CONCLUSION

In our first exploratory regression analysis, the census variables chosen had a low adjusted R-squared value. After running an OLS and then GWR, we were able to visually assess the model’s strength and the relationship between each explanatory variable and the dependent variable. Our model’s low R2 limited our analysis to the city of Edmonton (Census subdivision), where our model had the highest local R2 value. Our GWR model explained 41.97% of the density of substance incidents however the relationships between the dependent variable and the explanatory variables did not follow those we found through literature on environmental justice.

The results of the study show that our model was unable to explain the density of substance release incidents with our census variables. It seems that in Alberta, the influence of socio-economic and racial factors such as education, visible minority, and immigrants are not as evident as we expected.

 

FURTHER ANALYSIS

We shifted our analysis away from an urban focus to the less populated western portion of census division no.11. In our first model we did not take into account the magnitude of the oils spills and the type of substances released. Both these are important because the type of spill and size may determine what is perceived to be an environmental justice issue.

For our second GWR, we focused on crude oil spill data and incorporated census variables related to labour force. Our exploratory regression analysis gave us a much higher adjusted R-squared value (0.09). Our model fit the best in the western portion of the Census division where the highest incidence of crude oil spills occurred. This time around our model explained 46.13% of our dependent variable and we found that unlike traditional studies where low income and education were most burdened by environmental impacts, those most burdened by crude oil spill incidents were those working in the mining industry outside of the urban area and those working in natural resources and agriculture. Despite the improvement seen in our model, the local R2 values of our GWR model were still quite low and more variables are needed to better explain the occurrence of crude oil spills.

 

LIMITATIONS

We chose substance release data from 2006 to 2015 however our census data was from 2011. This can be a possible source of error. In addition, when considering volume released, we were unable to get data from the National Energy Board. This meant we could not include data from spills pertaining to the biggest pipelines (crossing provincial and national borders) in our further analysis. Much of our literature findings was focused on Alberta and while we first attempted to run this study for the whole province, the scale was not appropriate for analysis with a GWR. Our choice of census division would definitely affect our results and conclusions. In addition, the limited data on Aboriginal populations could have been a major factor in the exclusion of it as an explanatory variable.

 

RECOMMENDATIONS

This study has shown that Environmental justice issues in Alberta should be investigated with a different focus rather than the typical traditional environmental justice perspectives. It would be useful if future studies were able to incorporate pollution down river streams in their investigation of environmental justice and Aboriginals in Alberta. It would also be interesting if future studies could relate environmental justice to not only the density of incidents but the health impacts of oil spills as well.