Table 6.-Estimated equations and means for eating and drinking es-
 tablishments group, Dade and Monroe Counties, 1975-76.
Equation
Number Estimated Equationa
4.1 W= -15.2308 13.973r+11.155 1n DH+8.055 In BH
 (26.435) (3.080)' (16.393) (3.051)'
 R2=0.405 R2-0.250 SER=32.945 n=24
 Standard
 Variable Identification Mean Deviation


 W=quantity water purchased, thousands
 of gallons per month
 r=price of water
 DH=dining area (tens of square feet)
 times hours open per week
 BH=bar area (tens of square feet) times
 hours open per week
 aFigures in parenthesis are the standard errors.
the regression.
 'Significant at 0.01 level.
 eSignificant at 0.05 level.


53.38 39.82


 0.66
949.21


 0.46
983.30


 656.79 1,036.80

SER is standard error of


crease in price and purchases increased at a decreasing rate with
respect to increases in the size of dining and drinking areas
(which were multiplied times hours open). The relationship be-
tween two of the variables is shown in Figure 4. Demand declines
for increases in water price and shifts for increases in the dining
area times hours open interaction term. That is, at any given
price, the water demand is higher if the dining area of the res-
taurant is larger and the restaurant stays open more hours.
This was as expected. Of particular interest is that for any given
price, water use increased at a decreasing rate as the dining
area-hours open interaction term increased. This suggests there
are certain economies that result in larger restaurants with re-
spect to water use. Also, certain economies in water use can be
realized by keeping any given sized establishment open more
hours. The same type of relationships hold for the barroom area
times hours open interaction term.
 As noted above, the water price coefficient has the proper
sign but is not significant at even the 0.20 level. Proceeding with
this in mind, the water price responsiveness is inelastic over the
entire price range of the data given the mean quantity of 53.4
thousand gallons consumed per month. The finding that the in-
elastic region dominated the range of data was expected, but
further research needs to be completed to either support or deny
this conclusion.