Social Barometer of Spain Methodology

The way the 180 indicators are presented, the methodology used to develop the synthetic indexes and the process for users to modify the indexes according to their own criteria, is all explained.

Social Indicators

The start point in this Social Barometer was an exploration of the methodology used in other countries, and also in Spain, to evaluate society’s wellness or unease. Taking this into account, we developed a methodological proposal that makes possible to elaborate synthetic indexes of the main dimensions of social life from a broad selection of indicators. The election of the indicators has been made possible because of a systematic search of the most adequate statistical sources in order to cover the chosen social spheres. The indicators should bear three qualities: to be accessible, to be reliable and to count with temporal series since 1994. Finally, 180 indicators have been used, distributed in different numbers according to the information available.

In the database, each indicator is shown on a standard sheet, that can be printed on a A4  format sheet, including a statistical series with the evolution between 1994 and 2006 (to be updated on a yearly basis), together with a number of links to access the indicator’s information sources, and in its case, linked to the operations used to obtain the series, together with the charts –published in the book- related with the indicator (and that also will be updated every year). The standard sheet contains the following elements (click on them to see the examples; clicking on the image its size changes):

• Sphere and dimension
• Indicator's number and name
• Link to the Internet statistical source where the data is collected. These links were available at the time when publishing this site, but we must take into account that internet sites sometimes expire or change its name, which could eventually deny access to them. In other cases, the entry can only be obtained via a fee based subscription (as the online historical database of the World Bank).
• Definition and eventuals precitions around the indicator. When it relates to a complex indicator, its components are listed. For instance, "Available national net income based on year 2000 constant euros” and “population officially registered in Spain” are used to develop the indicator “Available national net income per inhabitant”. We also point out if the statistical series has experienced any important methodological changes and how they influence o are resolved in order to assure the continuity of the series. On top of this, we stress the “provisional” or “advance” status of some of the data. 

• The following chart depicts the way the indicator variation means an improvement or a worsening in the social order, and consequentially, if the “best” figure is the highest or the lowest one. All chosen indicators have been selected under the condition of expressing an improvement or worsening of social values broadly recognized by the population (a more extensive explanation of this point can be found in the Introduction of the book Barómetro Social de España, 2008).

  Below this is the indicator’s statistical series, colored in yellow, usually referred to whole Spain (as it is written on the top) with accurate data from 1994 thru 2006 (to be updated on a yearly basis). On the right, a bar or line chart shows explicitly the indicator’s evolution. Both in the table as in the chart, the best datum in the series is highlighted with blue color, and the worst datum with red color. The eventual methodological changes or disruptions in the statistical series are represented with a continuous line among the corresponding years (using italics in the lowest part of the table on the year or years of the series disruption). On the highest row of the table, in gold color, it shows if the indicator is directly fed from the source or if contains our own made sources series. In this latter case, the source can be accessed using a hyperlink.

  Between two black lines we collect three inputs related to the indicator: better datum meaning, highlighted in blue in the table and in the chart; the worst datum, highlighted in red. If any data in the series were estimated, there would be highlighted in bold text and italics in the table and with white bars in the chart. The specific criteria followed to establish the each estimation is also explained (using interpolation between the precedent and following years, by means of the continuity of the tendency, by proximity or by extrapolation).

  Below is a table with the normalized series of the indicator and the data used  to normalize the series. The normalized values are necessary to develop the synthetic indexes.

  At last, we show the title of those charts appearing in the book Barómetro social de España and directly or indirectly related to the indicator. After a link you can access them and the corresponding data series.
•  

Synthetic indexes

The Social Barometer of Spain has elaborated 46 synthetic indexes, being 35 of them corresponding to social dimensions or groups, fed by the 189 indicators, and 11 to the general spheres, fed by the preceding 35 dimensions.Furthermore, while incorporating the data from 2009 is been included a synthetic global index of social welfare and three global sub-indices, as follows:

1.                      economic and environmental sphere (Income and Wealth, Employment and Environment).

2.                      social policies (Health, Education, Social Protection and Housing).

3.                      welfare conditions on collective level (Security and Justice, Participation and International Relations).

In these cases the criteria for establishing the weight of the spheres in the development of the indices based on the dominant social opinion, expressed by the older population in a survey “ad hoc” held at the state level for the social barometer of Spain. (see justification and results of the survey Colectivo IOÉ, Barómetro social de España, Traficantes de Sueños, Madrid, 2008, pp. 31-32).

The dimension indices are developed from the combination of some indicators related to some socially relevant issues, for instance the synthetic index of poverty (sphere of Income and Wealth), synthetic index of working conditions (sphere of Employment) or synthetic index of access to housing (sphere of Housing). The process used has been tested many times, together with a contrasting of information with experts, which has elaborated various alternative forms of implementation. The latest version of the Barometer uses the following operative form:

 In principle, our proposal was to standardize the basic series transforming a scale of 0 to 10. However, this method has the problem of a high sensitivity to variations and trends to transform small changes (suppose in a range between 12.8% and 13.4%) in large variations (between 0 and 10). To avoid such problems, our methodology currently operates differently depending on the variability of the basic series.

1)             The series with low variability (less than 25% between minimum and maximum values) are normalized at short ranges, in order to avoid large fluctuations in the normalized series. Furthermore, in the measure in which exists sufficient information for it, there are three scales, depending on, if the indicator is moving in low or bad social levels (scale from 1 to 3), intermediate values (scale from 4 to 6) or in high numbers or good values (range from 7 to 9). In practice, there are only taken low and high segments if there exist obvious arguments for it, otherwise is applied the intermediate level.

2)             The rest of the series are normalized by default on a scale form 2 to 8 (it is not been applied the series 0-10 as it is very unlikely that the absolute minimum or maximum is set to one particular end): therefore is given a “2” to the worst value and a “8” to the best value of the scale from the social point of view, for the rest of the series is been used the simple rule of three. This general rule does not apply when there are elements that allow us to locate the series in a range of variation socially qualified as “bad-negative” (in that case the values are transferred to a scale from 1 to 5) or “good-positive” (therefore are been transferred to a scale from 5 to 9).

Examples:

If the empirical data of the basic series are clearly positive from the social point of view is assigned the value “9” to the best value and “5” to the worst; the rest of the values obtained a scale used the simple rule of three. As i.e. the indicator of remittance of emigrants, in which Spain is at the head of Europe.

If the empirical data are clearly negative, as the unemployment rate in which Spain is at the last position in Europe, it is been assigned the value “5” to the best and “1” to the worst value of the series; the rest of data obtain a scale used the simple rule of three.

 
For second instance, we proceed to the aggregation of the normalized indicators for every specific dimension. For example, “Working conditions” (including 6 indicators, among them the temporary rate of Salary Earners), we give each one of these indicators a given weight, which addition must result in 10 (as the result is obtained in the scale 0 to 10).

 Formula to aggregate normalized indicators
(Dimension "Working Conditions")
 
 
Formula applied for year 1994:
 

 The operations in order to result into synthetic indexes of the eleven social spheres is similar to the one explained before, except that in this case the values added are the synthetic indexes of the dimensions corresponding to each sphere (when referred to “Employment” there are two dimensions: Employment Access and Working Conditions), giving to each of them a determined weight (using the same mechanics explained for the dimensions indexes). The resulting chart shows the general tendency between 1994 and 2006 of the corresponding sphere, and will be updated on a yearly basis.

 In the lower part of the pages, referred to the synthetic indexes of the eleven spheres, a link grants access to a chart that lets you see the time evolution of the general index (thick black lined) and of the indexes of the dimensions there are made of (colored thin lines). This way it can be seen which are the dimensions that influence on increases or decreases of the general index along the years. In case of modifying the indexes weight and/or the dimensions feeding the general index, the chart automatically changes, evidencing the new changes effect.

 The main advantage of the chosen procedure to produce synthetic indexes is its great sensitiveness to variations along time, as it moves between 0 and 10 differences between the best and the worst datum. So it works for knowing the tendency; in other words, if a given matter is doing better or worse, and the oscillations it has gone through a period of time. Nevertheless, it has two cons, not frequent, but not to be forgotten: on the one hand, the result of normalization when existing variations are very small, will be shown as huge differences (we will always find a “very good” and a “very bad” case). On the other hand, as combining indicators to produce a synthetic index, each one of them will be homogenized (“0” thru “10”), not showing if most of the database is positive, negative or intermediate (though to compensate this problem, we just need to check the series with no homogenization). When trying out which would be the best procedure, we tested other methods that tried to avoid the disadvantages before described, but we encountered some difficulties and/or more important application problems.

 All things considered, the indexes used are useful to detect trends but must be interpreted with caution, always referring to the base indicators and taking into account, whenever possible, broad reference frameworks (historical series, the position of Spain amidst the European context, the theories explaining phenomena) so we can value in a qualitative way the temporary trends that the data shows us.

How to modify the indexes weighting

The indicators aggregation stage necessary implies an element of subjectivity, as it must be decided which weight is attributed to each indicator. To mitigate the risk of the authors’ slipped into arbitrariness, the user is granted the opportunity to change the weight given to each indicator, as long as the adjustments as a whole add up 10. While doing this, results and charts are automatically updated.

EXAMPLE. If it is wanted to change the weight of the indicators feeding the synthetic index of the dimension “Working conditions” the Employment sphere worksheet must be downloaded, and the corresponding tab of the synthetic index of Working conditions must be opened. Then it is possible to change the values of the cells highlighted in yellow, which express the attributed weight of each indicator (as being the best/worst Working conditions of workers in Spain). We have rated with 3.5 points the stability/temporality of the employment (22 points to the indicator 6 and 1.5 to indicator 7); 3.5 points to the workers’ salary (2 points to indicator 8 and 1.5 to indicator 9), and 3 points to security/labor health (1.5 point to indicator 10 and 1.5 to indicator 11). The user might think, for instance, that more weight should be given to “Salary purchasing power” (indicator 8) and less for “Labor illnesses with sick leave”. In other words, instead of attributing each 2 and 1.5 points, you match them with, for example, 2.5 and 1 point respectively (when writing these figures in its respective cells). When doing this, the results will automatically change in the tables and charts related to “Working conditions” dimension and in the “Employment” sphere.

It was also possible to detract or to add indicators, but this operations are more complex, as it requires the user to re-write the formulae so in one hand the series will normalize, and on the other hand, to calculate the indexes.