05.11.2014 Views

national multiple family submetering and allocation billing program ...

national multiple family submetering and allocation billing program ...

national multiple family submetering and allocation billing program ...

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

confidence interval ranged from 49.9 to 53.3 kgal. This means that in 95 of 100 samples are<br />

drawn this way, the actual annual water use per unit would be found within a range similarly<br />

constructed.<br />

Bivariate analyses. In order to examine the association of property characteristics with<br />

water use, two types of bivariate analyses were performed. Using ANOVA (analysis of<br />

variance), the relationship between water use <strong>and</strong> “categorical” variables such as the designation<br />

of a property as a senior citizen/retirement community (“yes,” “no” or “don’t know”) was<br />

examined. This test examines whether differences in the levels of the variable (water use in this<br />

example) are different in the specified subgroups. Factors with p-values less than 0.05 were<br />

considered “statistically significant,” meaning that if there were no difference, the probability of<br />

seeing a result as or more extreme than that seen in the sample was less than 5%.<br />

The relationship of continuous variables, such as average bedroom size or average rent,<br />

with water use were examined using Pearson correlations. A correlation statistic is produced<br />

which ranges from -1 to +1. Correlations of 0 indicate there is not a relationship between the<br />

two variables, while those close to -1 or +1 indicate strong negative or positive relationships. A<br />

p-value is shown, <strong>and</strong> in general, the 0.05 level is chosen to indicate statistical significance.<br />

In addition, differences in characteristics among properties based on the water <strong>billing</strong><br />

method used (a categorical variable, designated as “in-rent or HOA,” “submetered,” “RUBS,”<br />

“hot water hybrid,” or “other”) were examined using ANOVA for continuous variables <strong>and</strong> chisquare<br />

for categorical variables. The chi-square test examines whether proportions are different<br />

between two groups; for example, whether a similar or different proportion of properties are<br />

considered retirement communities among the different <strong>billing</strong> types. A p-value is produced,<br />

<strong>and</strong> factors with p-values of less than 0.05 were considered “statistically significant,” meaning<br />

that if there were no difference, the probability of seeing a result as or more extreme than that<br />

seen in the sample was less than 5%.<br />

T-tests were used to test differences between water use estimates by <strong>billing</strong> method or by<br />

survey source (e.g., comparing water use estimates of postcard survey respondents to manager<br />

survey respondents). P-values are also produced for these tests.<br />

Multivariate analyses. There were differences in the characteristics of in-rent properties<br />

compared to the impact properties; for example, submetered properties were more likely to be<br />

newer (41% were built after 1994 compared to 7% of in-rent properties), while RUBS properties<br />

were more likely to be larger complexes (71% had 100 or more units, compared to 32% of in-<br />

57

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!