**3. Materials and Methods**

The research based on the convenience sample consisted of 362 students from the University of Novi Sad, Republic of Serbia, and was conducted in 2019 and 2020 (see Table 1).


**Table 1.** Sample characteristics.

For measuring physical activity, we relied on the item already used in similar research [11] (p. 183) "whether the individual had taken any exercise (e.g., sport, physically active pastime) in the past 2 weeks". Although the active traveling to work or study could also be considered in this context, the contribution of walking and bicycling for the purpose of transportation amongst students to the overall level of physical activity is very small [11] (p. 188), especially when dormitories are located near the students campuses, as is the case with the University of Novi Sad. Hence, in accordance to the criterion of ACSM (American College of Sports Medicine Position Stand), by which the recommended general level of physical activity refers to exercising 3 or more times a week, i.e., 6 times in 2 weeks [11], all respondents have been divided into three groups: inactive, low-frequency activity, and recommended frequency activity. Besides physical activity, the questionnaire included socio-demographic characteristics, among which three of them were continuous (age, household size, and living standard) and four categorical (gender, emotional status, accommodation, and year of study).

Bearing in mind that physical activity was presented through an ordinal variable, an ordered logit model was implemented for examining its relations with socio-demographic characteristics. In addition, for more detailed analysis, the concept of marginal effects [39] was applied as well. Furthermore, in order to distinguish specific segments, predictive probabilities have been calculated for different combinations of analyzed variables (those that had significant relations with physical activity level segments). Data processing was carried out by the use of STATA statistical package.

#### **4. Results**

Generally, when it comes to the level of physical activity, the results of this research show that 15.7% of students are inactive, 22.9% have low-frequency activity, while 61.3% met the recommended level of physical activity. In addition, the results of a logit model are presented in Table 2.


**Table 2.** Logit model.

The p value of a model equals 0.00 (Prob > chi2 = 0.00), which confirms its statistical significance at p < 0.01. As can be seen, at three (gender, accommodation, and living standard) out of seven independent variables, coefficients are statistically significant with p lower than 0.01.

In regard to the living standard, the obtained result (0.44) points to the existence of positive relation between this variable and students' physical activity. Therefore, it can be concluded that students with a better living standard are more physically active.

As two other independent variables with significant coefficients are categorical, we extended the analysis with the concept of marginal effects. The results related to gender are presented in Figure 2.

**Figure 2.** Predictive margins—gender.

In the case of inactive and low-frequency levels, the average probabilities are larger for female students (0.19 and 0.26) in comparison to male students (0.09 and 0.18), whereby their differences of 0.10 (for inactive) and 0.08 (for low-frequency level) are statistically significant with p < 0.01. On the other hand, in the case of recommended level of physical activity, the average probability for male students (0.73) is higher than the average probability for female students (0.54), with a statistically significant difference of 0.19.

When it comes to the accommodation (Figure 3), the highest average probabilities for inactive and low-frequency levels of physical activity were obtained for students who live in a dormitory (0.22 and 0.27), followed by those who live in a private accommodation (0.18 and 0.25), and those who live with their families (0.10 and 0.19). Contrarily, the average probability for the recommended level of physical activity was the highest for students who live with their families (0.71), followed by two other student categories, those who live in a private accommodation (0.57), and those who live in a dormitory (0.51).

**Figure 3.** Predictive margins—accommodation.

In addition, the differences in average probabilities of physical activity between these three groups have been examined for all three segments as well. They are shown in Table 3.


**Table 3.** Marginal effects—accommodation.

The differences in average probabilities for all three segments based on physical activity between students who live with their families and two other groups are statistically significant, with p < 0.01. On the other hand, the p values for differences in average probabilities between students who live in a dormitory and those who live in a private accommodation were higher than 0.05 in all three physical activity segments. The results considering the influence of independent variables from the gender aspect are shown in Table 4.


**Table 4.** Marginal effects—gender aspect.

As can be seen in Table 4, there are no larger differences in their values between male and female. Predictive probabilities for different combinations of significant independent variables can be seen in Table 5.


**Table 5.** Predictive margins—combinations.

When it comes to the inactive category, the highest average probability (0.57) refers to female students who live in a dormitory, with low living standard (1 out of 5). As for students with low frequent level of exercise, there are several profiles with an average probability of higher than 0.30 to belong to this segment: female, living with families, assessing standard with 1 or 2, female living privately with standard marked with 2 or 3, female living at dormitory, with living standard assessed with 3 and 4, male living in dormitory with assessed standard of 1 and 2, male living privately and evaluating standard with 1. The largest probability (0.88) to be active can be noticed for men, living with families and assessing standard with 5.

#### **5. Discussion and Conclusions**

When it comes to the level of Serbian students' physical activity, it should be noticed that 15.70% of students are inactive, 22.90% have low-frequency activity, while 61.30% meet the recommended level of physical activity. The results suggest that almost 40% of the students exercised less than six times in two weeks before the interviewing. Those relatively negative tendencies are partly in accordance to the previously described situation worldwide [7,10,11], as well as in Serbia [12].

Having previous results in mind, as well as the significance of physical activity for social sustainability, some general recommendations can be provided. Positive influence of celebrity endorsers are already proven in the literature especially when they are famous because of sport [16]. In the concrete case, trying to cooperate with Novak Djokovic could be a good idea. He is, at the moment of writing this paper, the world tennis player number one at ATP list, and is also Serbian and very popular within the country. When it comes to suggestions regarding student population, there can often be identified the stress on education: "changes to current college physical education programs" [6] (p. 124), "paying attention to the health education and the behaviors related to the health promotion" [17] (p. 205). However, if relying dominantly on the university in that process, the special caution should be paid at implementation of intervention measures [18]. Some other recommendations rely on providing low-cost programs of physical activity or bicycles [19].

However, in addition to general recommendations, within this research is conducted market segmentation. Not only that the research is focused on a specific segment of population, but it is additionally segmented regarding socio-demographic characteristics. The obtained results can be compared to some of the previous research. Hereby, out of seven independent variables—students' gender, age, household size, emotional status, accommodation, year of study and living standard, only three had significant influence, two of them at 0.10 level. The three variables that had significant influence are gender, accommodation, and living standard. Concretely, the average probability of physical activity is larger for male students in comparison to female students. That is in accordance with the results of some previous studies [19,30,32–37]. On the other hand, the existence of gender differences is not always confirmed in the literature [31]. Furthermore, the students with better living standard are also more physically active, what is in accordance to previous research [33]. Finally, the average probability of physical activity decreases starting from students who live with their families, followed by those who live in a private accommodation, to the students who live in a dormitory. The existence of the difference in the context of students' accommodation is in line with some of the authors [37], although their results are in favor of living on the campus. However, the comparability of the results is limited due to exclusion of students living with parents from their research. The highest activity of students living with parents in this research can be explained by the smallest change in life when started studying in comparison to other segments, and having the largest amount of free time because of relying on parents for performing certain activities in the house.

Contrarily, students' age, household size, emotional status, and year of study are not proven as significant predictors of physical activity in this research. The absence of the influence of age is in line with the studies [31] (when applying linear regression model) [34], but is not in accordance to other studies [30,33,36] (in the last case, for females). The lack of influence of the year of study is in accordance to [31] (when applying linear regression model), but not with [37]. Emotional status not influencing physical activity was proven in this research in line with [34], but in contrast to [33].

The description of the profiles of marketing segments led to several conclusions. Firstly, the greatest probability to be inactive can be noticed in the case of women living in dormitories and having very low living standard. However, weakly active students belong in most of the cases of women, living in different places, and assessing living standard with lower and higher marks. When men belong to this segment, they always have low standard and do not live with their families. Finally, most physically active are men living with families and have high living standard.

Starting from the descriptions of the segments, there is a need to target each of them. In the case of active students, the stress should be on maintaining their level of activity. When it comes to students being less active or inactive, they should be tried to be translated to more active segments in each of the cases.

From the context of marketing mix, several recommendations can be provided. As a starting point can be used, the means-end approach to consumer behavior, meaning consumers are not interested in products per se, but are interested in them regarding the way the product helps them attain their life values [40]. Hereby, physical activity can be presented as a mean for accomplishing different ends. Having in mind that it is the case of younger population, there is a great possibility (although additional studies should confirm this hypothesis) that being healthy is still not the primary focus of their interest, since it is usually immanent to their age. However, "being attractive" or "having a good time with friends" or "being interesting and adventurous" or "being strong and successful" can be of more interest to them and motivate them to spend their time in accomplishing those goals. Launching a campaign named, for example, "People inspired with me" (with ambiguous meaning—being inspired together or being inspired by) and asking the students to post Instagram photos when having physical activity with friends can fulfill if not all, but most of the listed goals. Providing equipment for being active can be of special importance—free bicycles, gyms under the open sky, balls for different sports, or even walking routes. As a part of the campaign, the participants can be given designed shirts or hats. Besides, the number of likes could be understood as non-monetary incentive. An application offering students information about places nearby where they can accomplish "being attractive," "having a good time with friends," "being interesting and adventurous," or "being strong and successful" by being physically active and having content available at those places could also be helpful. Engaging students from faculty of sport to show other students different options of physical activity might be useful. In addition to using social networks, direct contact with a target audience is possible. From the beginning of the studies, faculties could provide students information about accomplishing quality of life, including possibilities of performing physical activities. Such information could increase the popularity of the faculties and bring them more interested students, with whom communication can be performed by using, among others, e-mail marketing in accordance to permission marketing approach [41].

Some of the recommendations from researches abroad might be appropriate to be implemented in domestic conditions, as well, especially in the case of students' segments being less physically active. For example, bearing in mind that for people with a lower living standard, certain physical activity contents are less accessible, they can be attracted with incentives in the first period of intervention (free period, novel activities). Informing them about it can be done in a classical manner, since in the subsequent phases, there would be increasing influence of word-of-mouth, which is expected to be stronger if the number of participants recruited at the beginning was larger [23]. Community-based interventions had a positive effect not only in case of population with lower living standard [23,24], but for women as well [25]. Wearable technology is already proven to be successful for increasing physical activity of women [26], so it might be appropriate to inform them how to measure the level of their physical activity by using disposable technology. In this regard, activities of ambush marketing, to which attention is being paid in domestic conditions as well, can be used [42]. For example, under the billboard which promotes some possibility of physical activity, there could be added information about disposable free wearable technology.

It should be noticed that communication with all those segments can be performed directly, and by using online and offline marketing communication channels. The possibilities of identifying such segments on social networks are great. Besides leaving the information about the gender when creating profiles, the living standard can often be predicted considering the devices by which is being logged-in to profiles, while at the same time information about changing the place of living are also available, together with frequent check-ins at certain places, including dormitories. However, there should be taken into account conclusions from previous research that "solely Web-based intervention seems to be ineffective in promoting PA among universities students" and that "face-to-face lifestyle modification interventions have greater effects than Web-based interventions" [27] (p. 1608). Finally, considering a social marketing approach, it is of great importance to provide adequate monitoring regarding all previously described issues.

Future research may measure physical activity more precisely, include other determinants in addition to socio-demographic (especially lifestyle), reach larger and more representative sample, and monitor the participants through time. In addition, consultations with representatives of the country/city/university/sponsors could be performed in advance for obtaining their opinion about suggested intervention.

**Author Contributions:** All of the authors formulated goals of the research and interpreted available literature; conducting and analyzing research was performed by N.M. and N.D., while implications were developed by I.D. and A.G. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The authors declare no conflicts of interest.

### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
