International comparison of active citizenship by using Twitter data, the case of England and the Netherlands
First Monday

International comparison of active citizenship by using Twitter data, the case of England and the Netherlands by Cristina Rosales Sanchez

Can social media become the new data source for certain social indicators? What does social media offer in comparison with classical sources for official statistics? This research analyzes the potential of Twitter data to obtain indicators of active citizenship. We use a methodology developed for a case study in Spain and test its replicability by applying it to England and the Netherlands. Twitter data offers an advantage to assess change in active citizenship over time and at high spatial resolution, while survey data — traditional sources for official statistics — are more authoritative and robust. Collecting and analyzing Twitter data is also faster than traditional survey processes. However, we found that Twitter usage specificities in each country ease or hinder the process, affecting the time to secure an indicator. Whilst future research will focus on other social indicators and sources of data, this research is an important step in developing an evidence-based understanding of the strengths and weakness of social media to inform policies.


I. Introduction
II. Objectives
III. Methodology
IV. Official data identification and data preparation
V. Analysis of official data
VI. Analysis of Twitter data
VII. Assessment
VIII. Discussion and conclusions



I. Introduction

Recent developments on information and communication technologies (ICT) have changed the way individuals communicate, work, and develop their daily lives. Accessibility to Internet-connected mobile devices, such as smartphones or tablets, allows instant communication and generates real-time geo-located data (Kamel Boulos, et al., 2011; Elwood, et al., 2012; Goodchild and Glennon, 2010). This user-generated content (UGC) contributes to a “big data” landscape, in which not only the volume or velocity of data are of importance, but also the variety of sources and data typologies to be encountered. The unprecedented availability of data leads to an emerging field of research in data science.

The potential of new data sources has attracted studies from a variety of domains. Research ranges from the integration of UGC into decision-making processes during disasters (Bruns and Liang, 2012; Goodchild and Glennon, 2010; Roche, et al., 2011) to public health monitoring (Culotta, 2014), exploitation of Twitter data to improve systems for fire detection (Craglia, et al., 2012), or track activities of tourists through an analysis of photo-sharing and mobile telephony (Girardin, et al., 2008).

Regarding the potential of new data sources, this research focuses on the strengths and challenges of using social media data to obtain indicators of urban quality-of-life.


Recent studies in official statistics focus on the potential of integrating new data sources to official statistics processes (Agafiţei, et al., 2015; Reis, et al., 2014; Struijs, et al., 2014; Tam and Clarke, 2015; Lalor, 2014). Data availability has been a constant constraint for generating indicators for policy-making and socio-economic development. Measuring the quality-of-life is a good example where official statistics struggle to provide relevant and timely data. As a proxy for quality-of-life, often GDP per capita is considered, based on the rationale that a wealthy country benefits quality-of-life. Using GDP as an indicator presents an advantage of worldwide data availability, facilitating spatial and temporal comparisons. However, GDP ignores certain aspects for policy analysis, such as environmental sustainability and societal inclusion (Costanza, et al., 2009).

Aware of the GDP limitations to measure quality-of-life, recent initiatives called Beyond-GDP (Organisation for Economic Co-operation and Development (OECD), 2013; Stiglitz, et al., 2009) propose new ways to understand living environments, considering environmental, social, and economic aspects (European Statistical System, 2011). These initiatives recognize the multidimensionality of quality-of-life and propose measurements that include the assessment of different integrated dimensions. Eurostat, for instance, considers nine different dimensions for quality-of-life (De Smedt, 2013; Eurostat, 2017): material living conditions, productive or main activity, health, education, leisure and social interactions, economic and physical safety, governance and basic rights, natural and living environment, and overall experience of life.

Beyond-GDP initiatives however, face major challenges regarding data availability. Traditional data sources for quality-of-life indicators are mainly statistics gathered from census activities and individual surveys, which can be expensive and time-consuming as well as burden respondents. In this context, new data sources faciltated by recent ICT developments offer potential solutions that may enrich the collection of data to analyse quality-of-life.

Research in this field is supported by international statistical organizations. The Global Working Group (GWG) on Big Data for Official Statistics (United Nations. Department of Economic and Social Affairs. Statistics Division (UNStats), n.d.), UNECE High-Level Group for the Modernisation of Official Statistics (HLG-MOS) (United Nations Economic Commission for Europe (UNECE), n.d.), or the European Commission’s Scheveningen Memorandum on ‘Big Data and Official Statistics’ (European Commission. Eurostat. Collaboration in Research and Methodology for Official Statistics (CROS), 2013) are examples of international efforts promoting research on the possibilities of integrating new sources of data into official statistics.



II. Objectives

Our research aims at analyzing the potential of social network data for quality-of-life indicators. We focus on active citizenship as one of the indicators included in the “governance and basic rights” dimension of quality-of-life, according to Eurostat (Rosales Sánchez, et al., 2017). Active citizenship is a relevant aspect of democratic societies and its measurement offers a sense of citizen involvement in their communities.

Our study utilizes a methodology that offers measurements of active citizenship at a local level using data gathered from Twitter. The method was developed in a case study based in Spain (Rosales Sánchez, et al., 2017), and will be tested for other countries to assess its validity for transnational studies.

Specifically, this study aims at answering the following questions:

  • What lessons can be extracted from this methodology about the use of Twitter data as a proxy of active citizenship, compared with traditional data sources?

  • To what extent is this methodology replicable in different countries; What are the main issues encountered in replicating this method?



III. Methodology

According to Hoskins and Mascherini (2009), active citizenship can be assessed by considering the Active Citizenship Composite Indicator (ACCI), which include four dimensions: participation in political life, protest and social change, community life, democratic values and representative democracy. The method applied in this study focuses on measuring one of the indicators included in the “community life” dimension of the ACCI, namely “contacting a politician”. Official data sources traditionally assess this indicator based on survey data, in which questionnaires ask individuals if they contacted a politician over the past 12 months.

The present research includes the following steps:

1. Evaluation of official data available and accessible in different countries. Those with accessible official data are chosen as targets for research;
2. Data collection from official sources and data processing;
3. Data collection from unofficial sources (Twitter) and data processing;
     3.1. Tweets data collection;
     3.2. Data processing, including:
          — Twitter data filtering, considering only information indicative of active citizenship.
          —Estimating geolocations of Twitter data. Usually only a small percentage of Twitter messages offer geolocation data. This method includes a geocoding process to investigate a larger number of Twitter messages.
4. Assessment. Evaluation by comparing results from official data sources and Twitter data, including differences between different countries.

Figure 1 summarizes the workflow to develop the comparative analysis between Twitter and official data sources to generate indicators of active citizenship. The workflow includes steps from data collection to data cleaning and analysis, as specified in Rosales Sánchez, et al. (2017).


Workflow to measure indicator
Figure 1: Workflow to measure the indicator “contacts to politicians to express opinions” using official and unofficial data sources.
Note: Larger version of figure available here.


1. Evaluation of official data accessibility in different countries and selection of targets
This research focused on the survey indicator “contacts to politicians to express an opinion”. In this regard, we analysed the availability and accessibility of official data sources on active citizenship at the city level in different countries of the European Union. Data availability from Italy, Germany, the Netherlands, and United Kingdom were analysed initially.

2. Data collection from official sources and data processing.
Traditionally, official data sources on active citizenship are based on surveys. When available and accessible, data at the level of individual respondents (microdata) is usually offered in proprietary statistical software formats, as SPSS or STATA. Packages for data management and statistical analysis — such as “haven” (Wickham and Miller, 2015), “sjmisc” (Lüdecke, 2015a), and “sjPlot” (Lüdecke, 2015b) — permit importing, analyses, and visualizations of microdata into the open source software environment R (
We use microdata to obtain a measure of the target indicator. In this case, the value of the indicator “contacts to politicians” for a specific city is the percentage of respondents who indicated having contacted a politician divided by the total number of respondents from that city:

Contacted a politician (city, year) (%) =

formula 1

Survey weights are considered to make statistics more representative of a given population.

3. Data collection from unofficial sources and data processing.
Twitter facilitates instant communication between its users, allowing for tweets to be transmitted even when relationships between users are not reciprocal. To address a tweet to a specific user, a sender can include the combination of the “@” sign with the Twitter username of the targeted user (@username), what in the Twittersphere is known as a “mention”. The consistent form of mentions enables an automation of text analysis, identifying elements within tweets (Highfield and Leaver, 2015). A user mentioned in a tweet is notified, and the tweet is publicly available through a Twitter search. Thus, the methodology adopted in this research considers tweets containing those “mentions” as direct communications.
As such, this step aims to obtain a measurement of the indicator “contacts to politicians” by considering tweets sent by individual citizens, in a specific region, that contain “mentions” of politicians and municipal representatives in that region. This includes the following assumptions:

  • An individual citizen — intending to contact a politician to express an opinion — would send a tweet including a mention of local government representatives or Twitter official accounts from municipalities (usually described on municipal Web sites).
  • Tweets should be generated by citizens. The method includes a filtering process to select only those tweets generated by Twitter user profiles categorized as individual citizens. Accordingly, messages generated from Twitter accounts of politicians, political parties, media, brands, or corporations are excluded.
  • Tweets should be sent by citizens at a specific location. The objective is to obtain an indication of citizen engagement at a specific location, i.e., in selected cities, thus the method considers the geolocations of tweets.

Under those conditions, the measurement of the indicator “contacts to politicians” using Twitter as a data source, includes the following steps (see Figure 1 ):

3.1. Tweets data collection
We collect tweets using the Twitter Streaming API and store tweet information in a PostgreSQL database. The tweets collected include related metadata, crucial for further analysis. Metadata includes required fields, i.e., tweet numerical identifier, user identifier, username, tweet text, time of tweet generation, number of followers, and number of tweets sent by a given user. Optional (but relevant) fields are retained as well, i.e., user description, user location, and geographical coordinates associated with the tweet.

3.2. Data processing
For the indicator we will only consider those tweets sent by individual citizens from selected cities. With that aim, the method includes two steps: Twitter user categorization and geographic location.

Categorization of Twitter users
We investigated metadata of tweets to identify Twitter profiles not corresponding to individual citizens (hereinafter, non-common citizens or “others”). The method excluded tweets from Twitter users representing an organization, brand, political party, institution, special interest group, other politicians, media, or journalists. To categorize the Twitter users, we evaluated particularly the presence of specific keywords in the user description on Twitter profiles.
Two types of keywords were applied. First, we focused on the most popular Twitter users in data collection (number of followers greater than the 99th percentile, equivalent to more than 12,500 followers). We consider that a high number of followers could be indicative of non-common citizens. From the 100 most frequently used words to describe those users, we selected a group of significant keywords closely related to non-common citizens, like “official account”, “deputy”, “journalist”, “newspaper”, and “information”. For example, in Table 1 the Twitter user whose description included the word “Councillor” was categorized as “others”, while the user whose description did not include any of the selected keywords was considered as a “citizen”.
The second type of keywords was related to political parties, politicians, and municipal representatives targeted by our study. This method assumed that accounts showing those particular names in their user descriptions would have special associations with them. Thus, they were classified as “non-common citizens”.


Table 1: Examples of Twitter users categorized as “citizen” or “others”.
Metadata fieldsTwitter user classified as “individual citizen”Twitter user classified as “other”
user nameNeil Arre!Joe Cullinane
user descriptionLaughter is contagious. Lets start an epidemicLabour Councillor for Kilwinning. @scottishlabour and @scotcoopparty candidate for Cunninghame South in 2016. Socialist. Trade unionist. Dad to Rosie.
user locationManchester 
user followers count221930
user friends count3581,279
user statuses count1,8826,743
Text@ManCityCouncil please may we have a paved foot path across this patch of grass, everyone would be grateful :)
On a serious note, huge congrats to @CllrJimMcMahon and all those who campaigned for him. Great result #IbackJim #Labourwin #OldhamWest
created at14/02/2016 14:0604/12/2015 08:59
user created at23/12/2010 01:4309/09/2011 21:57
geo coordinates latitude00
geo coordinates longitude00
place name  


This user categorization phase was facilitated by the use of data and text analysis packages in the R environment, as “plyr” (Wickham, 2011), “tm” (Feinerer, et al., 2008), and “stringr” (Wickham, 2015).

Twitter user geolocation
Twitter users can elect to make their geolocations public. Generally, only one percent of tweets are geolocated (Leetaru, et al., 2013), offering latitude and longitude data which can be highly accurate when using Twitter from a device equipped with a global navigation satellite system (GNSS). However, the geolocation of tweets can also be determined by analyzing information about locations contained in other metadata attributes, i.e., tweet location and profile location (Twitter Developer, n.d.). To do so, this research uses the geocoder library (, which allows exploiting different online geocoding services, such as Google, Bing, OpenStreetMap, and Geonames, among others. These geocoding services parse textual words and phrases in tweet attributes and assigns to them geographic identifiers, i.e., latitude and longitude information.

4. Assessment
As stated previously, this methodology was developed in a case study in Spain (Rosales Sánchez, et al., 2017). We assessed its replicability by comparing its execution and results for different countries selected for this research (including the case of Spain from the previous study). The comparison was completed between different types of data sources (official vs. unofficial). The evaluation was based on assessment criteria, adapted from the Australian Bureau of Statistics “Data quality framework” (Australian Bureau of Statistics, n.d.), that considered the following:

  • accuracy (assessed in terms of the major sources of error leading to inaccuracies);
  • scope and coverage (identification of target population, representativeness of data, biases);
  • spatial dimension (spatial coverage and granularity);
  • temporal dimension (period of data collection, exceptions during collection and granularity);
  • timeliness (any delay between data collection and data availability and frequency of data collection);
  • coherence (data comparability with other sources of information and consistency over time), and;
  • cost and effort (amount of work and time required to define, develop, apply, and reproduce the method).



IV. Official data identification and data preparation

Looking for official data on active citizenship in Italy, Germany, the Netherlands, and United Kingdom, we found that the main sources were usually national institutes of statistics, national institutes of social science, centres of social research, and research data repositories. In general, these national data sources offered well-designed Web platforms for data access with clear information on how to access data, conditions for use, and contact points. Access to microdata usually required registration as a user in a given online system and acceptance of specific agreements regarding data privacy.

Variables related to active citizenship aimed at offering a national perspective such as part of international surveys, like the ISSP (International Social Survey Programme) or the European Social Survey (ESS); or national surveys aimed at providing a regional vision. However, those sources usually offered small sample sizes at a city level; thus using their data became problematic due to low representativeness and privacy concerns. Official sources can even block access to microdata, as was the case for the German General Social Survey (ALLBUS) and the Dutch survey on social cohesion and well-being (sociale samenhang). In addition, economic restrictions limited access to detailed data. Microdata from the Dutch Parliamentary Election Study (NKO), initiative from the Centraal Bureau voor de Statistiek (CBS), and the Foundation of Electoral Studies in the Netherlands (SKON), were only accessible by the use of On Site and Remote Access facilities, with high associated costs.

Other sources on active citizenship, such as the Italian “Aspetti della vita quotidiana” (launched by Istat) and the German Survey on Volunteering (FWS, developed by the German Centre of Gerontology) were not taken into account because they did not include “contacts to politicians.” Finally, only two official sources presented available and accessible data on “contacts to politicians” at a city level with no costs — the English Community Life Survey (CLS) and Dutch data of the International Social Survey Programme (ISSP).



V. Analysis of official data

Since suitable official data was only available and accessible for England and the Netherlands, the following steps focus on those countries.


For the English case, we used data from the Community Life Survey (CLS), conducted by the Cabinet Office from 2012–2013 (U.K. Cabinet Office, 2013). The CLS includes information on citizen participation in the section “Influencing political decisions and local affairs”. The microdata was available from the U.K. Data Service and registration to their services was required to download data.

Our research focused on the most recent results from the CLS, the 2014–2015 wave (U.K. Cabinet Office, 2015a). The study contained 2,022 observations from face-to-face interviews, and another 2,323 answers of a Web-based survey. For privacy reasons, the data released offers geo-related variables with different levels of aggregation, but not the exact city where respondents were located. Hence, identifying observations from a certain city became difficult and only possible when the combination of classifiers within a region was unique. For instance, Manchester was the only city classified as “Centres with Industry” at the ONS district level (subgroup) within the North West England region. More information on the ONS classifiers can be found in the S1 Appendix and at the U.K. Office for National Statistics (ONS) Web site (

To explore the potential influence of city size in this study, this analysis focuses on three cities of different sizes. Between cities that can be uniquely identified through the geo-related variables of the CLS2014–2015 survey, we focused on London, with 8,538,689 inhabitants (ONS, 2015), Birmingham, with a population of 1,101,360 (ONS, 2015), and Manchester, with 520,215 individuals (ONS, 2015). Table 2 illustrates the values of the variable “contacted a politician” obtained from CLS for year 2014–2015 in London, Birmingham, and Manchester.


Table 2: Contacts to politicians in London, Birmingham, and Manchester according to CLS 2014–2015 (U.K. Cabinet Office, 2015a).
Note: (*) Weighted.
(ONS, 2015)
CLS2014–2015 observations (*)CLS2014–2015 contacts to politicians (*)Inferred contacts/city
London8,538,689668.590.6 (13.5%)1,156,992.0
Birmingham1,101,360119.87.3 (6.1%)67,183.0
Manchester520,21523.91.5 (6.1%)31,941.2


The Netherlands

Information on “contacts to politicians” for the Dutch case was obtained from the International Social Survey Programme (ISSP). The ISSP is a continuing annual programme of cross-national collaboration on surveys covering topics important for social science research. The surveys in 2004 and 2014 were focused specifically on active citizenship. Information for 2014 was available from national organizations in charge; microdata accessible through the Data Archiving and Networked Services (DANS) Web site (

The Dutch ISSP studies for 2013 ( (‘National Identity III’) and 2014 (‘Citizenship II’) (Ganzeboom, 2014) contained 1,638 observations. Similar to the previous Spanish and English cases, our analysis focused on three cities of different sizes. The selected cities were Amsterdam, with 837,155 inhabitants (Centraal Bureau voor de Statistiek (CBS), 2016); Utrecht, with a population of 339,852 (CBS, 2016); and, ’s-Hertogenbosch, with 151,781 individuals (CBS, 2016). The values of the variable “contacted a politician” from the Dutch studies for Amsterdam, Utrecht, and ’s-Hertogenbosch are shown in Table 3.


Table 3: “Contacts to politicians” in Amsterdam, Utrecht, and ’s-Hertogenbosch according to Dutch data (Ganzeboom, 2014).
Note: (*) Weighted.
(Feb 2016)
ISSP observations (*)ISSP contacts to politicians (*)Inferred contacts/city
Amsterdam837,15585.46.9 (8.0 %)67,307.3




VI. Analysis of Twitter data

The workflow to obtain information on active citizenship — based on Twitter — and with results for the Dutch and English cases is shown in Figure 2.


Workflow of Twitter data
Figure 2: Workflow of Twitter data processing to obtain “contacts to politicians” in the Netherlands (Amsterdam, Utrecht, and ’s-Hertogenbosch) and England (London, Birmingham, and Manchester).
Note: Larger version of figure available here.


The first step was the collection of tweets containing “mentions” to municipal representatives in the selected cities of the Netherlands and England. Mainly we considered the mayor and councillors of each city, while in London we also recognized London Assembly members and leaders of different boroughs. The full list of municipal representatives considered and information on their Twitter profiles is detailed in S2 Appendix, while Table 4 summarizes the number of representatives by city included in this analysis. During the three months of data collection (from 19 November 2015 to 26 February 2016) a total of 30,858 tweets were collected related to the Dutch case and another 271,199 tweets to the English case.


Table 4: Summary of tweet collection for Dutch and English cases.
Notes: (*) Number of encountered Twitter profiles /number of representatives considered; (**) 4 Mayor of London related, 24/26 London assembly related, 60/64 related to the 32 boroughs.
InhabitantsAmsterdam: 837,155
Utrecht: 339,852
’s-Hertogenbosch: 151,781
Total: 1,328,788
London: 8,538,689
Birmingham: 1,101,360
Manchester: 520,215
Total: 10,160,264
Municipal representatives consideredAmsterdam: 9/11
Utrecht: 7/8
’s-Hertogenbosch: 6/7
Total: 22/26
London: 88/94 (**)
Birmingham: 4/6
Manchester: 8/12
Total: 100/112
Number of tweets collected during data collection30,858271,199


The next step was to categorize Twitter users, to consider only those tweets generated by individual citizens. We analyzed the content of Twitter user descriptions, looking for the presence of a selection of keywords that allowed a distinction between “citizens” and “others”. The S3 Appendix describes the details considered during this process. As a result, just 50 percent of Dutch users and 74 percent of English users were identified as “citizens”. The accuracy of this filtering process was tested with a randomly selected set of users (about one percent of the total sample), revealing 74 percent accuracy in the Dutch case and 85 percent for the English case.

The next step was the geolocation of those users identified as individual citizens. The ratio of users including latitude and longitude information in their metadata was noticeably low (0.15 percent in England and 0.34 percent in the Netherlands), following a general trend for Twitter (Twitter Developer, n.d.). To increase the number of Twitter profiles that could be considered in the analysis, the method included a geocoding of “tweet location” and “profile location” given in tweet metadata.

The tweet location field was automatically generated when a Twitter user activated an option to display a place name, usually the name of a neighbourhood, city, or region. These toponyms follow standard formats, easing the geocoding process. In contrast, the user location attribute is a free-form character field in which information is not necessarily a location or parseable, making geocoding difficult. The geocoding service OpenStreetMap ( was selected in our previous research, because it offered geolocation results with high accuracy. In the case of England, 95.4 percent of tweets presenting tweet locations and 83.9 percent of the ones with profile locations were geolocated. Results in the Dutch case were along the same lines, with 98.7 percent and 86 percent of tweets geolocated, respectively. Finally, in both cases the process allowed geolocation of over 60 percent of Twitter users categorized as citizens.



VII. Assessment

Table 5 shows the values of the indicator “contacts to politicians to express opinion” in selected cities from England, the Netherlands, and Spain (on the results from a previous exercise, see Rosales Sánchez, et al., 2017) for more details) based on official data and Twitter information. The values from the use of Twitter were a linear extrapolation from three months of data collection to a one-year period, allowing an easier comparison of results obtained from surveys.


Contacts to politicians in selected cities of England, the Netherlands, and Spain
Table 5: “Contacts to politicians” in selected cities of England, the Netherlands, and Spain, obtained from official sources (ONS, 2015; IPPS, 2014; CIS, 2015), and from Twitter.
Note: Larger version of table available here.


Survey data for selected cities revealed that in the English cases the ratio of citizens that contact their politicians was higher than in the Dutch and Spanish cases. London, with 13.5 percent, presented the highest ratio of citizens contacting their politicians, while in the Dutch capital of Amsterdam the proportion was still high, but not in Madrid, where it was around two percent. Birmingham and Manchester presented high values, which might be related to traditions in English culture and politics, where the engagement of citizens and stakeholders has long been considered in policy-making. However, the analysis of the remaining cities considered in the study was hindered by a low number of observations from official surveys, a main disadvantage for official sources.

On the other hand, values obtained with Twitter were much lower than results from surveys. In London and Birmingham ratios are around 15 times lower than the surveys. Along the same lines, in Amsterdam the values obtained from Twitter were much lower. However, in Spain this was not the case, where Twitter offered higher values than surveys. In this last case, Twitter data collection took place during political elections in Spain which probably influenced the overall activity of Twitter users.

The method proposed to secure values for the indicator “contacted a politician to express an opinion” based on Twitter was evaluated to assess its replicability. This evaluation also allowed an identification of the main issues considered when using this methodology for transnational studies. The assessment criteria considered accuracy, scope and coverage, spatial dimension, temporal dimension, timeliness, coherence, cost, and effort.


The results from official data were obtained by using methods statistically designed to obtain a representation of the target universe. Specifications on sampling method, data processing, and standard errors are usually offered in technical reports (such as U.K. Cabinet Office, 2015a; ISSP, 2014) for the cases analyzed in this research. Meanwhile Twitter data presented several biases to consider. Twitter users are self-selected and not representative of a larger population. On the other hand, the results of the query to filter Twitter rarely exceeded one percent of the Twitter real-time stream in any of the cases, so Twitter data collection was not influenced by limitations of the public Twitter API [1].

Scope and coverage

The coverage of survey data in the English and Dutch cases was defined by a target universe, i.e., Dutch and English populations, respectively, over 18 years of age. On the other hand, the Twitter universe might not be representative of the whole population. Demographics of Twitter users in our data collection were unknown. However, Eurostat data offered a general idea of the use of the Internet, the different purposes for its use, and the kinds of technologies utilized. As shown in Table 6, in 2015, 83 percent of households in EU–28 had Internet access at home, and half of the population used the Internet to participate in social networks, from which the most active users were young people (86 percent of citizens between 16–24 years old), followed by individuals between 25–54 years old (58 percent rate), with the less active users on social networks being those between 55–74 years old (21 percent rate). Lastly, both Dutch and British citizens aged between 16 and 54 were more likely to use social networks to post opinions on civic or political issues than those over 54. This information offered a sense of what could be the potential profile of Twitter users in our sample.


 Indicators on the information society for the EU-28
Table 6: Indicators for the information society for the EU–28, United Kingdom, the Netherlands, and Spain, according to Eurostat (2017).
Note: Larger version of table available here.


Spatial dimension (spatial coverage and granularity)

Microdata from surveys are useful for obtaining indications at a local level for cities of a medium or large size. In small cities, the sample size is insufficient to be representative of the population. However, Twitter offers accurate geo-information when tweets are generated from devices equipped with GNSS.

In our study, most of the geolocated tweets were obtained by geoparsing spatial information in tweet metadata (tweet and profile locations). Thus, due to the capacities of geoparsers used, tweets were mainly geolocated at a city level.

Temporal dimension (period of data collection, exceptions during the collection and granularity)

Data collection from the surveys lasted several months, six months for the Dutch ISSP (April–September 2014) and 10 months for the Community Life Survey (July 2014 — April 2015). Survey results offered a perspective of the last year, asking respondents if they contacted a politician in the last 12 months. On the contrary, using the public Twitter Stream API, contacts to politicians could only be monitored in real time. It was not possible to obtain a measurement of who contacted a politician during the last year. The only possible data collection from the past was limited to approximately the past seven days, based on the limitations of the public Twitter Search API for tweet acquisition.

Concerning data collection, real-time data from Twitter could be downloaded at any time, for a period defined for the purposes of analyses.

Timeliness (any delay between data collection and data availability and frequency of data collection)

The English Community Life Survey (CLS) is held annually, obtaining information on issues related to social action and community empowerment. The data and main findings are published after a few months. In the case of the ISSP, there were only two modules dedicated to citizenship, in 2004 and 2014. The publication of data depends on the country participating. In the Dutch case, the results were published within a few months.

By using Twitter as a data source for a given indicator, replication could be done at any time. Once the method was settled for a specific city, results were obtained very quickly, depending on the number of tweets that needed to be geolocated.

Coherence (data comparability with other sources of information and consistency over time)

Periodically programmed data collection from surveys allows long-time series analysis. Data is comparable across the data collected, and with previous releases. The CLS results allowed yearly comparisons within the U.K., while cross-national comparisons were possible from ISSP data.

On the other hand, the method proposed to use Twitter ensured data comparability over time and space. However, the consistency of results was unknown due to unclear provisions of tweets from the streaming API.

Cost and effort (amount of work and time required to define, develop, apply, and reproduce the method)

Surveys costs were not documented, but considering the labour of the statisticians, the surveyors, data management, and publication, it could be assumed to be expensive. Nonetheless, ‘contacted a politician’ was but one of the questions included in the questionnaire, thus overall costs could be divided across all of the questions posed.

The method based on Twitter implied a substantial effort in design and development of the process, while its reproduction appeared nearly costless. Discovery of Twitter user profiles of politicians selected was time-consuming, depending on the popularity of a given politician and the ways in which users described themselves. For instance, most of the Dutch politicians of interest to this research followed a clear pattern of description, specifying their roles and affiliations to particular parties. This particular issue allowed a much faster recognition of users of interest.

To sum up, assessment demonstrated that the proposed methodology was replicable in different countries, offering results at a local level. According to the assessment, one of the main issues to consider for transnational studies was the coverage of Twitter information, in the sense that the level of Twitter use depends on the country to be analysed, where the demographic profiles of users could change, as shown in Table 6. Also relevant was the coherence and timeliness of survey data, that allowed temporal comparisons, depending on the periodicity of a given survey in each country. We found that modifications might be necessary depending on the targeted country, mainly due to the way Twitter users described their profiles in the network, creating in turn different levels of difficulty towards user discovery and categorization.



VIII. Discussion and conclusions

This research analyzed the potential of social media to obtain social indicators by investigating differences between three countries by applying a method proposed in Rosales Sánchez, et al. (2017). The robustness of the method to use Twitter as a source for the indicator ‘contacting a politician’ was replicated in English and Dutch cases. The different steps included in the method — data collection, user filtering, and user geolocation — were followed with no particular modification. The results yielded by the method, however, diverged from survey values more than expected. Using Twitter, the numbers of contacts to politicians obtained were much lower than those from surveys in the cases of London, Birmingham, Manchester, and Amsterdam. Several circumstances might have influenced the results. One of the main differences between English, Dutch, and Spanish cases (Rosales Sánchez, et al., 2017) was the political situation under which data collection took place. In the Spanish case, data collection was carried out during a pre-election period. Activity on Twitter is highly correlated to particular events; thus the electoral period in the Spanish case may have raised the number of tweets directed specifically to politicians.

Cultural differences may have also been a factor that influenced results. We found significant differences in the way that Twitter users from different countries provided their bios. Being a politician in the Netherlands, for instance, was usually clearly recognized from user descriptions, which did not happen in the Spanish case. This aspect may have influenced filtering process of users, leading to a higher recognition of political party members in the Netherlands. Another cultural effect might have been related to the differences in citizen engagement across different countries. In the U.K. there is a long history of civic participation, and an electoral system “first-past-the-post” that favours direct links between the electorate and their representatives. Hence Twitter might represent just one of many possible ways to communicate with local representatives.

Apart from cultural differences, language seemed to be a relevant factor, that is language comprehension on the part of researchers in analyzing data. Twitter user categorization was based on metadata content and the accuracy of results was quite different for different countries, with high accuracy in the Spanish case, less accuracy in the English case, and the lowest accuracy for the Dutch case. The level of accuracy corresponded with the lingustic knowledge of the research team. This might have led to an over-filtering of users recognized as non-citizens, potentially decreasing the number of the contacts to politicians in the selected cities for research.

Furthermore, Twitter users are unlikely to be representative of an entire diverse population; Twitter was not the only channel used for contacting politicians. In addition, to compare survey data with Twitter information, the three months of Twitter data collection were extrapolated to one year, which might not be representative. Moreover, the results obtained might vary, depending on the proportion of political representatives from selected cities who maintained a Twitter account. In addition, the geolocation process is based on information given by Twitter users with an unknown level of accuracy.

Considering all those aspects, we conclude that our methodology, as established in (Rosales Sánchez, et al., 2017), was replicable, allowing a measurement of active citizenship in different countries. However, the accuracy of the results varied, depending on how difficult it was to identify types of Twitter users (politician, media, citizen, or others), and on how familiar researchers were with a given language. Taking into account the results of this study, we deduce that Twitter data is not a source that can replace data obtained from traditional surveys. In contrast to all of the positive assumptions about the potentials of Twitter, the results obtained when using Twitter data are far from providing an accurate measure of an indicator that could replace measurements obtained by survey data. In line with the results from our previous research, this methodology provided results at a city level. As such, it might complement data from national surveys, casting new light on the evolution of indicators over time and offering the possibility to analyze relative differences across urban areas of interest, even in cities of smaller sizes where national surveys are not always representative. Twitter data also presents an opportunity to detect political events of particular interest to citizens, offering a measure of their relative impacts on society, as shown for the case of Spain where the influence of an election clearly was visible. Further research could reveal if adaptations of this methodology could offer more timely and local indications of other social indicators. In addition, considering that the wealth of digital traces will continue to increase over time, this methodology establishes an early view of online data processing to better understand some societal behaviors. End of article


About the author

Cristina Rosales Sánchez is a Research Fellow at the European Commission, DG Joint Research Centre (JRC), and Ph.D. candidate at the Laboratory of Geo-Information Science and Remote Sensing ( at Wageningen University & Research. Her research interests include spatial analysis, data mining and quality of life in urban areas.
E-mail: rosales [dot] cris [at] gmail [dot] com



This research has been possible due to the contribution of Massimo Craglia, Arend Ligtenberg and Arnold Bregt in the conceptualization and supervision of the research, and in the writing review part of the process.



1. About filter limitations on the Twitter streaming API, see



Australian Bureau of Statistics (ABS), n.d. “ABS Data Quality Framework,” at, accessed 6 July 2016.

M. Agafiţei, F. Gras, W. Kloek, F. Reis, and S. Vhataju, 2015. “Measuring output quality for multisource statistics in official statistics: Some directions,” Statistical Journal of the IAOS, volume 31, number 2, pp. 203–211.
doi:, accessed 21 May 2018.

A. Bruns and Y. Liang, 2012. “Tools and methods for capturing Twitter data during natural disasters,” First Monday, volume 17, number 4, at, accessed 24 May 2018.
doi:, accessed 24 May 2018.

Centraal Bureau voor de Statistiek (CBS), 2016. “Bevolkingsontwikkeling; regio per maand [Population growth regions per month],” at, accessed 24 May 2018.

R. Costanza, M. Hart, J. Talberth, and S. Posner, 2009. “Beyond GDP: The need for new measures of progress,” Pardee Center for the Study of the Longer-Range Future, paper number 4, at, accessed 7 August 2014.

M. Craglia, F. Ostermann, and L. Spinsanti, 2012. “Digital Earth from vision to practice: Making sense of citizen-generated content,” International Journal of Digital Earth volume 5, number 5, pp. 398–416.
doi:, accessed 24 May 2018.

A. Culotta, 2014. “Estimating county health statistics with Twitter,” CHI ’14: Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, pp. 1,335–1,344.
doi:, accessed 24 May 2018.

M. De Smedt, 2013. “Measuring subjective issues of well-being and quality of life in the European statistical system,” Social Indicators Research, volume 114, number 1, pp. 153–167.
doi:, accessed 24 May 2018.

S. Elwood , M. Goodchild, and D. Sui, 2012 “ Researching volunteered geographic information: Spatial data, geographic research, and new Social practice,” Annals of the Association of American Geographers, volume 102, number 3, pp. 571–590.
doi:, accessed 24 May 2018.

European Commission. Eurostat. Collaboration in Research and Methodology for Official Statistics (CROS), 2013. “Scheveningen Memorandum on Big Data and Official Statistics,” at, accessed 24 May 2018.

European Statistical System, 2011. “Well-being and sustainable development: Final report,”, accessed 24 May 2018.

Eurostat, 2017. “Final report of the expert group on quality of life indicators” (28 March), at, accessed 24 May 2018.

I. Feinerer, K. Hornik, and D. Meyer, 2008 Text mining infrastructure in R, Journal of Statistical Software, volume 25, number 5, at, accessed 24 May 2018.
doi:, accessed 24 May 2018.

H. Ganzeboom, 2014. “ISSP-NL 2013 & 2014: Citizenship and national identity,” at, accessed 24 May 2018.
doi:, accessed 24 May 2018.

F. Girardin, F. Calabrese, F. Fiore, C. Ratti, and J. Blat, 2008. “Digital footprinting: Uncovering tourists with user-generated content,” IEEE Pervasive Computing, volume 7, number 4, pp. 36–43.
doi:, accessed 24 May 2018.

M. Goodchild and J. Glennon, 2010. “Crowdsourcing geographic information for disaster response: A research frontier,” International Journal of Digital Earth volume 3, number 3, pp. 231–241.
doi:, accessed 24 May 2018.

T. Highfield and T. Leaver, 2015. “A methodology for mapping Instagram hashtags,” First Monday, volume 20, number 1, at, accessed 24 May 2018.
doi:, accessed 24 May 2018.

B. Hoskins and M. Mascherini, 2009. “Measuring active citizenship through the development of a composite indicator,” Social Indicators Research, volume 90, number 3, pp. 459–488.
doi:, accessed 24 May 2018.

International Social Survey Programme (ISSP), 2014. “ISSP 2013 — ‘National identity III’ — ZA number 5950,” at, accessed 24 May 2018.
doi:, accessed 24 May 2018; “IISSP 2014 — ‘Citizenship II’ — number 6670” at, accessed 24 May 2018.
doi:, accessed 24 May 2018.

M. Kamel Boulos, B. Resch, D. Crowley, J. Breslin, G. Sohn, R. Burtner, W. Pike, E. Jezierski, and K.-Y. Slayer Chuang, 2011. “Crowdsourcing, citizen sensing and sensor Web technologies for public and environmental health surveillance and crisis management: Trends, OGC standards and application examples,” International Journal of Health Geographics, volume 10, at, accessed 24 May 2018.
doi:, accessed 24 May 2018.

T. Lalor, 2014 “How big Is big data? Exploring the role of big data in official statistics,” United Nations Economic Commission for Europe (UNECE) (version 0.1, March), at, accessed 24 May 2018.

K. Leetaru, S. Wang, G. Cao, A. Padmanabhan, and E. Shook, 2013. “Mapping the global Twitter heartbeat: The geography of Twitter,” First Monday, volume 18, number 5, at, accessed 24 May 2018.
doi:, accessed 24 May 2018.

D. Lüdecke, 2015a. “sjmisc: Miscellaneous data management tools,” at, accessed 24 May 2018.

D. Lüdecke, 2015b. “sjPlot: Data visualization for statistics in social science,” at, accessed 24 May 2018.

Organisation for Economic Co-operation and Development (OECD), 2013. “OECD guidelines on measuring subjective well-being” (20 March), at, accessed 27 February 2014.

F. Reis, P. Ferreira, and V. Perduca, 2014. “The use of Web activity evidence to increase the timeliness of official statistics indicators” (12 November), at, accessed 24 May 2018.

S. Roche, E. Propeck-Zimmermann, and B. Mericskay, 2011. “GeoWeb and crisis management: Issues and perspectives of volunteered geographic information,” GeoJournal, volume 78, number 1, pp. 21–40.
doi:, accessed 24 May 2018.

C. Rosales Sánchez, M. Craglia, and A. Bregt, 2017 “New data sources for social indicators: the case study of contacting politicians by Twitter,” International Journal of Digital Earth volume 10, number 8, pp. 829–845.
doi:, accessed 24 May 2018.

J. Stiglitz, A. Sen, and J.-P. Fitoussi, 2009. “Report by the Commission on the Measurement of Economic Performance and Social Progress,” at, accessed 24 May 2018.

P. Struijs, B. Braaksma, and P. Daas, 2014. “Official statistics and big data,” Big Data & Society (1 April).
doi:, accessed 24 May 2018.

S.-M. Tam and F. Clarke, 2015. “Big data, official statistics and some initiatives by the Australian Bureau of Statistics,” International Statistical Review, volume 83, number 3, pp. 436–448.
doi:, accessed 24 May 2018.

Twitter Developer, n.d. “Tutorials: Tweet geospatial metadata,” at, accessed 24 May 2018.

U.K. Cabinet Office, 2015a. “Community Life Survey, 2014–2015: Special licence access,” at, accessed 24 May 2018.
doi:, accessed 24 May 2018.

U.K. Cabinet Office, 2015b. “Community Life Survey 2014 to 2015: Technical report” (23 October), at, accessed 24 May 2018.

U.K. Cabinet Office, 2013. “Community Life Survey 2012 to 2013 findings” (19 July), at, accessed 24 May 2018.

U.K. Office for National Statistics (ONS), 2015. “Population estimates for UK, England and Wales, Scotland and Northern Ireland,” at, accessed 24 May 2018.

United Nations. Department of Economic and Social Affairs. Statistics Division (UNStats), n.d. “Big data for official statistics,” at, accessed 24 May 2018.

United Nations Economic Commission for Europe (UNECE), n.d. “High-Level Group for the Modernisation of Official Statistics,” at, accessed 24 May 2018.

H. Wickham 2015. “stringr: Simple, consistent wrappers for common string operations,” at, accessed 24 May 2018.

H. Wickham, 2011. “The split-apply-combine strategy for data analysis,” Journal of Statistical Software, volume 40, number 1, at, accessed 24 May 2018.
doi:, accessed 24 May 2018.

H. Wickham and E. Miller, 2015. “haven: Import SPSS, Stata and SAS files,” at, accessed 24 May 2018.


Appendix 1: Excerpt from the data dictionary of the “Community Life Survey, 2014–2015”.

Pos. = 106 Variable = CivParta Variable label = In the last 12 months have you done any of the following: Contacted a local official such as local councillor, MP etc.
This variable is numeric, the SPSS measurement level is NOMINAL
SPSS user missing values = -1.0 thru -9.0 and -8.0
Value label information for CivParta
Value = -8.0 Label = Dont know
Value = 0.0 Label = No
Value = -9.0 Label = Refused
Value = -1.0 Label = Item not applicable
Value = 1.0 Label = Yes

Pos. = 591 Variable = GOR Variable label = Region (former Government Office Region)
This variable is numeric, the SPSS measurement level is NOMINAL
Value label information for GOR
Value = 1.0 Label = North East
Value = 2.0 Label = North West
Value = 3.0 Label = Yorkshire & Humberside
Value = 4.0 Label = East Midlands
Value = 5.0 Label = West Midlands
Value = 6.0 Label = East of England
Value = 7.0 Label = London
Value = 8.0 Label = South East
Value = 9.0 Label = South West

Pos. = 570 Variable = Wrdsupgp Variable label = ONS Ward Classification : Supergroup (2001 Wards)
This variable is numeric, the SPSS measurement level is NOMINAL
Value label information for Wrdsupgp
Value = 1.0 Label = Industrial Hinterlands
Value = 2.0 Label = Traditional Manufacturing
Value = 3.0 Label = Built-up Areas
Value = 4.0 Label = Prospering Metropolitan
Value = 5.0 Label = Student Communities
Value = 6.0 Label = Multicultural Metropolitan
Value = 7.0 Label = Suburbs and Small Towns
Value = 8.0 Label = Coastal and Countryside
Value = 9.0 Label = Accessible Countryside

Pos. = 571 Variable = Wrdgrp Variable label = ONS Ward Classification : Group (2001 Wards)
Value label information for Wrdgrp
Value = 7.1 Label = Suburbs
Value = 2.4 Label = Transitional Economies
Value = 3.5 Label = Built-up Areas
Value = 4.6 Label = Prospering Metropolitan
Value = 8.13 Label = Countryside
Value = 7.12 Label = Commuter Suburbs
Value = 1.2 Label = Out of Town Housing
Value = 8.14 Label = Senior Communities
Value = 7.11 Label = Prospering Suburbs
Value = 6.9 Label = Inner City Multicultural
Value = 8.15 Label = Out of Town Manufacturing
Value = 1.1 Label = Industrial Areas
Value = 9.17 Label = Accessible Countryside
Value = 6.8 Label = Multicultural Areas
Value = 2.3 Label = Built-up Manufacturing
Value = 5.7 Label = Student Communities

Pos. = 572 Variable = Wrdsubgpg Variable label = ONS Ward Classification : Subgroup (2001 Wards)
Value label information for Wrdsubgpg
Value = 8.13.21 Label = Countryside A
Value = 7.10.16 Label = Suburbs A
Value = 8.13.22 Label = Countryside B
Value = 4.6.10 Label = Prospering Metropolitan B
Value = 6.9.15 Label = Inner City Multicultural
Value = 7.10.17 Label = Suburbs B
Value = 9.17.26 Label = Accessible Countryside
Value = 3.5.7 Label = Built-up Areas A
Value = 2.3.4 Label = Built-up Manufacturing
Value = 1.2.3 Label = Out of Town Housing
Value = 5.7.11 Label = Student Communities A
Value = 6.8.14 Label = Multicultural Areas
Value = 7.11.18 Label = Prospering Suburbs
Value = 10 Label = Other
Value = 2.4.6 Label = Transitional Economies B
Value = 5.7.12 Label = Student Communities B
Value = 7.12.19 Label = Commuter Suburbs A
Value = 8.15.24 Label = Out of Town Manufacturing
Value = 7.12.20 Label = Commuter Suburbs B
Value = 1.1.1 Label = Industrial Areas A
Value = 4.6.9 Label = Prospering Metropolitan A
Value = 5.7.13 Label = Student Communities C
Value = 1.1.2 Label = Industrial Areas B
Value = 2.4.5 Label = Transitional Economies A
Value = 3.5.8 Label = Built-up Areas B
Value = 8.14.23 Label = Senior Communities

Pos. = 573 Variable = ladsupgp Variable label = ONS District Level Classification: Supergroup (2003)
This variable is numeric, the SPSS measurement level is NOMINAL
Value label information for ladsupgp
Value = 1.0 Label = Cities and Services
Value = 2.0 Label = London Suburbs
Value = 3.0 Label = London Centre
Value = 4.0 Label = London Cosmopolitan
Value = 5.0 Label = Prospering UK
Value = 6.0 Label = Coastal and Countryside
Value = 7.0 Label = Mining and Manufacturing

Pos. = 574 Variable = ladgrp Variable label = ONS District Level Classification: Group (2003)
Value label information for ladgrp
Value = 1.2 Label = Centres with Industry
Value = 4.6 Label = London Cosmopolitan
Value = 5.8 Label = New and Growing Towns
Value = 7.12 Label = Manufacturing Towns
Value = 1.3 Label = Thriving London Periphery
Value = 6.10 Label = Coastal and Countryside
Value = 2.4 Label = London Suburbs
Value = 5.7 Label = Prospering Smaller Towns
Value = 3.5 Label = London Centre
Value = 1.1 Label = Regional Centres
Value = 5.9 Label = Prospering Southern England
Value = 7.11 Label = Industrial Hinterlands

Pos. = 575 Variable = Ladsubgpg Variable label = ONS District Level Classification: Subgroup (2003)
Value label information for Ladsubgpg
Value = 4.6.10 Label = London Cosmopolitan - A
Value = 4.6.11 Label = London Cosmopolitan - B
Value = 1.2.3 Label = Centres with Industry - B
Value = 5.7.14 Label = Prospering Smaller Towns - C
Value = 6.10.18 Label = Coastal and Countryside - B
Value = 6.10.17 Label = Coastal and Countryside - A
Value = 1.2.2 Label = Centres with Industry - A
Value = 2.4.6 Label = London Suburbs - A
Value = 5.7.12 Label = Prospering Smaller Towns - A
Value = 5.8.15 Label = New and Growing Towns - A
Value = 1.3.4 Label = Thriving London Periphery - A
Value = 2.4.7 Label = London Suburbs - B
Value = 7.11.20 Label = Industrial Hinterlands - A
Value = 7.11.21 Label = Industrial Hinterlands - B
Value = 1.1.1 Label = Regional Centres - A
Value = 7.12.22 Label = Manufacturing Towns - A
Value = 1.3.5 Label = Thriving London Periphery - B
Value = 5.7.13 Label = Prospering Smaller Towns - B
Value = 10 Label = Other
Value = 3.5.8 Label = London Centre - A
Value = 5.9.16 Label = Prospering Southern England - A
Value = 3.5.9 Label = London Centre - B

Pos. = 576 Variable = oa_sup11 Variable label = Output Area Classification - Supergroup (8 categories) (2011 Census definition)
This variable is numeric, the SPSS measurement level is NOMINAL
Value label information for oa_sup11
Value = 1.0 Label = Rural residents
Value = 2.0 Label = Cosmopolitans
Value = 3.0 Label = Ethnicity central
Value = 4.0 Label = Multicultural metropolitans
Value = 5.0 Label = Urbanites
Value = 6.0 Label = Suburbanites
Value = 7.0 Label = Constrained city dwellers
Value = 8.0 Label = Hard-pressed living

Pos. = 577 Variable = oa_grp11 Variable label = Output Area Classification - Group (26 categories) (2011 Census definition)
This variable is numeric, the SPSS measurement level is NOMINAL
Value label information for oa_grp11
Value = 1.0 Label = 1a – Farming communities
Value = 2.0 Label = 1b – Rural tenants
Value = 3.0 Label = 1c – Ageing rural dwellers
Value = 4.0 Label = 2a – Students around campus
Value = 5.0 Label = 2b – Inner city students
Value = 6.0 Label = 2c – Comfortable cosmopolitan
Value = 7.0 Label = 2d – Aspiring and affluent
Value = 8.0 Label = 3a – Ethnic family life
Value = 9.0 Label = 3b - Endeavouring Ethnic Mix
Value = 10.0 Label = 3c – Ethnic dynamics
Value = 11.0 Label = 3d – Aspirational techies
Value = 12.0 Label = 4a – Rented family living
Value = 13.0 Label = 4b – Challenged Asian terraces
Value = 14.0 Label = 4c – Asian traits
Value = 15.0 Label = 5a – Urban professionals and families
Value = 16.0 Label = 5b – Ageing urban living
Value = 17.0 Label = 6a – Suburban achievers
Value = 18.0 Label = 6b – Semi-detached suburbia
Value = 19.0 Label = 7a – Challenged diversity
Value = 20.0 Label = 7b – Constrained flat dwellers
Value = 21.0 Label = 7c – White communities
Value = 22.0 Label = 7d – Ageing city dwellers
Value = 23.0 Label = 8a – Industrious communities
Value = 24.0 Label = 8b – Challenged terraced workers
Value = 25.0 Label = 8c – Hard pressed ageing workers
Value = 26.0 Label = 8d – Migration and churn

Pos. = 578 Variable = oa_sub11 Variable label = Output Area Classification - Subgroup (76 categories) (2011 Census definition)
This variable is numeric, the SPSS measurement level is NOMINAL
Value label information for oa_sub11
Value = 1.0 Label = 1a1 – Rural workers and families
Value = 2.0 Label = 1a2 – Established farming communities
Value = 3.0 Label = 1a3 – Agricultural communities
Value = 4.0 Label = 1a4 – Older farming communities
Value = 5.0 Label = 1b1 – Rural life
Value = 6.0 Label = 1b2 – Rural white-collar workers
Value = 7.0 Label = 1b3 – Ageing rural flat tenants
Value = 8.0 Label = 1c1 – Rural employment and retirees
Value = 9.0 Label = 1c2 – Renting rural retirement
Value = 10.0 Label = 1c3 – Detached rural retirement
Value = 11.0 Label = 2a1 – Student communal living
Value = 12.0 Label = 2a2 – Student digs
Value = 13.0 Label = 2a3 – Students and professionals
Value = 14.0 Label = 2b1 – Students and commuters
Value = 15.0 Label = 2b2 – Multicultural student neighbourhood
Value = 16.0 Label = 2c1 – Migrant families
Value = 17.0 Label = 2c2 – Migrant commuters
Value = 18.0 Label = 2c3 – Professional service cosmopolitans
Value = 19.0 Label = 2d1 – Urban cultural mix
Value = 20.0 Label = 2d2 – Highly-qualified quaternary workers
Value = 21.0 Label = 2d3 – EU white-collar workers
Value = 22.0 Label = 3a1 – Established renting families
Value = 23.0 Label = 3a2 – Young families and students
Value = 24.0 Label = 3b1 – Striving service workers
Value = 25.0 Label = 3b2 – Bangladeshi mixed employment
Value = 26.0 Label = 3b3 – Multi-ethnic professional service workers
Value = 27.0 Label = 3c1 – Constrained neighbourhoods
Value = 28.0 Label = 3c2 – Constrained commuters
Value = 29.0 Label = 3d1 – New EU tech workers
Value = 30.0 Label = 3d2 – Established tech workers
Value = 31.0 Label = 3d3 – Old EU tech workers
Value = 32.0 Label = 4a1 – Private renting young families
Value = 33.0 Label = 4a2 – Social renting new arrivals
Value = 34.0 Label = 4a3 – Commuters with young families
Value = 35.0 Label = 4b1 – Asian terraces and flats
Value = 36.0 Label = 4b2 – Pakistani communities
Value = 37.0 Label = 4c1 – Achieving minorities
Value = 38.0 Label = 4c2 – Multicultural new arrivals
Value = 39.0 Label = 4c3 – Inner city ethnic mix
Value = 40.0 Label = 5a1 – White professionals
Value = 41.0 Label = 5a2 – Multi-ethnic professionals with families
Value = 42.0 Label = 5a3 – Families in terraces and flats
Value = 43.0 Label = 5b1 – Delayed retirement
Value = 44.0 Label = 5b2 – Communal retirement
Value = 45.0 Label = 5b3 – Self-sufficient retirement
Value = 46.0 Label = 6a1 – Indian tech achievers
Value = 47.0 Label = 6a2 – Comfortable suburbia
Value = 48.0 Label = 6a3 – Detached retirement living
Value = 49.0 Label = 6a4– Ageing in suburbia
Value = 50.0 Label = 6b1 – Multi-ethnic suburbia
Value = 51.0 Label = 6b2 – White suburban communities
Value = 52.0 Label = 6b3 – Semi-detached ageing
Value = 53.0 Label = 6b4 – Older workers and retirement
Value = 54.0 Label = 7a1 – Transitional Eastern European neighbourhood
Value = 55.0 Label = 7a2 – Hampered aspiration
Value = 56.0 Label = 7a3 – Multi-ethnic hardship
Value = 57.0 Label = 7b1 – Eastern European communities
Value = 58.0 Label = 7b2 – Deprived neighbourhoods
Value = 59.0 Label = 7b3 – Endeavouring flat dwellers
Value = 60.0 Label = 7c1 – Challenged transitionaries
Value = 61.0 Label = 7c2 – Constrained young families
Value = 62.0 Label = 7c3 – Outer city hardship
Value = 63.0 Label = 7d1 – Ageing communities and families
Value = 64.0 Label = 7d2 – Retired independent city dwellers
Value = 65.0 Label = 7d3 – Retired communal city dwellers
Value = 66.0 Label = 7d4 – Retired city hardship
Value = 67.0 Label = 8a1 – Industrious transitions
Value = 68.0 Label = 8a2 – Industrious hardship
Value = 69.0 Label = 8b1 – Deprived blue-collar terraces
Value = 70.0 Label = 8b2 – Hard pressed rented terraces
Value = 71.0 Label = 8c1 – Ageing industrious workers
Value = 72.0 Label = 8c2 – Ageing rural industry workers
Value = 73.0 Label = 8c3 – Renting hard-pressed workers
Value = 74.0 Label = 8d1 – Young hard-pressed families
Value = 75.0 Label = 8d2 – Hard-pressed ethnic mix
Value = 76.0 Label = 8d3 – Hard-Pressed European Settlers


Appendix 2: List of Twitter profiles from selected politicians in the Netherlands and England.


The Netherlands
NameTwitter profileComments
City councilors
Kajsa Ollongren@KajsaOllongren
wethouder economische zaken, zee- & luchthaven, kunst en cultuur #D66 Amsterdam
Laurens Ivens@LaurensIvens
Wethouder Bouwen, Wonen en Dierenwelzijn in Amsterdam, lid van de SP. Amsterdam
Abdeluheb Choho@AbdeluhebChoho
Wethouder Duurzaamheid (D66), Amsterdam
Arjan van Gils@ArjanvanGils
Gemeentesecretaris/Algemeen Directeur gemeente Amsterdam
No tweets, not active? But recent picture
Eric van der Burg@ericvanderburg
Wethouder Zorg, Sport, Ouderen, Ruimtelijke Ordening en grondzaken. Vader van @martvanderburg en Daan van der Burg. #vvd #nsgk #ntr Amsterdam Noord
Simone Kukenheim@simonekukenheim
Wethouder Amsterdam - Onderwijs - Jeugd - Diversiteit - inburgering - volwasseneneducatie #D66 Amsterdam
Pieter Litjens@PieterLitjens
man van Saskia, vader van Joep, Teun en Neeltje. Wethouder Verkeer, Vervoer en Organisatie gemeente Amsterdam
Udo Kock No Twitter account
Eberhard van der Laan No Twitter account
Official Twitter channel of the municipality of Amsterdam
Gemeente Amsterdam@AmsterdamNL
Het officiële Twitterkanaal van de gemeente Amsterdam. Volg ons en blijf op de hoogte. Gebruik voor contact, klachten etc: Amsterdam
It is conceived for informing citizens. The official channel for contacts and complains ( ) is offered in the Twitter profile description. However, it receives large amount of tweets from citizens asking for answers about diverse themes.
City councilors
Lot van Hooijdonk@lotvanhooijdonk
wethouder verkeer&mobiliteit, duurzaamheid en milieu in Utrecht
Victor Everhardt (D66) No Twitter account
Jeroen Kreijkamp@JeroenKreijkamp
Wethouder Financien, Economische Zaken, Onderwijs, internationale zaken, Citymarketing en wijken Oost en Overvecht in Utrecht voor D66
Margriet Jongerius@jongerius_mfm
Wethouder Welzijn, Zorg en maatschappelijke ondersteuning, Wijkgericht werken en participatie, Cultuur en wijk Noordwest van@GemeenteUtrecht voor GroenLinks
Kees Geldof@KeesGeldofVVD
Wethouder gemeente Utrecht
Paulus Jansen@PaulusJansenSP
wethouder Utrecht (wonen, ruimtelijke ordening, sport, dierenwelzijn, vastgoed; vanaf mei 2014); woordvoerder wonen-energie-water-RO SP Tweede Kamer (2006–2014)
Utrecht-Lunetten ...
Jan van Zanen@janvanzanen030
Dit wordt misschien het officiele account van de nieuwe Burgemeester van Utrecht Jan van Zanen
Not so reliable?
Official Twitter channel of the municipality of Utrecht
Gemeente Utrecht@GemeenteUtrecht
Het officiële Twitteraccount van de gemeente Utrecht. Persberichten | Crisiscommunicatie | Stel je vraag | Antwoorden ma t/m vr tussen 9 en 17 uur
City councilors
Jan Hoskam@JanHoskam
wethouder VVD / gemeente 's-Hertogenbosch / financiën / economische zaken / duurzaamheid / afvastoffendienst
huib van olden@huibvanolden
CDA #werkwerkwerk en inkomen, aanpak armoede, cultuur, toerisme en erfgoed, wethouder ’-Hertogenbosch, CAO-onderhandelaar getrouwd en 3 kinderen
Eric Logister@ericlogister
Wethouder Onderwijs, Stedelijke Transformatie, Grondbeleid, Wonen en Brabant Stad
Jos van Son@JosvanSon
Wethouder in ’s-Hertogenbosch namens Rosmalens Belang met in portefeuille o.a. Wijken en Dorpen, Verkeer en Vervoer, Sport, Water en Groen
Rosmalen, The Netherlands
Paul Kagie@PaulKagie
Wethouder ’s-Hertogenbosch Zorg en Welzijn
Mr. Dr. A.G.J.M. (Ton) Rombouts
Burgemeester (CDA)
T (073) 615 51 55 | E
 No Twitter account
Official Twitter channel of the municipality
Het officiële twitteraccount van de gemeente ’s-Hertogenbosch
Lord Mayor
Paul Murphy  
Deputy Lord Mayor
Carl Austin@CarlAustinBehanCarl Austin-Behan
Deputy Lord Mayor of Manchester, Ward Councillor for Burnage.
Manchester, England
Leader of the Council
Sir Richard Leese@SirRichardLeeseSir Richard Leese
Sir Richard Leese, Labour Politician. Leader of Manchester City Council , Councillor for Crumpsall and Manchester City fan —
Statutory Deputy Leader of the Council
Sue Murphy@smurph99Sue Murphy
Labour Councillor Brooklands. Deputy Leader Manchester City Council. Chair of Governors, The Manchester College. Novus Board. Man City fan. Personal capacity.
Deputy Leader
Bernard Priest  
Executive Member for Adult Health and Wellbeing
Paul Andrews@cllr_pPaul Andrews
#Labour Councillor for Baguely and Newall Green. Executive member for Adult Social Care. My views are my own.
Executive Member for Children’s Services
Sheila Newman  
Executive Member for Culture & Leisure
Rosa Battle@rosa_battleRosa Battle
Manchester Labour Councillor for Beswick and Openshaw, Executive Member for Culture and Leisure, Lead Member for Schools
Executive Member for Environment
Kate Chappell@ChappellKateKate Chappell @ChappellKate
Exec Member for Environment at Manchester City Council and Labour Councillor for Rusholme. Cyclist, greenie etc. Not very good at twitter.
Executive Member for Finance & Human Resources
John Flanagan  
Executive Member for Neighbourhood Services
Nigel Murphy@CllrNigelNigel Murphy
Labour Councillor, listening and working for residents in Hulme Manchester. Executive Member for Neighborhoods@ManCityCouncil
Hulme, Manchester
Verified account
Official Twitter page of Manchester City Council. Monitored by the local authority’s Communications Team. We’ll try our best to find answers to your questions.
Manchester, U.K.
official Twitter feed from the Council
 Bham City Council
Birmingham City Council (UK), providing services for a million people. For service enquiries tweet @BCC_Help. Profile pic by @ross_jukes
Birmingham, UK
Joined November 2011
Lord Mayor
 Raymond Hassall
Leader of the Council
@johnclancyJohn Clancy
Leader, Birmingham City Council, Labour Councillor for Quinton Birmingham
Joined April 2008
Deputy Leader of the Council
@Cllr_IanWardIan Ward
Deputy Labour Leader of Bham City Council SHARD END ward
Just 5 tweets
Conservative Group
Leader of Group
 Robert AldenNo tw
Liberal Democrat Group
Leader of Group
@1oldcodgerpaul tilsley
Active LibDem Cllr, with very dry humour which can mean trouble! Semi detached Villa supporter & VP Moseley RFC and up the Bears.
Joined February 2012
No pic
Weird user-name — FAKE?
Chief Executive
@MKMRogersMark Rogers
First and foremost, me — Mark Rogers. Also, Chief Executive at Birmingham City Council. And SOLACE President 2013–16. Light relief? Vinyl — LP, 12, 10 & 7.
West Midlands
Joined March 2009
Lord Mayor
Boris JohnsonBoris Johnson
Verified account
This is the official Twitter account for the Mayor of London, Boris Johnson. See all of City Hall’s official social media accounts:
Deputy Mayor of London
Roger EvansRoger Evans
Deputy Mayor of London — Official tweets at @DepMayorLondon
London Gov
 London Gov
Updates on the work of the Mayor of London and staff at City Hall. See all our official social media accounts:
City Hall, London
London Assembly
 London Assembly
The London Assembly holds Mayor Boris Johnson to account and investigates issues that matter to Londoners
London Assembly Members
Tony Arbour  
Jennette ArnoldJennette Arnold AM
Labour & Co-op London Assembly Member working for Hackney, Islington & Waltham Forest and Deputy Chair of the London Assembly
London, City Hall
Gareth Bacon  
Kemi BadenochKemi Badenoch
London Assembly Member, GLA Conservative Spokesman for Policing & Crime, South London Area Tories, Spectator Digital stuff, Twitter-Sceptic, Wife and Mummy
London, England
John BiggsJohn Biggs
I am member of the London Assembly for City and East. I represent about 650,000 people across Barking & Dagenham, the City of London, Newham & Tower Hamlets
London, UK
(seems old)
Mayor John Biggs
Labour Mayor of Tower Hamlets
Tower Hamlets, London
Jennette ArnoldJennette Arnold AM
Labour & Co-op London Assembly Member working for Hackney, Islington & Waltham Forest and Deputy Chair of the London Assembly
London, City Hall
Andrew BoffAndrew Boff
Candidate for Mayor. GLA Conservative member of the London Assembly. We do this:
Barking Reach ...
James CleverlyJames Cleverly
Verified account
Conservative Member of Parliament for Braintree. London Assembly Member for Bexley & Bromley
Braintree, Bexley & Bromley
Tom CopleyTom Copley
London Assembly Member.@CityHallLabour Housing spokesperson and Chair of the Housing Committee. Democratic socialist. Trustee of@bhahumanists @newdiorama
Andrew DismoreAndrew Dismore
Verified account
Labour London Assembly Member for Barnet and Camden.
Barnet and Camden, London
Len DuvallLen Duvall
London Assembly Member for Greenwich & Lewisham and Leader of the Labour group on the London Assembly
City Hall
Roger Evans@RogerEvansAM
Nicky GavronNicky Gavron AM
Former Deputy Mayor of London. London Assembly Member, Chair of Planning Committee, and GLA Labour Spokesperson for Planning and Housing Supply. NPF Member.
Darren JohnsonDarren Johnson AM
London Assembly Member — Green Party, Chair of Environment Committee, Brockley resident. Tweeting about music and politics. Music blog:
London SE4 ...
Jenny JonesJenny Jones
Verified account
Green Party Assembly Member. Also a peer. Concerned esp about climate change & civil liberties. Labelled by Met Police as a domestic extremist.
London ...
Jenny Jones
Stephen KnightStephen Knight AM
Lib Dem London Assembly Member and Councillor for Teddington Ward, London Borough of Richmond upon Thames. Left-wing liberal.
Teddington, England
Kit MalthouseKit Malthouse
Member of Parliament for North West Hampshire. London Assembly Member. I don't use twitter to communicate. Please email via website contact form.
Andover & Westminster
Joanne McCartneyJoanne McCartney
London Assembly Member for Enfield and Haringey, Chair of the Assembly’s Police & Crime Committee, and member of the Budget & Performance Committee.
Enfield/Haringey, London, UK
Steve O’ConnellSteve O’Connell
GLA Member for Croydon and Sutton Kenley Ward Councillor Lifelong Crystal Palace FC fan lover of Animals and Real Ale.
Caroline PidgeonCaroline Pidgeon
Lib Dem Leader on the London Assembly. Lib Dem London Mayoral Candidate 2016.
Murad QureshiMurad Qureshi
Mancunian by birth, bred in London. W9er living in NW1. London Assembly Member. RTs certainly not endorsement but note of interest.
Dr Onkar SahotaOnkar Sahota
Chair London Assembly Health Committee, Labour Assembly Member for Ealing & Hillingdon. Chair Ealing Save NHS. NHS GP. Views are personal
London , UK
Navin ShahNavin Shah
Labour Assembly Member for Brent and Harrow. On the Planning Committee and the Fire Authority
Brent, Harrow and City Hall
Valerie Shawcross Valerie Shawcross
I’m the London Assembly Member for Lambeth and Southwark. I spend my time fighting for better public Transport in London.
Richard TraceyRichard Tracey
Wandsworth, London
(Different profile style than the rest, but political content and picture recognized)
Fiona Twycross Fiona Twycross
Labour London Assembly Member. Spokesperson on Fire, Economy and Welfare (and Food). Contact me on
London boroughs
City of London
(City of London Corporation)
 City of London
Verified account
Follow us for news, events and more. Monitored during office hours 8am–6pm Monday to Friday.
Guildhall, London
Town Clerk and Chief Executive, City of London
(Key City officers)
John BarradellJohn Barradell
Town Clerk & CEO of the City of London Corporation. NPR, Machin, social history, cycling and family fan. Views are mine etc.
City of Westminster
 Westminster Council
Verified account
Tweeting news, views and events from Westminster City Council. Monitored Mon–Fri, 9am–6pm. Out of hours call 020 7641 6000.
Westminster, London
Joined April 2009
Leader of the CouncilPhilippa Roe
Verified account
Conservative leader of Westminster City Council.
London, England
Low number of tweets
New? Date not available
Royal Borough of Kensington and Chelsea
Verified account
Official Twitter feed for the Royal Borough of Kensington and Chelsea
Kensington and Chelsea, London
Joined February 2009
Leader of the Royal Borough of Kensington and Chelsea.Nicholas Paget-Brown
Blogger, but not Tw
Hammersmith and Fulham
Official Twitter feed from Hammersmith & Fulham CouncilH&F Council
Official Twitter feed from Hammersmith & Fulham Council in west London (monitored Mon–Fri, 9am–5.30pm). News, events and consultations in your borough.
London, UK
Leader of the CouncilStephen Cowan
Leader, the London Borough of Hammersmith & Fulham. Labour councillor for Hammersmith Broadway. Resident needing help? Please email:
Joined March 2009
official Twitter feed from the CouncilWandsworth Council
The official Wandsworth Council Twitter feed. News, events, jobs & more. Monitored during office hours. Report flytips etc online at
Wandsworth, London
Joined December 2008
Leader of the Council
Ravi Govindia
official Twitter feed from the CouncilLambeth Council
Verified account
The official twitter account of Lambeth Council, the London Borough of Lambeth. We monitor this account Monday to Friday, 9am to 5pm. ...
Lambeth, London
Joined March 2009
Leader of Lambeth Council.Lib Peck
Leader of Lambeth Council. Local councillor for Thornton ward. All views are my own.
Joined January 2012
Leader of the Council
Ravi Govindia
official Twitter feed from the CouncilSouthwark Council
News, information and events from Southwark Council. We monitor tweets Monday - Friday, 9am—5pm.
Southwark, London
Joined December 2008
LeaderPeter John
Leader of Southwark Council and Labour Councillor for South Camberwell. Executive Member for Children, Young People, Employment & Skills — London Councils
Joined February 2009
Tower Hamlets
official Twitter feed from the CouncilTower Hamlets
Get the latest news and service updates from Tower Hamlets Council. For service requests visit our website or call 020 7364 5020.
London, E14 2BG
Joined March 2010
Leaderabdul mukit
Low number of tweets
official Twitter feed from the CouncilHackney Council
News, events and activities from Hackney Council. Monitored during office hours only. For service enquiries go to
Joined October 2010
Mayor / Political leader of the council
* distinctly in this borough
no tw
official Twitter feed from the CouncilIslington Council
Follow us for council news and updates. This account is monitored 9am–5pm Mon–lFri. For customer services please call 020 7527 2000.
Islington, London
Joined February 2009
LeaderRichard Watts
I’m @islingtonlabour Leader of Islington Council and co-founder of @FSM_4_all
Islington, London.
Joined March 2009
official Twitter feed from the CouncilCamden Council
Verified account
Follow us for news and updates. We monitor this account during office hours, Monday to Friday. For service enquiries visit our website
London Borough of Camden
Joined January 2009
LeaderSarah Hayward
Labour leader of Camden. Makes a difference, got in to politics to change things. Tweets littered with predictive text errors. Usual caveats re RTs etc.
Camden, London
Joined March 2009
official Twitter feed from the CouncilBrent Council
Hi! We’re Brent Council. Follow us for news, events, jobs and more. We’re usually here Mon–Fri, 9am to 5pm. Check out our house rules
London, UK
Joined March 2009
LeaderCllr. Muhammed Butt
Official account of Labour Leader of @Brent_Council. Passionate about empowering local communities, fairness for all & non-spicy curries ...
London Borough of Brent
Joined June 2008
official Twitter feed from the CouncilEalingCouncil
Verified account
News, events and jobs from Ealing Council. Tweets Mon- Fri 9–5
Ealing, West London.
Joined November 2008
LeaderJulian Bell
Labour Leader of @EalingCouncil, Chair of @LondonCouncils Transport & Environment Ctte, Researcher to Virendra Sharma MP, Sheff Utd fan & keen cyclist!
Joined March 2009
official Twitter feed from the CouncilHounslow
News and info from the London Borough of Hounslow
Hounslow, west London
Joined September 2010
LeaderCouncillor Steve Curran
Richmond upon Thames
official Twitter feed from the CouncilRichmond Council
Verified account
News and tips from the council and community | Support: @LBRuT_Help | Keep up to date with our latest news:
Richmond upon Thames
Joined October 2009
LeaderLord True.
Kingston upon Thames
official Twitter feed from the CouncilKingston Council
Official Twitter for The Royal Borough of Kingston upon Thames. Follow us for the news, events and other information from the council.
Kingston upon Thames, UK
Joined January 2012
 Contact Kingston
Official Twitter feed for the Royal Borough of Kingston Council Customer Service Team. We’re here to offer assistance and updates from 9am until 5pm, Mon — Fri
Kingston upon Thames
Joined October 2012
LeaderKevin Davis
Leader of Royal Borough of Kingston upon Thames, Musician, Husband, Father
Kingston upon Thames
Joined June 2007
official Twitter feed from the CouncilMerton Council
The official account for Merton Council. To report a problem, for example fly-tipping or a missed bin collection please visit
Joined August 2009
LeaderStephen Alambritis
Labour Politician. Former Business Leader. Now Leader Merton Council. Member @EU_CoR. Fulham Fan. LSE. Cypriot. FCIPR. Email:
England. London. Merton.
Joined August 2012
official Twitter feed from the CouncilSutton Council
Verified account
Open green fair smart. We are the London Borough of Sutton. Fbook: . News: . Social T&Cs:
Sutton, London
Joined March 2009
LeaderRuth Dombey
Lib Dem campaigner, councillor Sutton North ward, Leader Sutton Council, lover of all things Italian (except Italian politics!)
Joined April 2009
official Twitter feed from the CouncilYour Croydon
Verified account
News and events from Croydon Council. This account is monitored Mon to Fri, 9am to 5pm. For enquiries please visit
Croydon, South London
Joined March 2009
LeaderTony Newman
Leader of Croydon Council, AFC Croydon Athletic #Ambitious4Croydon, LGA Housing, RSA, Digital Champion, White Ribbon Ambassador
Joined February 2011
official Twitter feed from the CouncilBromley Council
News and events from Bromley Council. Tweets office hours.
Joined July 2009
Tweet to Bromley
LeaderStephen Carr
NO tw
official Twitter feed from the CouncilLewisham Council
News, events, service updates and alerts. While we monitor during office hours, please don’t use this in an emergency. Report issues via
Greater London, London
Joined October 2008
Executive mayorSteve Bullock
Directly Elected Mayor of Lewisham. Please email casework to
London Borough of Lewisham
Joined April 2009
official Twitter feed from the CouncilRoyal Greenwich
Verified account
Greenwich Council. News, services, events, offers. Account mainly monitored in office hours. T: 020 8854 8888 Email:
Greenwich, London, UK
Joined July 2009
LeaderCllr Denise Hyland
Leader of Royal Borough of Greenwich. Abbey Wood Cllr. Out of office hours call 020 8854 8888. E: All views — my own
Royal Borough of Greenwich
official Twitter feed from the CouncilLB Bexley
London Borough of Bexley’s official Twitter account. Monitored Mon–Fri 9am–5pm. Please make service requests via or 020 8303 7777
Bexleyheath, Kent DA6 7AT
Joined March 2009
LeaderTeresa O’Neill
NO tw
official Twitter feed from the CouncilHavering Council
News, info and updates from Havering Council’s official twitter feed.
Joined July 2011
Leaderroger ramsey
NO tw
Barking and Dagenham
official Twitter feed from the CouncilBarking and Dagenham
Barking and Dagenham is an outer London Borough to the east of the City on the north bank of the River Thames and within the M25 London Orbital Motorway.
Joined August 2011
LeaderDarren Rodwell
Barking and Dagenham Councillor, Campign Organiser for Barking CLP, Passionate about the community and keeping it at forefront of Labour policy
Barking and Dagenham
Joined January 2013
Cllr Darren Rodwell
official Twitter feed from the CouncilRedbridge Council
Verified account
Official Twitter feed for Redbridge Council. This account is only staffed Mon-Fri 9am-5.30pm so we may not respond outside of these times
London Borough of Redbridge
Joined February 2009
LeaderJas Athwal
Leader of Redbridge Council, Mayfield ward Councillor.
Joined March 2011
official Twitter feed from the CouncilNewham London
News and service updates from #Newham Council. Monitored Mon — Fri, 9am — 6pm. For service requests use
London Borough of Newham
Joined August 2011
Executive mayorSir Robin Wales
Mayor of Newham
London Borough of Newham ...
Joined January 2013
Waltham Forest
official Twitter feed from the CouncilWalthamForestCouncil
News, events and service updates from Waltham Forest Council. Monitored Monday-Friday, 9am–5pm. Contact us at
Waltham Forest, London, UK
Joined August 2008
LeaderChris Robbins
Leader of Waltham Forest Council.
Waltham Forest, London
Joined April 2009
official Twitter feed from the CouncilHaringey Council
Verified account
Official council account featuring local news, events and updates. To report service problems, send us feedback or make a complaint please use our website.
Haringey, London
Joined January 2011
LeaderClaire Kober
Leader of @HaringeyCouncil. Deputy Chair of @LondonCouncils. Chair of @LGAcomms Resources Board. Kids, cooking, tennis plus an occassional run. All view mine.
Joined June 2012
official Twitter feed from the CouncilEnfield Council
Enfield Council’s Twitter feed. Monitored from 9am — 5pm Mon to Fri. Please report any issues at
Enfield, London, UK
Joined April 2009
LeaderCllr Doug Taylor
Leader of Enfield Council
official Twitter feed from the CouncilBarnet Council
Verified account
The official Twitter account of The London Borough of Barnet
Barnet, London, N11 1NP
Joined May 2008
LeaderTweet to Barnet
Richard Cornelius
Leader of the London Borough of Barnet since 2011. Conservative Councillor for the Totteridge Ward. Elected member on Barnet Council since 2006.
official Twitter feed from the CouncilBorough of Harrow
Tweeting news & events from the London Borough of Harrow via Harrow Council. You can also email us on
Joined July 2009
Leadercllr david perry
official Twitter feed from the CouncilHillingdon Council
Verified account
News & info for the London Borough of Hillingdon. Tweets Mon–Fri 9–5 by Caroline, Charlotte (CS), Hana, Pauline & Linda
Joined June 2008
LeaderCouncillor Raymond Puddifoot MBE
no tw



Appendix 3: Specifications of the process for the categorization of Twitter users.

➢ Selection of keywords from the most common words describing Twitter profiles with high number of followers:
"actueel", "belangrijkst", "bestuurders", "bnr", "columnist", "d66", "dagelijk", "gemeent", "heldenmagazine", "joopnl", "journalist", "kanaal", "krant", "laatst", "mail", "media", "nieuw", "nieuws", "offici", "officiël", "radio", "rtl", "twitteraccount", "uur"

➢ Self-selection from Twitter accounts analysis:
"politie", "wijkagenten", "alderperson", "mp", "minister", "d66", "sp", "vvd", "chairman", "pvda", "lid tweede kamer", "woordvoerder", "parlement", "parliament", "kamerlid", "fractievoorzitter", "verslaggever"

➢ Keywords obtained from the Spanish case study (Dutch version)
"informatie","dagelijks","communicatie","actualiteiten","directeur","radio","plannen","kanaal" "president", "krant", "plaatsvervanger", "politieke", "televisie", "website", "drukpers", "secretaresse", "leider","officiële","journalistiek","kandidaat"

➢ Keywords related to the politicians or municipal representatives that have been contacted:
"@kajsaollongren", "@laurensivens", "@abdeluhebchoho", "@arjanvangils", "@ericvanderburg", "@simonekukenheim", "@pieterlitjens","@amsterdamnl", "@lotvanhooijdonk", "@jeroenkreijkamp", "@jongerius_mfm", "@keesgeldofvvd", "@paulusjansensp", "@janvanzanen030", "@janvanzanen", "@gemeenteutrecht", "@janhoskam", "@huibvanolden", "@ericlogister", "@josvanson", "@paulkagie", "@shertogenbosch"

Features for the classification of Twitter users

➢ Feature 1 → Providing User Description (Y/N) → N indicates citizen["f.user_desc"]<-ifelse($user_desc.proc=="", "0", "1")

➢ Feature 2 → Ocurrence of media/politics/business related terms → ">0" indicates non-citizen
keywords.pol ← c ("news", "radio", " media ", "press", "journalist", "newspaper", " magazine", "broadcasting", "channel \\d", "[0-9]\\s?[ap]m", "actueel", "belangrijkst", "bestuurders", "bnr", "columnist", "d66", "dagelijk", "gemeent", "heldenmagazine", "joopnl", "journalist", "kanaal", "krant", "laatst", "mail", "media", "nieuw", "nieuws", "offici", "officiël", "rtl", "twitteraccount", "uur", "politie", "wijkagenten", "alderperson", "mp", "minister", "d66", "sp", "vvd", "chairman", "pvda", "lid tweede kamer", "woordvoerder", "parlement", "parliament", "kamerlid", "fractievoorzitter", "verslaggever", "informatie","dagelijks","communicatie","actualiteiten", "directeur","radio", "plannen", "kanaal", "president", "krant", "plaatsvervanger", "politieke", "televisie", "website", "drukpers", "secretaresse", "leider", "officiële", "journalistiek", "kandidaat")$f.pol ← str_count($user_desc.proc, paste(keywords.pol,collapse="|"))

➢ Feature 3 → Ocurrence of terms related to the targeted politicians/representatives → ">0" indicates non-citizen
keywords.repres ← c("@kajsaollongren", "@laurensivens", "@abdeluhebchoho", "@arjanvangils", "@ericvanderburg", "@simonekukenheim", "@pieterlitjens","@amsterdamnl", "@lotvanhooijdonk", "@jeroenkreijkamp", "@jongerius_mfm", "@keesgeldofvvd", "@paulusjansensp", "@janvanzanen030", "@janvanzanen", "@gemeenteutrecht", "@janhoskam", "@huibvanolden", "@ericlogister", "@josvanson", "@paulkagie", "@shertogenbosch")$f.repres ← str_count($user_desc.proc, paste(keywords.repres,collapse="|"))

➢ Feature 4 → Ocurrence of terms related to personal opinion → ">0" indicates citizen
keywords.self ← c("my own", "personal", "views are mine", "mijn eigen opvattingen", "persoonlijke")
noself ← c("persoonlijke train", "personal train", "personal coach")$f.self ← str_count($user_desc.proc, paste(keywords.self,collapse="|"))$f.noself←-str_count($user_desc.proc, paste(noself,collapse="|"))$f.self ←$f.self -$f.noself

➢ Feature 5 → Firms/Organizations related
keywords.firms ← c("we ", "join us", "leader", "leading", "firm", "brand", " us", "our ", "Wij", "bij ons", "leider", "leidt", "merk", "ons", "onze")$f.firms ← str_count($user_desc.proc, paste(keywords.firms,collapse="|"))


Editorial history

Received 6 February 2018; accepted 27 May 2018.

Creative Commons License
“International comparison of active citizenship by using Twitter data, the case of England and the Netherlands” by Cristina Rosales ánchez is licensed under a Creative Commons Attribution 4.0 International License.

International comparison of active citizenship by using Twitter data, the case of England and the Netherlands
by Cristina Rosales Sánchez.
First Monday, Volume 23, Number 6 - 4 June 2018

A Great Cities Initiative of the University of Illinois at Chicago University Library.

© First Monday, 1995-2019. ISSN 1396-0466.