Main content area

Quantitative approaches in life cycle assessment—part 2—multivariate correlation and regression analysis

Grant, Aneurin, Ries, Robert, Thompson, Carla
The international journal of life cycle assessment 2016 v.21 no.6 pp. 912-919
acidification, computer software, data collection, durability, ecotoxicology, environmental assessment, environmental indicators, global warming, life cycle assessment, models, regression analysis, transportation, value added
PURPOSE: This study examines the use of inferential statistics, specifically multivariate correlation and regression, as a means of interpreting LCA data. It is believed that these methods provide additional context in understanding data and results, and may serve as a way to present the uncertain results that are inherent to LCA. METHODS: Nine building envelope combinations were analyzed according to five service life models (N = 45). Three environmental indicators were used: global warming potential, atmospheric ecotoxicity, and atmospheric acidification from the Tool for the Reduction and Assessment of Chemical and Other Environmental Impacts assessment method. Multivariate correlation was performed using nine variables, including cumulative life cycle impact, major replacement, major replacement (frequency), minor replacement, major repairs, minor repairs, inspections 1 and 2, and total transportation (N = 45, 405 data points). The same data set was used for the regression analysis, although the variables were limited to major replacement, minor replacement, major repair, and minor repair (N = 45, 225 data points). SPSS software was used for all statistical calculations. RESULTS AND DISCUSSION: Multivariate correlation analysis showed strong, statistically significant correlations between cumulative life cycle impact and major replacement across all environmental indicators. Similarly, the regression analysis showed strong R ² values between cumulative life cycle impact and major replacement, such that the influence of all other variables was considerably diminished. CONCLUSIONS: The use of inferential statistics provides useful information with respect to the strength and statistical significance of correlations between variables as in multivariate correlation, and allows for predictive capacity of impact, as demonstrated through regression analysis. Further studies should be conducted to confirm the added value of these analytical tools.