We first investigate any difference in average NEP scores between
pre- and post-spill responses using a t-test. According to the variable of interest, changes in sociodemographic characteristics
between pre- and post-spill responses are also assessed using
either weighted t-tests or weighted Rao Scott chi-square tests. To
account for the possibility that changes in these characteristics may
confound observable changes in NEP scores between time periods,
we then specify a linear least-squares regression to regress respondents’
NEP scores on variables believed to influence environmental
attitudes.