Supplementary Material S4 Conditional Inference Trees - PLOS

0 downloads 0 Views 162KB Size Report
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