
0.6
0.6
0.6
0.4
0.4
0.4
0.4
0.2
0.2
0.2
0.2
0
0
0
0
Disagree
Agree
Disagree
Agree
Disagree
Agree
Disagree
Agree
2.2 Within the next 50 years CC is going to have a significant impact on forest ecosystems (Classif.accuracy = 87.5%) Politics p < 0.001 ≤3
>3
Education p = 0.005
Education p < 0.001 ≤ BSc Politics p = 0.004
≤ Non Univ. > Non Univ.
≤5 n = 22
> BSc
n = 615
>5
n = 136
n = 20
n = 178
1
1
1
1
1
0.8
0.8
0.8
0.8
0.8
0.6
0.6
0.6
0.6
0.6
0.4
0.4
0.4
0.4
0.4
0.2
0.2
0.2
0.2
0.2
0
0
0
0
0
DisagreeAgree
DisagreeAgree
DisagreeAgree
DisagreeAgree
DisagreeAgree
2.3 Within the next 100 years CC is going to have a significant impact on forest ecosystems (Classif.accuracy = 94.6%)
Stakeholder p < 0.001
{Acad, Stud}
{F.Gov, P.Gov, Indus, Priv.}
n = 355
n = 606
1
1
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
0
0 Disagree
Unsure
Agree
Disagree
Unsure
Agree
2.4 CC effects on forest ecosystems are predictable (Classif.accuracy = 80.5%) n = 970 1
0.8
0.6
0.4
0.2
0 Disagree
Unsure
Agree
2.5 There is certainty about the effects of CC on forest ecosystems (Classif.accuracy = 65%)
Province p < 0.001
QC
{BC, AB, ON, NB}
n = 331
n = 635
1
1
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
0
0 Disagree
Unsure
Agree
Disagree
Unsure
Agree
2.6 The effects of CC on forest ecosystems are understood by forest managers (Classif.accuracy = 66.7%) n = 970 1
0.8
0.6
0.4
0.2
0 Disagree
Unsure
Agree
2.7 Forest managers have the ability to control CC impacts on forest ecosystems (Classif.accuracy = 69.3%)
Age p < 0.001
≤ 35−44
> 35−44
Province p = 0.019
{BC, QC, NB} n = 510
{AB, ON}
n = 274
n = 187
1
1
1
0.8
0.8
0.8
0.6
0.6
0.6
0.4
0.4
0.4
0.2
0.2
0.2
0
0
0
Disagree Unsure
Agree
Disagree Unsure
Agree
Disagree Unsure
Agree
3.1 Current forest legislation takes into account the impacts of CC on forest ecosystems (Classif.accuracy = 67.6%)
Province p < 0.001
{BC, AB, ON, QC}
NB
Stakeholder p = 0.018
{F.Gov, Acad, Stud} {P.Gov, Indus, Priv.} n = 461
n = 430
n = 78
1
1
1
0.8
0.8
0.8
0.6
0.6
0.6
0.4
0.4
0.4
0.2
0.2
0.2
0
0
0
Disagree Unsure
Agree
Disagree Unsure
Agree
Disagree Unsure
Agree
3.2 CC is properly incorporated into calculations of timber supply (Classif.accuracy = 67.8%)
Age p < 0.001
≤ 3
Education p = 0.003
≤ BSc
> BSc
n = 190
n = 444
n = 333
1
1
1
0.8
0.8
0.8
0.6
0.6
0.6
0.4
0.4
0.4
0.2
0.2
0.2
0
0
0
Disagree Unsure
Agree
Disagree Unsure
Agree
Disagree Unsure
Agree
3.4 We need to create and design new forest practices to deal with the impacts of CC on forests (Classif.accuracy = 74.9%)
Stakeholder p < 0.001
{F.Gov, Acad, Stud}
{P.Gov, Indus, Priv.}
Stakeholder p < 0.001
{P.Gov, Priv.} n = 492
Indus
n = 397
n = 79
1
1
1
0.8
0.8
0.8
0.6
0.6
0.6
0.4
0.4
0.4
0.2
0.2
0.2
0
0
0
Disagree Unsure
Agree
Disagree Unsure
Agree
Disagree Unsure
Agree
3.5 We should wait to see the impacts of CC on forests before implementing adaptive practices (Classif.accuracy = 77.7%)
Politics p < 0.001
≤5
>5
Politics p < 0.001
≤2
>2
n = 340
n = 584
n = 38
1
1
1
0.8
0.8
0.8
0.6
0.6
0.6
0.4
0.4
0.4
0.2
0.2
0.2
0
0
0
Disagree Unsure
Agree
Disagree Unsure
Agree
Disagree Unsure
Agree
