Binary logistic regression
is carried out to compare the overall visits to Estonia with visits
made to the two smaller study areas in Estonia.
Regression analysis
is undertaken to predict the probability of visiting Saare or Tartu
counties against that of visiting the whole of Estonia. Logistic
regression is used because it is effective for describing relationships
between a categorical dependent variable (a binary one in the
present analysis) and one or more independent variables (Li, Fang,
Huang, & Goh, 2015; Peng, Lee, & Ingersoll, 2002). First, we use
models separately for each variable: country of origin, visitation
time, trip duration, and number of counties visited. Thenwe use full
models for the two areas including all the aforementioned variables.
The probability of visiting other counties in Estonia is also
assessed. Hereinafter the names Saaremaa and Tartumaa are used
interchangeably with Saare County and Tartu County, respectively.