Table 4.
Multivariate model results for the change in forest cover
| Variables | Model 1a: OLS | Model 1b: Spatial Error | Model 1c: Random Effects |
|---|---|---|---|
| INTERCEPT | −2.45 (3.57) | −2.24 (3.67) | −2.82 (3.81) |
| ROAD ACCESS: | |||
| None | 5.60 (2.51)* c | 5.84 (2.76)* | 5.68 (2.85) |
| In 1999 only | 1.76 (3.58) | 3.20 (3.81) | 3.09 (3.84) |
| In 1990 and 1999 | Ref | Ref | Ref |
| NEAREST CITY | |||
| Coca | 0.10 (2.89) | 0.08 (3.20) | 0.13 (3.32) |
| La Joya de los Sachas | −1.99 (3.28)* | −1.15 (3.64) | −2.04 (3.78) |
| Shushufindi | −5.12 (3.35) | −45.17 (3.76) | −5.14 (3.87) |
| Lago Agrio | Ref | Ref | Ref |
| DISTANCE (KM) a | |||
| Coca | −0.02 (0.16) | −0.02 (0.18) | −0.01 (0.19) |
| La Joya de los Sachas | 1.09 (0.62) | 0.85 (0.69)* | 0.99 (0.71) |
| Shushufindi | −0.34 (0.51) | −0.40 (0.57) | −0.30 (0.60) |
| Lago Agrio | −0.18 (0.18) | −0.16 (0.21) | −0.16 (0.21) |
| FINCA SETTLEMENT COHORT | |||
| ≤ 1975 | Ref | Ref | Ref |
| 1975–79 | −7.16 (3.35)* | −7.10 (3.26)* | −6.43 (3.39) |
| 1980–84 | −8.19 (3.40)* | −8.47 (3.28)** | −7.95 (3.45)* |
| 1985–1990 | −12.94 (3.73)** | −12.46 (3.63)** | −12.30 (3.79)** |
| 1990 FINCA POPULATION b | |||
| Males 12–49 | −1.65 (0.74)* | −1.36 (0.69)* | −1.57 (0.73)* |
| All 50 years and up | 2.24 (1.14)* | 1.88 (1.08) | 2.14 (1.14) |
| CHANGE IN POPULATION | |||
| Population 0–11 Years | −0.28 (0.33) | −0.29 (0.32) | −0.30 (0.33) |
| Females 12–49 Years | −0.44 (0.69) | −0.57 (0.67) | −0.48 (0.69) |
| Males 12–49 | −1.39 (0.60)* | −1.15 (0.56)* | −1.39 (0.59)* |
| OFF-FARM EMPLOYMENT, 1990 | 0.16 (0.15) | 0.09 (0.14) | 0.16 (0.15) |
| CHANGE IN OFE | −0.03 (0.04) | −0.05 (0.04) | −0.03 (0.04) |
| TENURE SECURITY | −0.39 (2.27) | −0.22 (2.17) | 0.38 (2.34) |
| SOIL QUALITY | −3.87 (2.60) | −4.39 (2.61) | −3.79 (2.69) |
| Spatial Autocorrelation | Morans I=0.11 (p=0.009) | ||
| LMlag = 3.32 (p=0.068) | λ = 0.18 (0.05)*** | ICC = 7.0 d | |
| LMerr = 4.24 (p=0.039) | |||
| AIC | 3184.3 | 3178.3 | 3115.7 |
Standard errors are in parentheses. ‘”Ref” refers to the Reference Category.
p<0.001;
p<0.01;
p<0.05;
Distances are centered around the mean distances to each city, i.e., Coca=Distance-24.3, Joya=Distance-11.4, Shushufindi=Distance-13.0, and Lago Agrio=Distance-19.1;
Population counts are centered on their grand mean, i.e., Males 12–49 = Males12_49 – 2.12; Pop 50up in 1990 = Pop90_50up – 0.71;
No road access in either 1990 or 1999 results in 5.6% more forest cover from 1990 to 1999 compared to road access in 1990 and 1999.
ICC, or the intra-cluster correlation coefficient, is computed as residual sector variance (27.01) divided by total variance (359.92+27.01).