Quantitative approaches in life cycle assessment—part 1—descriptive statistics and factor analysis

Aneurin Grant,Robert Ries,Carla Thompson
DOI: https://doi.org/10.1007/s11367-016-1099-4
2016-04-05
The International Journal of Life Cycle Assessment
Abstract:PurposeThis study examines the utility of descriptive statistics and factor analysis in the interpretation of an LCA data set. It is believed that the use of descriptive statistics and factor analysis provides a comprehensive summary of the data, and insight on internal validity, which improves interpretation of results. The current study also provides a pretext for subsequent analyses with inferential statistics.MethodsNine building envelope combinations were analyzed according to five service life models (N = 45, 45 data points). Three environmental indicators were used to characterize the data: global warming potential (GWP), atmospheric eco-toxicity (ECO), and atmospheric acidification (ACID) from the Tool for the Reduction and Assessment of Chemical and Other Environmental Impacts (TRACI) assessment method. Descriptive statistics measured minimum, maximum, mean, standard deviation, skewness, and kurtosis. For the factor analysis, data were disaggregated according to 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). All statistical operations were performed using SPSS software.Results and discussionDescriptive statistics show a relatively normal data distribution for GWP. Data distributions for ECO and ACID deviated slightly from normal, which was confirmed through Shapiro-Wilk and Kolmogorov-Smirnov tests. Factor analysis resulted in three factors for GWP and ECO, and two factors for ACID. In each analysis, cumulative life cycle impact, major replacement, and major replacement (frequency) aligned as variables in a common factor. This alignment corroborates previous findings. Other variables in the factor analysis were less clearly or inconsistently aligned.ConclusionsThe descriptive statistics provide important information about the distribution of data in the study. The data are determined to be suitable for further analysis using inferential univariate and multivariate statistical tests, a topic that is examined in part 2 of this study as reported by Grant et al. (Int J Life Cycle Assess, 2015). The factor analysis used in the study corroborates the conclusions of previous studies and provides a basis for future statistical analyses.
What problem does this paper attempt to address?