Skip to end of metadata
Go to start of metadata

 

 

 

 

 

 

 

 

 

 

 


 

 

 

 

 

 

 

 


 

 

 

 

 


 

 

 

 

 

 

 

 


 

 

 

 

 

 

 

 

The Active Ageing Index for Canada, Iceland, Norway and Switzerland

The calculation of 22 indicators of AAI, its domain scores and overall value for the four countries in question was undertaken within the research area of activities of AAI project. The calculations were implemented under a consultancy contract with the Southampton University and led by prof. Asghar Zaidi.

The idea behind this research activity was to explore possibilities of calculating AAI based on the data produced in UNECE countries from outside the European Union (EU) but with statistical systems not entirely different from the EU one and at least partly covered by the surveys used for the original AAI.

Given the lack of readily available data for Canada, Norway and Switzerland the work of the research consultant included extensive consultations with national statistical offices and other relevant agencies, as well as focal points on ageing in the named countries.

Computation of AAI for Canada turned out to be the most problematic given it is not covered by any of the surveys used for calculation of AAI for the EU countries. The research consultant worked closely with the Statistics Canada and the team of the Canadian Longitudinal Study on Aging (CLSA). These provided data for the majority of the indicators for the year 2010 (that is for 2012 AAI). The consultant urged the Statistics Canada to do the calculations using the more recent data, i.e. from the year 2012. The indicator 3.6 No Severe material deprivation appeared not possible to calculate. This should not prevent, however, the calculation and use of the third domain score and the overall AAI, given the weight of the indicator, and the fact that there are two more indicators referring to the financial security, namely 3.4 and 3.5. The main data sources used for Canada are Labour Force Survey, Survey of Labour and Income Dynamics, National Household Survey, Canadian Longitudinal Study on Aging (CLSA), and General Social Survey. The available results would have placed Canada at the second place in comparison to 28 EU countries. The comparison however is limited. Canada has particularly high results in Social participation and Capacity for active ageing domains.

Calculations for Iceland did not present difficulties as the country is covered with all the same surveys the EU countries are. Data are missing for the indicators 2.1–2.4 and 4.3 for the year 2007, as Iceland was not participating in the second wave of EQLS. Similarly, Iceland did not participate in the fifth wave of ESS in 2010. The AAI results for Iceland are very high in comparison to 28 EU countries. The only domain where Iceland has the result 0.5 points below EU maximum is the third domain; in the other three domains, results for Iceland are higher than for EU, which would result in the overall AAI score of 51.5 points. For comparison Sweden, the leading county among EU, has 44.9 points as overall AAI value.

Calculations for Norway were somewhat limited by the fact that Norway was not covered by the third wave of EQLS. The calculations are still possible on the basis of the 2007 EQLS. The research consultant also looked at possibilities of using alternative data sources for the concerned indicators, namely 2.1–2.4, 3.1 and 4.3, for which he used SILC and Health Interview Survey (2.1–2.4) and ESS (3.1 and 4.3). If these indicators are calculated based on second wave of EQLS, the results for Norway would place it above leading country in EU (Sweden) with 45.9 points for overall AAI, with the only domain score not higher than the maximum among the EU countries being Social participation. If the alternative sources are used, this domain score would be 1.2 points lower, not significantly influencing the overall AAI.

Calculations for Switzerland were limited as the country did not participate in EQLS at all. The research consultant worked in cooperation with the Swiss Federal Statistical Office and the Swiss Foundation for Research in Social Sciences (FORS). The following sources were used: Indicator 2.1 — Freiwilligen-Monitors, 2009; Indicator 2.2 — Family and Generations survey, 2013; Indicator 2.3 — Health survey, 2012; Indicators 2.4 and 3.1 — ESS; Indicator 4.3 — Special module of SILC, 2013. In comparison to EU Switzerland would be at the same place as Sweden with 44.9 points for overall AAI.

2014 AAI: indicators, domain scores and overall value

Executive summary of the report on the pilot study

UNECE/European Commission (2016) “Extending the Active Ageing Index to the local level in Germany: Pilot study”, Report prepared by Jürgen Bauknecht, Elias Tiemann, Jan Anye Velimsky of the Institute of Gerontology at the Technical University of Dortmund, under a contract with the United Nations Economic Commission for Europe (Geneva), co-funded by the European Commission’s Directorate General for Employment, Social Affairs and Inclusion (Brussels). See full report here

Key findings

It is possible to calculate a local Active Aging Index (AAI) based on German secondary data. However, using the nine data sources available during the project duration, the number of analysable territorial entities is between 20 and 30 (out of 403) and numbers of respondents are low in a few cases. Both problems could be solved with access to further data sources.

Background

The Active Ageing Index quantitatively depicts active ageing outcomes. It was calculated for all 28 European Union (EU) member States (for the purpose of this report, the AAI for the 28 EU countries is called EU-AAI) and for a number of non-EU countries for several years (2008, 2010, 2012). The index consists of 22 indicators, categorised into four domains, three of them assessing achievements of active ageing and the fourth one reflecting conditions for active ageing. To national policymakers and stakeholders the index can provide valuable information on their country´s strengths and weaknesses concerning active ageing, which can encourage appropriate action. However, several aspects of active ageing are mainly affected by local rather than national circumstances. A calculation at the local level for local policymakers and stakeholders could therefore provide crucial insights.

Pilot study

In the course of the pilot study ‘Gerontology Study – Extending the Active Ageing Index to the local level in Germany’ (carried out under the joint management project of the European Commission and the United Nations Economic Commission for Europe (UNECE)), a replication of the EU-Active Ageing Index at German NUTS[1] 3 level was conducted, based on secondary data analyses. The main goal of the pilot study was to ascertain the methodological feasibility of such calculations. The main advantage of an index calculated with secondary data is the low cost compared to primary data collection and therefore the high sustainability. Once established, a periodic re-calculation of the index would be possible with minimal effort.

Methodological proceedings

Territorial entities analysed: Initial selection

The German NUTS 3 level consists of 295 counties (including 3 ‘Special regional associations’) and 107 cities. Their mean size is about 35x35 kilometres (counties) or 12x12 kilometres (cities). Their population size is between 34,000 and 3.5 million, with a mean of about 200,000. Despite this mean size, only 87 out of the 403 territorial entities have more than 200,000 inhabitants. The 50 most populous counties and the 38 most populous cities entered the initial analysis.

Use of several surveys per indicator

Due to the high number of NUTS 3 units and the left-skewed distribution of their population size, most of the 88 NUTS 3 units initially selected have a very low population size when compared to the national population size. Although sampling in large-n surveys is not clearly geographically representative so that not all NUTS 3 units are represented in country-wide surveys and so that some numbers of respondents are higher than their population size would suggest, the overall expectation was that numbers of respondents would be low in the selected territorial entities. This was further aggravated by the restriction to persons 55 years old or older. Therefore, in order to increase numbers of respondents, in contrast to EU-AAI it was decided that several surveys per indicator would be used when possible for this pilot study.

Data sources

The Active Ageing Index is based on six data sources. For only one of them (European Social Survey) data at the German NUTS 3 level is available. Therefore, for the local level AAI new data sources had to be found. Out of 26 surveys fulfilling the criteria of (a) being repeatedly conducted and (b) covering the whole of Germany, nine surveys had readily available data at the NUTS 3 level.

Variables and answer categories

In these surveys, questions corresponding to those used for EU-AAI were identified. The relevant dimension was the ‘Goal (rationale)’ of the indicators defined in the EU-AAI methodology.[2] In EU-AAI, respondents are categorised into two groups, those being ‘active’ and those not. ‘Active’ here refers to the share of the population who contribute to the economy and society via participation in the labour market, unpaid activities (volunteering and care provision), through living independent lives, and who at the same time are enabled to do so by the environment they are living in. For example, the survey question on physical exercise/sports contains four answer categories. In order to categorise respondents into two groups, only those selecting category 1 (every day or almost every day) were categorised into the ‘active’ group and all the others – into the ‘inactive’ group. The categorisation was implemented in a similar way at the local level.

Weighting

The overall AAI results from the four domain scores are weighted according to theoretical considerations. The domain scores are calculated based on indicator values, which are also weighted according to theoretical considerations. In the local level AAI, as a result of using several variables from different data sources per indicator, another weighting step had to be added. This was done based on the inverted standard error of the surveys providing data for a given variable. This comes close to the square root of the number of respondents. Surveys with higher numbers of cases had a higher weight, yet with diminishing marginal increases.

Data gaps

Surveys covering the whole of Germany are not geographically representative at the NUTS 3 level. This implies that some territorial entities are not represented in some surveys. Resulting data gaps for single variables can distort indicator values and therefore domain scores and the overall AAI. For example, if for a given indicator for territorial entities A and B data are available for variables 1 and 2 and for territorial entity C just for variable 1, the indicator score would be distorted if the mean values between variables 1 and 2 differ. If the mean value for variable 1 is lower than for variable 2 (e.g. if variable 1 measures participation in political groups and variable 2 measures electoral participation), then the value for territorial entity C would be too high because only data for electoral participation would be available. In order to correct this, values were weighted so that mean values are similar.

Territorial entities analysed: Final selection

As expected, due to low numbers of respondents and data gaps, not all 88 territorial entities could be used for the final analysis. Error scores for each territorial entity were calculated in a way that territorial entities with high numbers of respondents in surveys with low standard errors had the lowest scores. This effectively resulted in low error scores for territorial entities with high numbers of respondents, or in other words territorial entities with high population numbers. The 30 territorial entities with the lowest error scores were used in the final analysis. 

Comparability between EU-AAI and the local level AAI

Due to different variables used and different dichotomisation of respondents into the categories, in most cases mean values of the 30 territorial entities differed from the German value in EU-AAI. Exceptions are the four indicators of the domain ‘Employment’ (employed yes/no in four different age groups) and the indicator ‘Independent living arrangements’.

Findings on the 30 territorial entities analysed

Results for the overall AAI value and the four domains were correlated with disposable income per person and population density. For the latter, positive, yet very weak, correlations have been found, so that there is a weak and statistically insignificant relationship between AAI score and urbanity. Only in the case of domain 4 (‘Capacity and enabling environment’) the correlation is significant at the 10% level, but with 30 cases also this is sensible to outliers. In contrast, there is an (expected) positive and statistically significant (5% level, Pearson´s r at around 0.5) relationship between disposable income and AAI scores for the overall score and all domains but domain 2 (‘Participation in society’), where no relationship between AAI score and wealth could be found. Therefore, the top of the Overall AAI table is populated by affluent Southern regions (Esslingen, Stuttgart, Rems-Murr, Rhein-Neckar and Frankfurt/Main if the latter can be categorised as ‘southern’) and the bottom – by regions in Eastern Germany (Halle, Mittelsachsen, Zwickau), still economically lagging behind Western Germany, and the Ruhr area (Dortmund, Duisburg) – a region bearing economic problems due to the decline of the steel and coal industry. This pattern prevails for the overall AAI score as well as three domains. However, the rank of regions is not without surprises. For example, in the overall AAI the cities of Dresden and Berlin rank above the affluent cities of Nürnberg and Düsseldorf. As the absence of any relationship to disposable income suggests, for domain 2 the ranking looks different, for example, with Eastern regions Chemnitz and Bautzen amongst the top 5 regions, and Frankfurt and Stuttgart below the middle of the table. It can be clearly concluded that there is the correlation with disposable income, which is expected given the construction of the index and the various components related to affluence, but disposable income does not clearly determine the scores. Affluent regions can have comparatively low scores and less affluent regions can have comparatively high scores.

Concerning the results and the ranking of the regions, it has to be mentioned that this results from a pilot study with (partly) low cases numbers. Further work on an improved data situation can lead to results with higher reliability.

Conclusion

It is possible to calculate a local active ageing index in Germany based on secondary data, yet this comes with various limitations. The most crucial limitation is the low case numbers for some indicators or territorial entities. The number of NUTS 3 areas for which index values can be calculated is somewhere between 20 and 30, depending on one’s perspective concerning how many respondents are necessary for the calculation of a score for a territorial entity. Most, but not all, indicator values can be interpreted content-wise thanks to sufficient numbers of respondents, i.e. they can be interpreted as first findings on the active ageing situation in the territorial entities. The low number of respondents restricts intertemporal comparison, since, for example, changes in low-n indicators can result from different samples and not reflect real changes. One solution could be the accumulation of different survey rounds. This would imply a less frequent re-calculation, perhaps once every four years instead of once every two years.

 

 



[1] Nomenclature des unités territoriales statistiques or Nomenclature of units for territorial statistics.

Dataset developed within the framework of the pilot study

A reference to the report should be made when using the data.