4.2. Meta-analysis results for GHGE studies
Table 7 presents the frequency distributions for the structural and outcome variables of the GHGE studies. Of the 195 observations, 67.7% exhibited positive outcomes and 32.3% exhibited neutral or negative outcomes. That is, in terms of GHGE, organic farming was favored over conventional farming. In all, 43.1% of the GHGE comparisons involved field crops and 72.8% of the comparisons involved farm sizes of more than 10 ha. The frequency distribution of the structural variables in the GHGE studies was similar to that of EE studies except for the variables of measurement unit and measurement method. EE studies more often employed area-based outcome measures with EAM as the measurement method, whereas GHGE studies more often used output-based outcome measures with LCA as the measurement method.
A logistic regression analysis was employed to identify the structural variables that were associated with GHGE benefits for organic farming. The analysis results (Table 8 and Table 9) identified the structural variables that were significantly associated with better GHGE effects for organic farming. The logistic regression model provided a good fit. The Chi-square value was statistically significant (p = .000) and the result of the Hosmer–Lemeshow test was not significant (p = .341).
The goodness of fit for the logistic regression model was confirmed by the accuracy of classifications based on the model, which was 72.72% (Table 9). The model did very well in predicting cases in which GHGE organic farming outcomes were superior, predicting the outcome variable with 86.90% accuracy. However, the structural variables in the logistic regression were poor predictors of neutral and negative GHGE outcomes, exhibiting an accuracy of only 42.20%. Thus, further research is needed to identify the determinants of neutral or negative GHGE outcomes for organic farming versus conventional farming. The classification accuracy of the GHGE model was similar to that of the EE model.
The structural variables of product, cropping pattern and measurement unit were statistically significant. Livestock exhibited a negative statistically significant value (p = .004), which indicated that better GHGE effects for organic farming were less likely for livestock than for field crops. Although the values for the vegetable and fruit categories were negative, they were not statistically significant. GHGE effects for organic farming might be less likely for vegetables (p = .126) than for field crops. The values for other product categories, which included dairy (p = .534) and mixed crops (p = .426), were positive but not statistically significant, which suggests that the GHGE effects for organic farming for these categories were similar to the outcomes for field crops.
The logistic regression results indicated that superior GHGE effects for organic farming were highly dependent on the measurement unit. Output-based (ratio/ton) outcome measures significantly reduced the superiority of GHGE effects for organic farming (p = .000) in comparison to area-based (ratio/ha) measures. These results are consistent with Lynch et al. (2011), which found output-based measures do not favor organic farming, particularly with respect to GHGE, due to the yield differences between conventional and organic farming. The significant value for cropping pattern was negative (p = .028), indicating that the GHGE superiority of organic farming was higher for monocropping than for multicropping.
Other structural variables in the analysis exhibited the expected values but were not statistically significant. Study period (p = .820), location (p = .860 and .534), duration (p = .890), data source (p = .679 and .703), measurement method (p = .407 and .886) and farm size (p = .816) were not associated with better GHGE effects for organic farming. Superior GHGE effects for organic farming were modestly related to sample size but were not statistically significant (p = .313 and .131). These findings suggest that study period, location, duration, data source, farm size, and measurement method were not strongly associated with superior GHGE effects for organic farming. However, superior GHGE effects for organic farming might be associated with larger sample sizes.
The results of the analysis indicated that superior GHGE effects for organic farming were more marked for studies that involved monocropping rather than multicropping and for studies that used area-based rather than output-based effect measures. The meta-analysis results confirmed the results of earlier meta-analyses that found superior environmental effects for organic farming per unit of land (Bengtsson et al., 2005, Mondelaers et al., 2009 and Tuomisto et al., 2012) rather than per unit of output. Superior GHGE effects for organic farming were less marked for livestock in comparison to other product categories. As Table 10 indicates, better GHGE effects for organic farming were found for 91.7% of the farm-level studies that involved monoculture cropping patterns, area-based measures, and the product categories of field crops, dairy, and mixed crops.
4.3. Comparison of meta-analyses for EE and GHGE
Table 11 presents comparison results of the meta-analyses for EE and GHGE. The structural variables of data source, sample size and product type significantly affected the EE of organic farming in comparison to conventional farming, whereas product type, cropping pattern, and measurement unit significantly affected the GHGE of organic farming in comparison to conventional farming. The better EE effects for organic farming were primarily associated with the field study's data source and sample size of more than 100, whereas better GHGE effects for organic farming were primarily associated with monoculture in cropping patterns and area based measurement unit. The results support previous studies that investigated EE with superior performances for organic farming when the data were obtained from field surveys and experiments rather than from secondary data. However, there were no differences based on data sources for studies that investigated GHGE.
Increases in sample size were significantly associated with superior EE effects for organic farming, whereas increases in sample size were only modestly associated with superior GHGE effects for organic farming. The superiority of organic farming was significantly reduced for output-based measurement of GHGE and weakly reduced for area-based measurement of EE. These findings support the Lynch et al. (2011) claim that output-based measures typically do not find benefits of organic farming, particularly for GHGE, due to yield differences between conventional farming and organic farming. Earlier studies found superior performances for organic farming with monocropping compared to multicropping patterns. However, this finding was only significant for GHGE.
For EE, the analysis indicated that superior performances for organic farming were associated with field crops, livestock, and mixed crop farms compared to vegetable and fruit farms. For GHGE, better performances for organic farming were associated with field crops, dairy, and mixed crop farms, whereas poorer performances were associated with livestock, vegetable and fruit farms. None of the other structural variables influenced differences in EE or GHGE between organic and conventional farming. Study publication date, location, measurement method, farm size, and duration did not significantly influence environmental effects for organic farming, which indicates that the influence of these variables on the differences between organic and conventional farming were negligible.
5. Conclusions
In this paper, logistic regressions were estimated to identify the structural variables that were associated with superior environmental effects for organic farming compared to conventional farming. Data source, sample size, and farm products were significant for EE performance and farm products, cropping patterns, and measurement unit were significant for GHGE outcomes of organic farming.
In the EE studies, the superiority of organic farming was more likely to be found in studies with larger samples, field studies, and experiments rather than secondary data. In the GHGE studies, the superiority of organic farming was more likely to be found for studies with monocropping compared to multicropping and with outcome measures based on area rather than output.
The results suggest that future studies should employ enough samples to improve confidence on the performance of organic farming compared to conventional farming. Future studies, especially on GHGE, should be cautious in identifying the appropriate measurement unit because output-based measures often do not favor organic farming (particularly for GHGE) due to yield differences between conventional and organic farming. When land use efficiency and energy productivity are considered, output-based (per weight) measures are more appropriate for assessing EE and GHGE than area-based (per ha) measures. The cropping pattern has a significant impact on GHGE. Therefore, we recommend that future studies investigate monocropping for direct and unbiased comparisons of the environmental effects of organic and conventional farming.
EE studies were more likely to find that organic farming was superior for field crops and dairy farms and less likely for vegetable and fruit farms. GHGE studies were more likely to find that organic farming was superior for field crops, dairy, and mixed crop farms and less likely for livestock, vegetable, and fruit farms. These findings indicate that the comparisons of the environmental effects of organic and conventional farming should take the type of farm product (e.g., field crops, livestock, fruits, or vegetables) into account and that comparisons should be based on the same types of farm products.
The variable of duration is not s