Evaluating the end results of the analysis the data

evaluating the end results of the analysis. The data was then used to generate fixed effects
multiple regression models, which control for differences in time and between subjects, and
simple regressions for data unable to fit within the fixed effects models. The dependent variables
in these models include all measures of smoking consumption and cessation rates, and the
independent variables are measures of cigarette excise tax rates and prices. While the models
attempt to control for some confounding variables, they do not directly control for income,
specific population demographics, education, and other characteristics that have been
hypothesized to affect cigarette consumption, due to the unavailability of current state-level
longitudinal data.
4. DATA ANALYSIS
The descriptive statistics for the study's most recent years dataset reflect the great
diversity in values across states (see Table 1 in Appendix for more detail). Additionally, the
preliminary ranking of states' cigarette tax rates, smoking rates, and tobacco prevention spending
hints to a mild link between the variables, as 10 out of the top 15 high-tax states in the Union
also appear to be in the top 15 states with the lowest adult smoking rates. Also, 9 of the states in
top 15 lowest tax rates appear within the 15 states with the highest smoking rates. Such a strong

11
pattern does not exist between the top 15 tobacco-prevention-spending states and low smoking
rate states, and vice versa (see Tables 2-5 for further comparisons). Isolating the MA-only time-
series data, preliminary graphic analysis illustrates an inverse trend between the increase of the
cigarette tax as a percentage in retail price and adult cigarette consumption. It also suggests a
direct trend between cessation rates and prices (see Figures 5-6).
A correlation matrix for the most recent year state-specific data begins the relationship
analysis between all the variables and gives a general suggestion as to the direction of the study.
The matrix displays a significant negative correlation (at the p < 0.05 level) between all measures
of cigarette consumption and cigarette excise taxes and prices, which is stronger for youth than
adults. Adult cigarette cessation rates also correlate positively to cigarette taxes and prices, and
they correlate negatively to tobacco-related death rates. However, cessation rates have no
significant correlation to gross cigarette tax revenues. There appears to be a positive correlation
between cigarette consumption and tobacco-related death rates and a negative correlation
between tax rates and tobacco-related death rates, as expected, but no such correlation between
cigarette consumption and tobacco healthcare costs. Surprisingly, no significant correlations
appear between gross cigarette tax revenue, tax rates and cigarette consumption either. Moreover,
funding for state tobacco control programs does not correlate to neither tax rates, cessation rates,
nor consumption rates; yet gross cigarette tax revenues correlate positively to funding for state