5. Results
5.1. Results of the regression analysis
The model specifications were estimated using panel corrected
least squares (Williams, 2000). Each of the specifications
has 189 observations allocated to 92 clusters. Table 3
shows the results for the combinations of C1 to C3 with the
landscape specification L1. R2 ranges from 0.68 to 0.70; coastal
specification C3 has the highest R2. Table 4 shows the results
for the coastal combinations C1 and C2 with the landscape
specification L2.6 The landscape specification L2 has R2 values
of 0.66 and 0.65 for the specifications C1 and C2 respectively;
these are slightly lower than for the combination with the
landscape specification L1. Each of the specifications passes
the Ramsay RESET test for correct functional form (the pvalues
range from 0.09 to 0.38).
The results for the variables describing the type of accommodation
are highly significant and positively related to price.
As can be expected the coefficient of the dummy variable
HOTEL is much higher than that of BANDB; hotels are a much
greater leap in quality than bed and breakfast establishments
are from rooms in private accommodation. Hence, a greater
price difference can be expected. The coefficient of the variable
HOTEL is significant at the 0.5% level for all specifications. The
coefficient of the variable BANDB for the specification combinations
L1 is significant at the 0.5% level, and f