*2.6. Stage 6*

To test the hypotheses, positive and negative impacts of *Xn* variables on the reduction of incidence rates of *Yn* were revealed separately for the four types of circumpolar territories. Positive effects were differentiated as high positive (HP), positive (P), and moderately positive (MP); the negative ones—extremely negative (EN), negative (N), and moderately negative (MN). To decide on the degree of positive or negative effect, maximum and minimum extremes (*Xmax* and *Xmin*, respectively) were excluded from the calculation, and then a mean value *Xmed* was determined for each of the multitudes (Equation (3)):

$$X\_{mcd} = \frac{\sum X\_n - X\_{\text{max}} - X\_{\text{min}}}{n - 2} \tag{3}$$

A degree of the effect of *Xn* on *Yn* was recognized, when a value of *Xn* fell into one of the intervals (Table 4).


**Table 4.** *Xn* intervals and effects on *Yn*.

Note: *Xn* – regressors, *Yn* – regressands. Source: authors' development.

#### **3. Results**

The results are presented across five sub-sections in accordance with stages 2–6 of the study flow algorithm (Figure 2). We first checked the array of *Xn* variables established at stage 1 for collinearity (Section 3.1.), then selected the best subsets from derived multitudes (Section 3.2.). After that, we performed multiple regression analysis in selected subsets and generalized the effects of *Xn* on *Yn* for the entire Arctic Zone of Russia (Section 3.3.). Based on the identified correlations, we then categorized the territories into types (Section 3.4.), revealed positive and negative determinants of incidence rates across them, and tested out hypotheses (Section 3.5.).
