Advertisement

Journal of Population Ageing

, Volume 10, Issue 1, pp 73–86 | Cite as

A Two-Step Multivariate Composite I-Distance Indicator Approach for the Evaluation of Active Ageing Index

  • Ivana Djurovic
  • Veljko JeremicEmail author
  • Milica Bulajic
  • Marina Dobrota
Article
  • 285 Downloads

Abstract

Ageing population represents one of the most pressing issues of the modern society. Recognising the need for a proactive attitude, a multidimensional Active Ageing Index (AAI) has been introduced. In order to consider the AAI as a useful tool for future policy making, several weaknesses of the current methodology must be stressed out. Specifically, the assignment of weights to individual indicators is essentially a subjective procedure. As a possible remedy to the issue, this paper proposes the application of the statistical Composite I-distance Indicator (CIDI) method. The aim of this research is to introduce the use of a two-step CIDI approach that has several benefits. Firstly, the I-distance method, as an underlying foundation of CIDI, defines which of the input indicators/domains are most significant for the ranking process; therefore it allows the calculation of statistically based weights. Secondly, the obtained weights integrated into AAI framework provide more impartial rankings by each of the four domains and in total. Moreover, uncertainty and sensitivity analysis was applied to both the official and the proposed methodology, and the results imply that our proposal establishes a more stable ranking. Additionally, analysis has shown that input indicators/domains with the highest significance are areas in which implemented policies would contribute the most to Active Ageing goals.

Keywords

I-distance CIDI Active ageing index Multivariate analysis Composite index Weights 

Introduction

The ageing population, a trend that has started in the second half of the twentieth century, refers to growing participation of senior citizens in the total population. It is typically seen as a problem of developed countries, especially the richest ones. However, even though this trend can be easily observed in developed countries, the world population is distributed unevenly, making the developing countries a home of the majority of senior citizens in the world. Expectations concerning elder population are that by the end of 2050, eight out of ten of senior citizens will reside in less developed countries (UN 2013).

One of the reasons for this demographic transition is the longer life expectancy; thus the introduction of policies that promote stronger inclusion of older age groups in everyday activities is becoming an urgent matter. The shift in the structure of age groups is not as momentous as the expansion of population that came before. However, it is more than enough to reshape the world economy (Age Invaders 2014). The “old-age dependency ratio” or the ratio of old people to those of working age is rapidly increasing. Giving the fact that existing pension and health systems need to be reformed (Disney 2016; Moreau 2016), we are witnesses that appropriate policies addressing this issue are more than urgent. With the extent of demographic changes in Europe being much more drastic than in the USA (Börsch-Supan et al. 2014), European countries were the first to take a proactive stand and focus on a necessity to increase older people’s well-being (Nordbakke and Schwanen 2014). More concretely, they decided to impose policies needed for the creation of living and working conditions suitable for Active Ageing. As a part of this initiative, European Union declared 2012 the Year of Active Ageing and Solidarity between Generations (EU 2011; European Commission 2012b) and within that framework Active Ageing Index (AAI) was introduced. AAI is a composite index that consists of four domains: Employment, Participation in Society, Independent, Healthy and Secure Living, and Capacity and Enabling Environment for Active Ageing, and 22 subdomains/indicators in total (see Table 1).
Table 1

Weights for official (implicit and explicit) and CIDI methodology assigned to individual indicators and domains

Indicators / Domains

Implicit weights

Explicit weights

Two-step CIDI weights

1.1. Employment rate 55–59

55 %

25 %

19 %

1.2. Employment rate 60–64

28 %

25 %

28 %

1.3. Employment rate 65–69

11 %

25 %

30 %

1.4. Employment rate 70–74

6 %

25 %

23 %

1st domain: Employment

29 %

35 %

19 %

2.1. Voluntary activities

20 %

25 %

22 %

2.2. Care to children, grandchildren

45 %

25 %

35 %

2.3. Care to older adults

18 %

30 %

33 %

2.4. Political participation

17 %

20 %

10 %

2nd domain: Participation in Society

18 %

35 %

21 %

3.1. Physical exercise

2 %

10 %

16 %

3.2. Access to health and dental care

25 %

20 %

13 %

3.3. Independent living

24 %

20 %

18 %

3.4. Relative median income

12 %

10 %

1 %

3.5. No poverty risk

13 %

10 %

9 %

3.6. No material deprivation

13 %

10 %

15 %

3.7. Physical safety

10 %

10 %

13 %

3.8. Lifelong learning

1 %

10 %

15 %

3rd domain: Independent, healthy and secure living

21 %

10 %

27 %

4.1. Remaining life expectancy of 50 at 55

33 %

33 %

15 %

4.2. Share of healthy life expectancy at 55

22 %

23 %

20 %

4.3. Mental well-being

20 %

17 %

24 %

4.4. Use of ICT

5 %

7 %

22 %

4.5. Social connectedness

12 %

13 %

18 %

4.6. Educational attainment

8 %

7 %

1 %

4th domain: Capacity and enabling environment for active ageing

32 %

20 %

33 %

Source (Zaidi et al. 2013, p. 17) and personal calculation

To this end, the initiative that resulted in the construction of the AAI is more than praiseworthy. However, to create a measurement which will serve as the best tool for policy-making (Dobbie and Dail 2013; Soo 2013; Tarantola and Saltelli 2007), we need to mention certain weaknesses of the official methodology. This issue is connected to the fact that the very nature of AAI is a complex and composite, and that it consists of a number of different individual indicators. Moreover, the weighting system and the operating method in aggregating composite scores play a predominant role in the development of any composite index (Singh et al. 2012). However, in the construction of the AAI subjectively assigned weights are used.

A possible solution to the issue and the goal of this paper is to propose the application of a statistical Composite I-distance Indicator (CIDI) method (Ivanovic 1973; Jeremic et al. 2011; Dobrota et al. 2016) as a new methodology for the construction of the AAI. Namely, the main aims of this paper are (i) to review the AAI index by its weaknesses, (ii) to use a two-step CIDI methodology in order to create the index that is based on AAI, but creates the new objective weights for the original indicators, and (iii) to analyse the stability of the new index created by the two-step CIDI methodology and compare it to the official AAI methodology.

This paper is organised as follows. In the second section, the two-step CIDI method is presented. The third section covers the analysis of the obtained results, combined with the short comparative analysis of the results derived from the new proposed methodology and the official one. The fourth section of the paper gives a detailed discussion and states the limitations of the research. The final section of the paper underlines key conclusions and future directions of the study.

Problem Definition and Methodology

A possible remedy to the issue stated above is to propose the application of a two-step CIDI as a new methodology for the construction of the AAI. The benefits of using CIDI methodology are numerous. Besides its ability to calculate a single index (by taking into account many indicators) and consequently ranking entities (countries), CIDI methodology uses the Pearson’s coefficients of correlation, provided by the I-distance method, which are calculated from the I-distance values and input indicators (thus presenting the relevance of each input indicator). Instead of defining subjective weights to input indicators, the I-distance method determines which of the input indicators are the most important for the ranking process, by putting them in a particular order according to these correlations. This paper proposes a new ranking methodology which also identifies indicators and domains contributing most to the country’s rank, thus providing solid ground for decision-making and implementing policies. Obtained results will provide decision makers with a powerful tool which directs attention to the areas in which improvements and implemented policies could contribute most to the goals of Active Ageing. Also, the rank list of countries enables the identification of the most successful ones, which consequently could lead to the recognition and implementation of their best practices in other countries.

In this research, the two-step CIDI approach was performed (Maricic and Kostic-Stankovic 2016). Firstly, we applied the method on indicators compounding each of the four domains: Employment, Participation in Society, Independent, Healthy and Secure Living, and Capacity and Enabling Environment for Active Ageing. The obtained I-distance values for these domains were then included in the procedure, and the total I-distance value was determined. Second, having implemented this approach, we were also able to present correlations of each indicator with its I-distance domain value, thus providing a more in-depth analysis of underlying dynamics of each domain (Maricic et al. 2014). Additionally, correlations of each I-distance domain value with the total I-distance value gave us a clear picture of the importance of each domain for the calculation of the total I-distance value. Third, according to the CIDI methodology, the obtained Pearson’s coefficients of correlation were used as a base for the calculation of new weights which were integrated into AAI framework, consequently leading to more impartial rankings by each of the four domains and in total.

Two-Step CIDI Approach

Composite indices represent a research area that has captivated the attention of various stakeholders. They allow presentation and analysis of an individual phenomenon with only one global measurement, by taking into account different variables from the designated dataset.

A common objection against the composite indices is the subjectivity of the weights assigning (Paruolo et al. 2013; Saltelli 2007). As a way to overcome this issue, we suggest application of the CIDI method. The foundation of the CIDI methodology is the I-distance method which was introduced as a method for ranking countries by the level of their development (Ivanovic and Fanchette 1973: Ivanović 1977). The proposed solution takes into account many different variables, stressing the importance of their right combination to create one synthesised indicator which represents a rank. The method is based on the creation of a fictive entity, the one with the minimum values of each variable, and measuring the distance of other entities in the observed data set from the fictive entity (Dmitrovic et al. 2015; Horvat et al. 2015; Išljamović et al. 2015; Savic et al. 2016b).

For the selected set of variables X = × 1 , × 2 ,. .., x k, and the set of entities P = p 1 , p 2 ,. .., p n which need to be compared, the distance between any two entities P r and P s is defined as:
$$ D\left(r,s\right)={\displaystyle \sum_{i=1}^k\frac{\left|{d}_i\left(r,s\right)\right|}{\sigma_i}}{\displaystyle \prod_{j=1}^{i-1}\left(1-{r}_{ji.12\dots j-1}\right)} $$
where σ i is the standard deviation of X i , r ji.12,..., j−1 is a partial correlation coefficient between X i and X j , (j < i) and d i (r,s) is the distance between the values of variable X i for P r and P s , also known as the discriminate effect (Dobrota et al. 2015a; Jovanovic-Milenkovic et al. 2016),
$$ {d}_i\left(r,s\right)={x}_{ir}-{x}_{is}\kern2.5em i\in \left\{1,\dots, k\right\} $$
In addition, the square I-distance (Dobrota et al. 2015c; Jeremic and Jovanovic Milenkovic 2014) is frequently used and given as:
$$ {D}^2\left(r,s\right)={\displaystyle \sum_{i=1}^k\frac{d_i^2\left(r,s\right)}{\sigma_i^2}}{\displaystyle \prod_{j=1}^{i-1}\left(1-{r}_{ji.12\dots j-1}^2\right)} $$

A two-step CIDI methodology proposes establishing adequate weights for selected indicators. For this purpose, it is necessary to calculate I-distance values. Subsequently, we have examined the stability of each of the compounding indicators by calculating the Pearson’s correlations between the I-distance values and input indicators. Pearson correlations are used due to a special feature of the I-distance method: it can present the relevance of input indicators. Instead of defining subjective weights to input indicators, as in AAI, the I-distance method sets input indicators that are most important for the ranking process. It puts them in a specific order of importance according to these correlations. Also, since both the I-distance values and compounding indicators are continuous variables, Pearson’s correlations are suitable for performing these analyses (Dobrota et al. 2015b).

The new weights for each of the compounding indicators are formed by weighting the empirical Pearson’s correlations: values of correlations are divided by the sum of correlations. The final sum equals 1, thus forming a new and appropriate weighting scheme:
$$ {w}_i=\frac{r_i}{{\displaystyle \sum_{j=1}^k{r}_j}} $$

r i , i = 1,…k is a Pearson correlation between the i-th input variable and I-distance value. Thus, instead of subjectively defining the values of weights, CIDI is based on methodological and statistical concept defined by the I-distance method (Dobrota et al. 2015b; Savic et al. 2016a).

Results of the Two-Step CIDI Approach

The aim of the paper is to overcome the problem of potential bias and to propose a methodology that has several beneficial outcomes. In the beginning, we have to mention that because of the incompleteness of data, two countries, Bulgaria and Malta, were excluded from the research. Consequently, new implicit weights (Zaidi et al. 2013) were recalculated (implicit weight for an indicator is obtained by multiplying the value of explicit weight with the value of the indicator when aggregating the indicators to a domain-specific index; likewise, the implicit weight for each domain is derived from a multiplication of an explicit weight for the domain and the value of the domain-specific index). Furthermore, the new method was applied to the data obtained from the official database that was used for the calculation of 2014 AAI (UNECE/ European Commission 2015).

Table 1 contains 22 indicators grouped into four domains, which ultimately compile the AAI. In the official methodology, explicit weights were obtained by assuming their initial value then re-adjusting them so that the values of the resulting implicit weights match with those recommended by the experts (Zaidi et al. 2013). For this reason, both implicit and explicit weights should be taken into account when comparing with the two-step CIDI results.

The official methodology considers that the first age group contributes most to the value of the first domain (Employment). On the other hand, the result obtained through the application of the two-step CIDI method shows a shift of significance towards older age groups. The results can be explained if we take into account that the average age for retiring in Europe is 64 (OECD 2012). The majority of countries included in the analysis have active population until this age so that if the right policies were implemented, they would continue to be active. Therefore, if we consider that results obtained below 65 years are easily achievable, and if the goal of active ageing is to have a population active beyond that age, we can conclude that more emphasis should be given to the countries which accomplish a higher level of employment among the eldest age groups. Additionally, in many countries elderly choose to continue to work because of the possibility to combine income from work with a pension (Eurofound 2012; European Commission 2012a). Consequently, the financial incentive is often highlighted as the strongest reason to work after the retirement age (OECD 2011), and it should serve as valuable information for policy making.

Two-step CIDI method gave the most intriguing results of the domain Participation in Society. Calculated weights, shown in Table 1, differ from the ones provided by the official methodology. The most extreme change is the drop in importance of the indicator Political participation, resulting in the decreased rank of Sweden (see Table 3). Scandinavian countries are the leading ones according to the level of politically active older generation (Goerres 2009), followed by Anglo-Saxon and Central-European countries, while the Mediterranean and Eastern-European countries are at the bottom (Hoskins and Mascherini 2009).

On the other hand, the weights proposed for the caring activities increased significantly. Further, because of the availability of the data by genders, a more in-depth analysis was done. As it is commonly known, grandparents are very involved in the upbringing of their grandchildren, which consequently led to the increase in the number of fully-employed women (Lewis et al. 2008). Additionally, because women tend to involve more in the caring activities than men, especially in the care of other older adults (Altschuler 2001), the two-step CIDI considers this activity slightly more significant for them. This result also supports the stand that additional increase in the government expenditure on formal residential and home help for the elderly can lead to the increase in the labour force participation rates of women across Europe by relieving their informal care burden (Viitanen 2007). For the males, weights proposed by the official methodology are in better compliance with the results obtained by the two-step CIDI because they have a bigger diversification of their activities. Due to the fact that traditionally males and females involve in different types of activities, especially as they age, we consider that for this domain separate gender analysis need to be done (see Table 2).
Table 2

Weights for the domain Participation in Society, by gender

Indicator

Men

Women

2.1. Voluntary activities

22 %

23 %

2.2. Care to children, grandchildren

26 %

32 %

2.3. Care to older adults

28 %

34 %

2.4. Political participation

24 %

11 %

Results for domain Independent, healthy and secure living are similar to the ones from the official methodology, and the first four countries remain at the top of the chart with slightly shuffled ranks. Regarding the results obtained by the two-step CIDI approach (Table 1), we can see that the most important indicator is Independent living. Achieving high levels of this indicator is crucial for the fulfilment of the Active Ageing, and it is the obligation of every country to provide this continuously growing part of the population with an environment that enables pleasant independent living (Mui and Burnette 1994). Additionally, high level of significance is given to indicator Lifelong learning, and here Nordic countries excel (Rubenson 2006). The interconnection between these two indicators was best described in the Policy Framework for Active Ageing where it stands that: “Education in early life combined with opportunities for lifelong learning can help people develop the skills and confidence they need to adapt and stay independent as they grow older” (WHO 2002).

As opposed to these results, a subtle weight was assigned to the indicator Relative median income, and it implies that its exclusion can be considered in future research. Reasons for that can be found in the results for other indicators and the whole domain. Namely, a new methodology gives the highest significance to the independent living, physical activities, lifelong learning and medical support, which are the core activities described in the domain’s name. Furthermore, although the financial aspect should not be overseen, within this domain it can also be observed from other indicators that measure the poverty and material deprivation.

Finally, the official methodology uses fourth domain Capacity and enabling environment for active ageing as the foundation for the analysis of the first three, thus emphasising its importance (Sen and Nussbaum 1993; Sen 1985, 2009). After applying the proposed two-step CIDI method, different values for the significance of indicators were obtained. Indicators that most contribute to the value of the domain are the second, third and fourth indicator, with almost equal weights. Here, we can see that weight proposed for the Use of ICT indicator has tripled. The fast advancement of technology made ICT inevitable part of a modern life, which is why EU’s initiative “Independent life for ageing society” includes action plan “Ageing well in the information society” (EU 2010). It underpins the use of ICT as a means for prolonging work life, having socially active and creative life, and achieving a higher degree of independence in the older generation, which perfectly meet the objectives of Active Ageing.

The indicator Educational attainment also demands more attention. As this is an important indicator and the obtained results are surprising, we need to have a closer look at the structure of the data. It represents the percentage of older persons aged 55–74 with upper secondary or tertiary educational attainment. Namely, it covers generations born between 1936 and 1955 which had entirely different living and studying conditions. Also, these inequalities can be observed from the geographical aspect, where countries can be divided into three large groups: North-Western (Nordic countries, Germany, Netherlands and Great Britain), South-Western (Belgium, France, Greece, Portugal and Spain) and Eastern (Bulgaria, Czech Republic, Estonia, Hungary, Poland, etc.) (Bartusek and Koucky 2013). The two-step CIDI method shows that this indicator is particularly problematic, which is why we consider it should be additionally revised and perhaps substituted with a similar indicator.

Composite Index – Active Ageing Index

Objectively defined weights are the ultimate goal of this paper, and a resulting rank list of countries should serve as a measurement and as a confirmation of countries’ success in Active Ageing policies. Weights proposed by the new method can be compared with the ones defined in the official methodology. The results (see Table 1) show that methodologies somewhat similarly value the contribution of the domains to the value of the AAI.

The main difference between them is that the whole procedure for assigning weights is inverted. The proposed two-step CIDI method first calculates the I-distance values and afterwards the weights, while the official one defines explicit weights (using the Expert Group and implicit weights) and applies them for the calculation of the final rankings. After application of the proposed methodology, the most successful country is Sweden, followed by Denmark and Netherlands. The resulting rank list by each of the four domains and in total is presented in Table 3. In addition, hierarchical clustering (Ward’s method) has been performed and Silhouette approach (Rousseeuw 1987; de Amorim and Hennig 2015) defined four clusters as optimal solution (p < 0.05).
Table 3

Official ranks of AAI, two-step CIDI ranks for each of the four domains and total AAI ranks and scores, with appropriate cluster membership

Country

Rank official AAI

Rank

Score CIDI AAI

Cluster CIDI AAI

A

B

C

D

CIDI AAI

Sweden

1

1

9

2

1

1

56.224

1

Denmark

2

4

17

3

2

2

52.788

1

Netherlands

3

6

5

4

4

3

51.757

1

Finland

4

7

4

1

5

4

51.575

1

United Kingdom

5

3

13

11

6

5

48.924

2

Ireland

6

12

2

6

9

6

48.679

2

Luxembourg

8

21

7

7

3

7

48.540

2

France

9

17

3

5

8

8

46.555

2

Belgium

15

24

6

12

7

9

45.804

2

Austria

10

15

11

10

10

10

45.013

2

Germany

7

5

25

9

11

11

44.869

2

Spain

17

16

12

14

12

12

42.874

3

Italy

14

18

1

17

14

13

42.454

3

Czech Republic

12

14

8

15

15

14

42.290

3

Portugal

16

8

20

19

13

15

41.919

3

Cyprus

13

11

10

16

17

16

40.979

3

Estonia

11

2

23

18

22

17

40.709

3

Croatia

20

20

15

13

16

18

40.541

3

Slovenia

21

25

14

8

18

19

39.921

3

Hungary

23

26

16

20

21

20

37.241

4

Latvia

19

9

21

26

20

21

36.918

4

Lithuania

18

13

18

23

24

22

36.848

4

Slovakia

24

22

22

22

19

23

36.640

4

Poland

25

19

26

21

23

24

35.520

4

Greece

26

23

19

24

25

25

34.470

4

Romania

22

10

24

25

26

26

34.038

4

Domains - A: Employment, B: Participation in society, C: Independent, healthy and secure living, D: Capacity and enabling environment for active ageing

To measure the stability and permanence of the official and the two-step CIDI weighting scheme, we have performed its uncertainty and sensitivity analysis (Saisana and D’Hombres 2008; Saltelli et al. 2008; Saisana et al. 2011; Dobrota et al. 2015b; Dobrota and Jeremic 2016). The relative contribution is estimated as a proportion of an indicator score multiplied by the appropriate weight with regard to the overall entity score. In this particular case, we collected official values for each domain and performed the same procedure with official explicit domain weights and the proposed two-step CIDI domain weights (Table 1). Subsequently, we have calculated the average relative contributions and their standard deviations. Using the Monte Carlo simulation method, we have simulated the score results 10,000 times, each time recording the results. In addition, according to uncertainty and sensitivity methodology (Saisana and D’Hombres 2008; Dobrota and Dobrota 2016), we have counted the ranks for all countries, thus measuring the amount of uncertainty of the official and the two-step CIDI ranking results. The results of MC simulation are presented in Fig. 1 and on the left graph the two-step CIDI results can be seen. Our results clearly imply that the new approach establishes a more stable ranking of countries.
Fig. 1

Uncertainty and sensitivity of the two-step CIDI (left) and the official AAI (right) ranks

Figure 2 sums up and presents the final results (two most significant indicators within each domain, altogether with appropriate coefficients of correlation). The significance of each domain and the most important indicators are presented, and this can be used to determine the priority areas in which measures should be implemented.
Fig. 2

Proposed framework for re-engineering AAI (circles represent the given domains and subdomains/indicators while arrows represent Pearson’s coefficients of correlation)

Conclusion

From the perspective of the younger generation, the ageing population is a problem that affects the sustainability of the current pension and social protection system. However, for the people above 65, it is a reality which they have to live with. This is one additional reason why the Active Ageing policies are crucial: they “shift the focus of policy away from older people, as a separate group who have aged, to all of us, who are ageing constantly” (Walker 2002 pp.137). In order to provide the present and the future elders with the system adequate for successful ageing (Walker 2008; Walker and Maltby 2012; Foster and Walker 2015), Active Ageing policies (which in its current form overemphasizes employment) need to be more effective. Authors point out that “policies related to active ageing claim to be responding to the impact of the ageing of the population on the sustainability of pension systems by extending professional activity. However, not everyone wants to (or is able) to continue working beyond retirement age” (Madero-Cabib and Kaeser 2016, pp. 37). In a line with this, the AAI strives to present itself as a tool that measures the effectiveness of Active Ageing policies and serves as a foundation for the future decision-making.

Through the application of multivariate methods the official AAI was built, still failing to address the subject of subjectively defined weights. In this paper, we wanted to propose a method for calculation of unbiased weights that also show in which areas implemented policies would have the strongest impact. Our proposed CIDI methodology presents a more statistically sound weights for the each indicator/domain. Besides the benefits mentioned above, the CIDI methodology points out indicators that need to be further analysed or excluded from the AAI framework.

Finally, the results show that Scandinavian countries top the two-step CIDI ranking list. This has come as a no surprise, considering how broadly praised Scandinavian welfare system is. Its reliance on high taxes, from which the system of social protection is financed, provides the environment and society apposite for Active Ageing. Sweden has introduced new forms of private pension, reduced pensions for those who retire early, all with aim to maintain a sustainable pension system (Foster and Walker 2013). As previously mentioned, current generations must build a system in which Active Ageing will be imminent to their lives, but that does not mean that the elders of today have to wait for it. It seems that the official methodology is more oriented to the construction of the future system, and not taking into account the current situation for people aged above 55. As we are witnessing, globalisation and technology brought changes to everyday lives, and it is incomparable to the conditions in which previous generations grew old. For this reason, it is essential to implement policies that build a system useful for the current ageing generation, which will also serve as the basis for further improvements in the desired state. Also, this can serve as an argument for the future revisions of the weights.

Future improvements in the methodology depend on the inclusion of other countries, and according to the project Global AgeWatch Index (http://www.helpage.org/global-agewatch/), we can see that the initiative for it does exist. Furthermore, there are several directions for future research. Namely, I-distance posthoc analysis (Marković et al. 2016) could be done to revise the number of indicators and hopefully trigger further research on this subject. In addition, benefit-of-doubt model (Van Puyenbroeck and Rogge 2017) and findings of Amado et al. (2016) should be carefully considered and potentially implemented by AAI policy-makers.

As one of the limitations of the study, one could argue that the CIDI methodology creates data-driven weights (Dobrota et al. 2016; Maricic and Kostic-Stankovic 2016), leading to their fluctuations over the years. As values of indicators change over the years, weights are also susceptible to changes. As a possible remedy to the issue longer time series of data observation is a way to go, placing the ageing research high on the agenda of various stakeholders (Wahl et al. 2013). Another limitation of our study is concerning the availability of data. As mentioned before, because of the incompleteness of data, two countries, Bulgaria and Malta, were excluded from the research. Malta is particularly interesting since their Government “pursued a novel direction in ageing policy – shift form focus on elderly care to active citizenship issues” (Formoza 2016). Not only for the two mentioned countries, but it should be a priority to gather the data for the as large number of countries as possible. Efforts must be made to collect the missing data over the years and add to the growing multidimensional framework for ageing research (Fernández-Mayoralas et al. 2015).

References

  1. Altschuler, J. (2001). Meaning and centrality of caring activities among older women. Journal of Women & Aging, 13(3), 79–100. doi: 10.1300/J074v13n03_06.CrossRefGoogle Scholar
  2. Amado, C. A., São José, J. M., & Santos, S. P. (2016). Measuring active ageing: a data envelopment analysis approach. European Journal of Operational Research, 255(1), 207–223.CrossRefGoogle Scholar
  3. Bartusek, A., & Koucky, J. (2013). Access to a degree in Europe. In Inequality in tertiary education attainment 1950–2011. Prague: Charles University in Prague.Google Scholar
  4. Börsch-Supan, A., Härtl, K., & Ludwig, A. (2014). Ageing in Europe: reforms, international diversification, and behavioral reactions. American Economic Review, 104(5), 224–229. doi: 10.1257/aer.104.5.224.CrossRefGoogle Scholar
  5. de Amorim, R. C., & Hennig, C. (2015). Recovering the number of clusters in data sets with noise features using feature rescaling factors. Information Sciences, 324, 126–145.CrossRefGoogle Scholar
  6. Disney, R. (2016). Pension reform in the United Kingdom: an economic perspective. National Institute Economic Review, 237(1), R6–R14. doi: 10.1177/002795011623700111.CrossRefGoogle Scholar
  7. Dmitrovic, V., Dobrota, M., & Knezevic, S. (2015). A statistical approach to evaluating the bank productivity. Management: Journal for Theory and Practice Management, 20(75), 47–56. doi: 10.7595/management.fon.2015.0010.Google Scholar
  8. Dobbie, M. J., & Dail, D. (2013). Robustness and sensitivity of weighting and aggregation in constructing composite indices. Ecological Indicators, 29, 270–277. doi: 10.1016/j.ecolind.2012.12.025.CrossRefGoogle Scholar
  9. Dobrota, M., & Dobrota, M. (2016). ARWU ranking uncertainty and sensitivity: what if the award factor was excluded? Journal of the Association for Information Science and Technology, 67(2), 480–482. doi: 10.1002/asi.23527.CrossRefGoogle Scholar
  10. Dobrota, M., & Jeremic, V. (2016). Shedding the light on the stability of university rankings in the ICT Field. IETE Technical Review, 1–8. doi: 10.1080/02564602.2016.1144487
  11. Dobrota, M., Jeremić, V., Bulajić, M., & Radojičić, Z. (2015a). Uncertainty and sensitivity analyses of PISA efficiency: distance based analysis approach. Acta Polytechnica Hungarica, 12(3), 41–58.Google Scholar
  12. Dobrota, M., Martic, M., Bulajic, M., & Jeremic, V. (2015b). Two-phased composite I-distance indicator approach for evaluation of countries’ information development. Telecommunications Policy, 39(5), 406–420. doi: 10.1016/j.telpol.2015.03.003.CrossRefGoogle Scholar
  13. Dobrota, M., Savic, G., & Bulajic, M. (2015c). A new approach to the evaluation of countries’ educational structure and development: the European study. European Review, 23(04), 553–565. doi: 10.1017/S1062798715000277.CrossRefGoogle Scholar
  14. Dobrota, M., Bulajic, M., Bornmann, L., & Jeremic, V. (2016). A new approach to the QS university ranking using the composite I-distance indicator: uncertainty and sensitivity analyses. Journal of the Association for Information Science and Technology, 67(1), 200–211. doi: 10.1002/asi.23355.CrossRefGoogle Scholar
  15. EU (2010). Ageing well in the information society: action plan on information and communication technologies and ageing.Google Scholar
  16. EU (2011). Decision no 940/2011/EU of the European parlament and the council. Retrieved from http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ:L:2011:246:0005:0010:EN:PDF.
  17. Eurofound (2012). Income from work after retirement in the EU. Luxembourg: Publications Office of the European Union.Google Scholar
  18. European Commission. (2012a). EEO review: employment policies to promote active ageing. Retrieved from http://ec.europa.eu/social/main.jsp?catId=738&langId=en&pubId=6783&furtherPubs=yes.
  19. European Commission. (2012b). European year for active ageing and solidarity between generations. Retrieved from http://ec.europa.eu/archives/ey2012/ey2012main9ef0.html.
  20. Fernández-Mayoralas, G., Rojo-Pérez, F., Martínez-Martín, P., Prieto-Flores, M. E., Rodríguez-Blázquez, C., Martín-García, S., et al. (2015). Active ageing and quality of life: factors associated with participation in leisure activities among institutionalized older adults, with and without dementia. Aging & Mental Health, 19(11), 1031–1041.CrossRefGoogle Scholar
  21. Formoza, M. (2016). Responding to the active ageing index: innovations in active ageing policies in Malta. Journal of Population Ageing, doi:  10.1007/s12062-016-9163-1
  22. Foster, L., & Walker, A. (2013). Gender and active ageing in Europe. European Journal of Ageing, 10(1), 3–10.CrossRefGoogle Scholar
  23. Foster, L., & Walker, A. (2015). Active and successful aging: a European policy perspective. The Gerontologist, 55(1), 83–90. doi: 10.1093/geront/gnu028.CrossRefGoogle Scholar
  24. Goerres, A. (2009). The political participation of older people in Europe: the greying of our democracies. Basingstoke: Palgrave Macmillan.CrossRefGoogle Scholar
  25. Horvat, A., Antic, S., & Jeremic, V. (2015). A new perspective on quality characteristics determining supply chain management of coffee production. Inzinerine Ekonomika-Engineering Economics, 26(3), 239–244. doi: 10.5755/j01.ee.26.3.5462.Google Scholar
  26. Hoskins, B. L., & Mascherini, M. (2009). Measuring active citizenship through the development of a composite indicator. Social Indicators Research, 90(3), 459–488. doi: 10.1007/s11205-008-9271-2.CrossRefGoogle Scholar
  27. Išljamović, S., Jeremić, V., Petrović, N., & Radojičić, Z. (2015). Colouring the socio-economic development into green: I-distance framework for countries’ welfare evaluation. Quality and Quantity, 49(2), 617–629. doi: 10.1007/s11135-014-0012-0.CrossRefGoogle Scholar
  28. Ivanovic, B. (1973). A method of establishing a list of development indicators. Paris: United Nations educational, scientific and cultural organization.Google Scholar
  29. Ivanović, B. (1977). Teorija klasifkacije. Beograd: Institut za ekonomiku industrije.Google Scholar
  30. Ivanovic, B., & Fanchette, S. (1973). Grouping and ranking of 30 countries of sub-Saharan Africa, two distance-based methods compared. Paris: United Nations educational, scientific and cultural organization.Google Scholar
  31. Jeremic, V., & Jovanovic Milenkovic, M. (2014). Evaluation of Asian university rankings: position and perspective of leading Indian higher education institutions. Current Science, 106(12), 1647–1653.Google Scholar
  32. Jeremic, V., Bulajic, M., Martic, M., & Radojicic, Z. (2011). A fresh approach to evaluating the academic ranking of world universities. Scientometrics, 87(3), 587–596. doi: 10.1007/s11192-011-0361-6.CrossRefGoogle Scholar
  33. Jovanovic-Milenkovic, M., Brajovic, B., Milenkovic, D., Vukmirovic, D., & Jeremic, V. (2016). Beyond the equal-weight framework of the networked readiness index a multilevel I-distance methodology. Information Development, 32(4), 1120–1136. doi: 10.1177/0266666915593136.CrossRefGoogle Scholar
  34. Lewis, J., Campbell, M., & Huerta, C. (2008). Patterns of paid and unpaid work in Western Europe: gender, commodification, preferences and the implications for policy. Journal of European Social Policy, 18(1), 21–37. doi: 10.1177/0958928707084450.CrossRefGoogle Scholar
  35. Madero-Cabib, I., & Kaeser, L. (2016). How voluntary is the active ageing life? A life-course study on the determinants of extending careers. European Journal of Ageing, 13(1), 25–37.CrossRefGoogle Scholar
  36. Maricic, M., & Kostic-Stankovic, M. (2016). Towards an impartial responsible competitiveness index: a twofold multivariate I-distance approach. Quality and Quantity, 50(1), 103–120. doi: 10.1007/s11135-014-0139-z.CrossRefGoogle Scholar
  37. Maricic, M., Jankovic, M., & Jeremic, V. (2014). Towards a framework for evaluating sustainable society index. Romanian Statistical Review, 62(3), 49–62.Google Scholar
  38. Marković, M., Zdravković, S., Mitrović, M., & Radojičić, A. (2016). An iterative multivariate post hoc I-distance approach in evaluating OECD better life index. Social Indicators Research, 126(1), 1–19. doi: 10.1007/s11205-015-0879-8.CrossRefGoogle Scholar
  39. Moreau, J. R. (2016). Social security and pension reform: international perspectives. Eastern Economic Journal, 42(3), 496–497. doi: 10.1057/eej.2014.70.CrossRefGoogle Scholar
  40. Mui, A. C., & Burnette, J. D. (1994). A comparative profile of frail elderly persons living alone and those living with others. Journal of Gerontological Social Work, 21(3–4), 5–26. doi: 10.1300/J083V21N03_02.CrossRefGoogle Scholar
  41. Nordbakke, S., & Schwanen, T. (2014). Well-being and mobility: a theoretical framework and literature review focusing on older people. Mobilities, 9(1), 104–129. doi: 10.1080/17450101.2013.784542.CrossRefGoogle Scholar
  42. OECD (2011). Pensions at a glance 2011: retirement-income systems in OECD and G20 countries. OECD Publishing.Google Scholar
  43. OECD (2012). Statistics on average effective age and official age of retirement in OECD countries. Retrieved from http://www.oecd.org/employment/emp/ageingandemploymentpolicies-statisticsonaverageeffectiveageofretirement.html.
  44. Paruolo, P., Saisana, M., & Saltelli, A. (2013). Ratings and rankings: voodoo or science? Journal of the Royal Statistical Society. Series A, Statistics in Society, 176(3), 609–634. doi: 10.1111/j.1467-985X.2012.01059.x.CrossRefGoogle Scholar
  45. Rousseeuw, P. J. (1987). Silhouettes: a graphical aid to the interpretation and validation of cluster analysis. Journal of Computational and Applied Mathematics, 20, 53–65.CrossRefGoogle Scholar
  46. Rubenson, K. (2006). The Nordic model of lifelong learning. Compare: A Journal of Comparative and International Education, 36(3), 327–341. doi: 10.1080/03057920600872472.Google Scholar
  47. Saisana, M., & D’Hombres, B. (2008). Higher education rankings: robustness issues and critical assessment. How much confidence can we have in higher education rankings? (EUR23487 ed.). Italy: Joint Research Centre, Publications Office of the European Union. doi: 10.2788/92295.Google Scholar
  48. Saisana, M., d’Hombres, B., & Saltelli, A. (2011). Rickety numbers: volatility of university rankings and policy implications. Research Policy, 40(1), 165–177.CrossRefGoogle Scholar
  49. Saltelli, A. (2007). Composite indicators between analysis and advocacy. Social Indicators Research, 81(1), 65–77. doi: 10.1007/s11205-006-0024-9.CrossRefGoogle Scholar
  50. Saltelli, A., Ratto, M., Andres, T., Campolongo, F., Cariboni, J., Gatelli, D., et al. (2008). Global sensitivity analysis: the primer. New York, NY, NY: John Wiley and Sons. doi: 10.1002/9780470725184.Google Scholar
  51. Savic, D., Jeremic, V., & Petrovic, N. (2016a). Rebuilding the pillars of sustainable society index: a multivariate post hoc I-distance approach. Problemy Ekorozwoju, 11(1), 125–134.Google Scholar
  52. Savic, D., Bogetic, Z., Dobrota, M., & Petrovic, N. (2016b). A multivariate approach in measurement of the sustainable development of European countries. Management: Journal for Theory and Practice Management, 21(78), 73–86. doi: 10.7595/management.fon.2016.0007.Google Scholar
  53. Sen, A. K. (1985). Well-being, agency and freedom: the Dewey lectures 1984. The Journal of Philosophy, 82(4), 169. doi: 10.2307/2026184.Google Scholar
  54. Sen, A. K. (2009). The idea of justice. London: Penguin.Google Scholar
  55. Sen, A. K., & Nussbaum, M. C. (1993). Capability and well-being, the quality of life. Oxford: Oxford University Press.Google Scholar
  56. Singh, R. K., Murty, H. R., Gupta, S. K., & Dikshit, A. K. (2012). An overview of sustainability assessment methodologies. Ecological Indicators, 15(1), 281–299. doi: 10.1016/j.ecolind.2011.01.007.CrossRefGoogle Scholar
  57. Soo, K. T. (2013). Does anyone use information from university rankings? Education Economics, 21(2), 176–190. doi: 10.1080/09645292.2011.561626.CrossRefGoogle Scholar
  58. Tarantola, S., & Saltelli, A. (2007). Composite indicators: the art of mixing apples and oranges. In R. Munich (Ed.), Composite indicators – boon or bane. Germany: Statistisches Bundesamt.Google Scholar
  59. UN (2013). World population ageing 2013. UN: Department of Economic and Social Affairs, Population Division.CrossRefGoogle Scholar
  60. UNECE/ European Commission. (2015). Active ageing index 2014: analytical report. Report prepared by Asghar Zaidi of centre for research on ageing, University of Southampton and David Stanton, under contract with UNECE (Geneva), co-funded by EC’s directorate general for employment, social affairs and inclusion (Brussels).Google Scholar
  61. Van Puyenbroeck, T., & Rogge, N. (2017). Geometric mean quantity index numbers with benefit-of-the-doubt weights. European Journal of Operational Research, 256(3), 1004–1014.CrossRefGoogle Scholar
  62. Viitanen, T. K. (2007). Informal and formal care in Europe. IZA discussion paper no. 2648. Institute for the Study of Labor.Google Scholar
  63. Wahl, H. W., Deeg, D. J., & Litwin, H. (2013). European ageing research in the social, behavioural and health areas: a multidimensional account. European Journal of Ageing, 10(4), 261–270.CrossRefGoogle Scholar
  64. Walker, A. (2002). A strategy for active ageing. International Social Security Review, 55(1), 121–139.CrossRefGoogle Scholar
  65. Walker, A. (2008). Commentary: the emergence and application of active aging in Europe. Journal of Aging & Social Policy, 21(1), 75–93.CrossRefGoogle Scholar
  66. Walker, A., & Maltby, T. (2012). Active ageing: a strategic policy solution to demographic ageing in the European Union. International Journal of Social Welfare, 17, 117–130. doi: 10.1111/j.1468-397.2012.00871.x.CrossRefGoogle Scholar
  67. WHO (2002). Active ageing - a policy framework. The contribution of the world health organization to the second united nations world assembly on ageing, Madrid, Spain, April 2002.Google Scholar
  68. Zaidi, A., Gasior, K., Hofmarcher, M. H., Lelkes, O., Marin, B., Rodrigues, R., … Zolyomi, E. (2013). Active ageing index 2012: concept, methodology and final results. report produced in collaboration with UNECE (Geneva) and European commission’s DG for employment, social affairs & inclusion, (Brussels).Google Scholar

Copyright information

© Springer Science+Business Media Dordrecht 2016

Authors and Affiliations

  • Ivana Djurovic
    • 1
  • Veljko Jeremic
    • 1
    Email author
  • Milica Bulajic
    • 1
  • Marina Dobrota
    • 1
  1. 1.Faculty of Organizational SciencesUniversity of BelgradeBelgradeSerbia

Personalised recommendations