Measuring and Forecasting Demand - Tourism: Principles, Practices, Philosophies - page 293

TABLE 13.1 Demand and Income Data

Number of Tourist Per Arrivals
(thousands)

Capita Income of Tourists
(dollars)

Year

1 75 $6,300

2 90 7,200

3 100 7,000

4 105 7,400

5 95 6,800

6 110 7,500

7 105 7,500

8 100 7,200

9 110 7,600

10 120 7,900

Simple Regression: Linear Least Squares Method

In , information on demand levels for past years is plotted against one important

determinant of demand, say, income or prices. Then, through the application of a statistical technique
called

simple regression

the relationship between demand
and the particular variable being considered (such as income levels of tourists). Consider, for example,
the hypothetical data in Table 13.1 for demand levels for ten years and the income levels of tourists for
these same years.
By plotting the pairs of arrivals and income data on a graph, we obtain a relationship between
income and travel demand, illustrated in Figure 13.2. The points represent the annual observa-
tions, and the line AB represents the line of

least squares regression , a straight line is used to '' explain ''

It is obtained by the least squares method.
We can now obtain demand projections from this method based on what we expect income levels
to be in the future. Suppose we wish to estimate demand for year 15. In this year, income is
projected to be $8,300 per capita. As shown in the figure, the estimate of demand for this income
level is 128,000.
Because income is a major determinant of demand, simple regression ''explains'' demand to some
extent. It is superior to trend analysis for this reason. Besides, the methodology is still relatively simple
and can be presented visually. Data needed for this method are relatively easy to collect, when
compared to the data needs of the two following projection methods.

'' best t. ''

Multiple Regression: Linear Least Squares Method

The major drawback of simple regression is that only one variable can be considered at a time. In
reality, demand is affected by all the factors that in uence propensity and resistance, as discussed
earlier. It may not be feasible to include all these variables at one time, but it is certainly practical to
isolate a few that are particularly relevant to determining demand and deal with these in one model.
Multiple regression is one way to do this. It is essentially the same as simple regression, except that

now more than one variable can be used to explain demand. Through a mathematical formula, a
relationship is established between demand and the variables that we have chosen to consider in the
model. For example, suppose that we had data on the prices of tourist services at a destination in
addition to the incomes of the tourists. We could then regress demand on these two variables (income