Periodic Report Summary 3 - SHARE_M4 (Multinational Advancement of Research Infrastructures on Ageing)
Project Context and Objectives:
SHARE, the Survey of Health, Ageing and Retirement in Europe, has been created in response to demands by the European Commission, the Council and the European Parliament to address the challenges of population ageing and its social and economic challenges. SHARE provides an infrastructure of micro data that combines information on health with the economic and social living conditions of individuals as they age and as they are exposed to the societal changes precipitated by the population ageing process. SHARE is unique and innovative for three reasons: First, the survey is ex-ante harmonized across all 20 participating countries which allows comparing the effects of the different health and welfare systems in the European countries on individuals and families. Second, SHARE is multi-disciplinary and fills an important research vacuum, namely the interaction between health and socio-economic factors. Third, SHARE is longitudinal, i.e. the same individuals are repeatedly being interviewed to understand their individual aging processes and their responses to on-going social and political changes. SHARE provides a high European added value since it is a supra-national survey which draws its scientific significance from comparing social, economic and health developments across countries.
The main challenge in managing SHARE is to keep a well-balanced and comparable survey in all countries involved and to create a harmonized data base. For that reason, the centrifugal forces due to country and disciplinary differences need to be controlled. This was the main aim of the SHARE-M4 project. It included all tasks that are essential to maintain the European added value of SHARE, especially to keep the national surveys well integrated. It complemented the national financing mode of data collection as result of the ESFRI process by financing the (a) central design of the questionnaire, especially the development of new modules, (b) central data base management, dissemination, and archiving, (c) central training for interviewers, user outreach, training, and feedback, (d) software development, and (e) coordination and communication.
Project Results:
We have accomplished keeping up our excellence in service provision as well as in science by the following activities:
(a) Central design of the questionnaire, especially the development of new modules:
The reporting period covers the second preparatory stage of wave 6 in which the conceptual revisions of the questionnaire were put to the test in the pilot study in February 2014 and the pretest in June 2014. A number of innovations were introduced during the reporting period. A new case in point for SHARE’s leadership in state-of-the-art interdisciplinary panel construction was bringing back the Social Networks module and adapting it so that panel respondents will be asked to compare their network in wave 6 with the one they reported in the fourth wave. This is the first cross-national, harmonized large-scale, population-based survey study to measure changes in respondents' social networks over time.
According to our principles of module rotation, we followed through with substituting our measure of lower body strength (the so-called Chair Stand module) which was conducted in wave 5 with the measurement of lung force (the so-called Peak Flow module). This is in line with best practices in panel surveys where modules are rotated to keep respondent burden in check in each wave.
Another innovation concerns the introduction of a short scale to assess the five dominant factors of personality, the so-called “Big Five”. We hope to enable our scientific data users to perform new analyses integrating measures of personality, a promising avenue in ageing research. Several nodes financed under the M4 proposal have done extensive work in further streamlining important longitudinal modules, such as modules on pensions and income. Several iterations have been conducted between the involved parties under leadership of Prof. Guglielmo Weber at Padua University and numerous changes and cuts were decided to reduce the burdensomeness of the modules and improve the conversational flow of the modules. To give but one example, we will be piloting a new automated coding of occupational categories based on the latest research by ISCO experts. In practice, respondents will state their occupation in their own words and an autocomplete mechanism will suggest a number of standardized occupational categories upon the interviewer’s entering of the first few letters, much like the google auto-complete function. The interviewer will then confirm with the respondent which suggested category makes for the best match. If successful in the pilot runs, this will be a very efficient way of coding occupational categories in congruence with an internationally harmonized standard.
After receiving translation verification of all 27 languages used in the fifth wave of SHARE we used this as input for the translation of the wave 6 questionnaire. Feedback from the country teams was very positive. The input provided by the translation verification served as one of the two independent translations that are stipulated by the TRAP-D process which is considered state-of-art in multinational and multi-cultural studies such as SHARE.
Further improvements were accomplished in devising a module for gathering Dried Blood Spots in wave 6. Country-specific adaptations had to be created to allow for different procedures in several countries, mostly due to legal requirements about the process. Several versions had been implemented and they have been shown to work quite well in the pretest.
Finally, we have increased our efforts in harmonizing the assessment of respondents’ willingness to have their survey data linked to administrative data of their countries. We had successfully done this in Germany with data of the German Pension Fund in waves 4 and 5 and extend this in wave 6 in other countries. As access rules and laws governing the usage of administrative data are highly heterogeneous between the European countries, it was a key challenge to create a unified survey instrument able to handle that heterogeneity.
(b) Central data base management, dissemination, and archiving:
To prepare the internal release 0 of wave 5 as well as the data collection of wave 6, data from all previous waves has been checked against the new wave 5 data. This includes consistency checks between waves and modules, e.g. correct linkage of panel respondents. At the same time the data base management team wrote programs for the first scientific release of wave 5 data. These programs are mainly designed to make the data more user friendly. They include for example reformatting of many variables to facilitate analyses, converting non-Euro currencies to Euro values, coding of open answers as well as labelling of variables and values. Information was collected to produce generated modules, e.g. with regional information or ISCED codes for education and last but not least the team worked on the documentation of the data.
In addition testcases files were prepared for the wave 6 pilot studies of SHARE countries. for the Pretest as well as for the Main Study so called preload data files were prepared for all countries. The preload data was also checked for inconsistencies with information from the all survey agencies to insure that interviewers address the right households. These preload files including information on respondents from previous waves are used to make the interview shorter and less repetitive for panel respondents. In addition refreshment samples from various countries were checked and prepared for the main study of wave 6.
Not only because interviews in SHARE are conducted in 25 languages, checking cleaning and dissemination of the data can only be done in close cooperation with the country teams. A lot of communication and coordination between the central data base management and the country teams, as well as special training, took place, e.g. at the SHARE operators meetings in March and July, to ensure the high quality standards and the unique ex-ante harmonisation of the SHARE data
(c) Central training for interviewers, user outreach, training, and feedback:
A prominent example of our success in user outreach and user feedback during the reporting period was the user conference in the fall of 2013 in Liège, Belgium: http://www.share-project.org/home0/news/article/share-user-conference.html. This conference, building on our experiences of earlier user conferences, brought together designers and users of SHARE. More than 100 international researchers from various disciplines presented and discussed their work. It provided a very fruitful opportunity for the scientific teams dealing with the revision of the actual questionnaire to learn from the scientific data users what was actually used. It is rewarding for all survey designers to see how the output of their efforts helped in advancing the understanding of the population ageing process. In addition, the release of a new training data set was celebrated during that meeting – and is expected to considerably push the use of SHARE data in university courses.
Another important accomplishment during the reporting period was holding the full circle of three intensive training sessions to prepare the fieldwork of the sixth wave (fieldwork start in January 2015) for national head interviewers, the people who later train the fieldwork interviewers that conduct the SHARE interviews. All technical, logistical and managerial aspects of successful fieldwork were conveyed and, even more importantly, trained with hands-on experience. SHARE puts great emphasis on multi-modal teaching methods that ensure that standardised best practices of interviewing (such as active listening and being prepared to deal with respondent reluctance) are being trained at the national level. In addition, we assessed the correct implementation in the SHARE countries with a standardised form and in some instances visited national training sessions to observe national trainings first-hand and compare the curriculum with the template provided through the central TTTs. All software innovations were trained, especially new protocols to update information on deceased households and difficult situations for the interviewer, such as nursing home interviews or refusals.
(d) Software development:
During the reporting period, we finalized all major (and minor) changes in all software tools. The online platform used to translate the questionnaire was completely revamped and hence renamed to “Translation Management Tool (TMT)”, formerly the “Language Management Utility”. New features of the TMT (inline editing of questions, multi-language views, full compatibility to manage multiple surveys simultaneously, a richer overview of the history of the questionnaire translations, advanced options to better document the translation process, a better local management dashboard, new exports and the option to implement more advanced translation workflows like TRAP-D) were put to the test during the translation period of wave 6 in 2014. Multiple iteration of translations were conducted by all country teams and we received valuable feedback on future improvements. Overall, the innovations in the translation tool were received very well by the SHARE researchers responsible for translating.
During the reporting period, the pilot and pretest periods served to enhance the sample management system (SMS) and the sample distributor (SD) before the actual start of fieldwork in January 2015. All modifications proved to facilitate the preparation and implementation of fieldwork. Among the major changes were the improvement of the name import and export functions which are important for panel care, the modification of the household state variable to obtain the most recent information about the current state of each household, and the extension of the SD technical export with a set of new variables.
A cutting edge innovation was the first fieldwork using a sequential mixed-mode software to interview a subsample of SHARE panel respondents. This fieldwork was conducted in the Czech Republic as this country was able to run the first in-between-waves data collection ever in SHARE. We provided the survey agency with a sample distributor that could seamlessly migrate households between survey modes (web-based self-completion, CATI interviewing and computer-assisted self-interviewing CASI). The actual questionnaire was based on the Day Reconstruction Method (DRM) Outcomes are being currently evaluated but first evidence suggest good success at administering the survey in different modes.
(e) Coordination and communication:
SHARE is in its essence a team-based effort, not only across scientific disciplines but also across professional boundaries and across different interests and goals of the groups involved. Thus, maintaining a culture of intensive cooperation remained a key challenge we had to deal with during the entire reporting period. This was accomplished by very tight communication between central coordination in Munich and the area coordinators on the one side, and the country teams on the other side. Most of the daily communication uses the Internet. Substantial personnel effort is devoted to the cooperation between the central coordination and the country teams. In addition, we had workshops and conferences about every two-three months.
We continued with the production of “Compliance Profiles” for the performance of fieldwork agencies in preparation, fieldwork and post-fieldwork delivery of contractually demanded deliverables. Like ion wave 4, we followed our proven and well-received quantified internal evaluation model which was based on quality standards outlined in the SHARE model contract. Again, we released this evaluation report on our internal website after presenting a first draft to the Scientific Monitoring Board.
Potential Impact:
The major steps were pilot and pretest in February and June 2014 and the fielding of wave 6 in January 2015 after the end of this project. The data of wave 6 will be unique as they contain information on the time use and well-being as well as dried blood spots from respondents from all participating SHARE countries.
List of Websites:
www.share-project.org/m4
SHARE, the Survey of Health, Ageing and Retirement in Europe, has been created in response to demands by the European Commission, the Council and the European Parliament to address the challenges of population ageing and its social and economic challenges. SHARE provides an infrastructure of micro data that combines information on health with the economic and social living conditions of individuals as they age and as they are exposed to the societal changes precipitated by the population ageing process. SHARE is unique and innovative for three reasons: First, the survey is ex-ante harmonized across all 20 participating countries which allows comparing the effects of the different health and welfare systems in the European countries on individuals and families. Second, SHARE is multi-disciplinary and fills an important research vacuum, namely the interaction between health and socio-economic factors. Third, SHARE is longitudinal, i.e. the same individuals are repeatedly being interviewed to understand their individual aging processes and their responses to on-going social and political changes. SHARE provides a high European added value since it is a supra-national survey which draws its scientific significance from comparing social, economic and health developments across countries.
The main challenge in managing SHARE is to keep a well-balanced and comparable survey in all countries involved and to create a harmonized data base. For that reason, the centrifugal forces due to country and disciplinary differences need to be controlled. This was the main aim of the SHARE-M4 project. It included all tasks that are essential to maintain the European added value of SHARE, especially to keep the national surveys well integrated. It complemented the national financing mode of data collection as result of the ESFRI process by financing the (a) central design of the questionnaire, especially the development of new modules, (b) central data base management, dissemination, and archiving, (c) central training for interviewers, user outreach, training, and feedback, (d) software development, and (e) coordination and communication.
Project Results:
We have accomplished keeping up our excellence in service provision as well as in science by the following activities:
(a) Central design of the questionnaire, especially the development of new modules:
The reporting period covers the second preparatory stage of wave 6 in which the conceptual revisions of the questionnaire were put to the test in the pilot study in February 2014 and the pretest in June 2014. A number of innovations were introduced during the reporting period. A new case in point for SHARE’s leadership in state-of-the-art interdisciplinary panel construction was bringing back the Social Networks module and adapting it so that panel respondents will be asked to compare their network in wave 6 with the one they reported in the fourth wave. This is the first cross-national, harmonized large-scale, population-based survey study to measure changes in respondents' social networks over time.
According to our principles of module rotation, we followed through with substituting our measure of lower body strength (the so-called Chair Stand module) which was conducted in wave 5 with the measurement of lung force (the so-called Peak Flow module). This is in line with best practices in panel surveys where modules are rotated to keep respondent burden in check in each wave.
Another innovation concerns the introduction of a short scale to assess the five dominant factors of personality, the so-called “Big Five”. We hope to enable our scientific data users to perform new analyses integrating measures of personality, a promising avenue in ageing research. Several nodes financed under the M4 proposal have done extensive work in further streamlining important longitudinal modules, such as modules on pensions and income. Several iterations have been conducted between the involved parties under leadership of Prof. Guglielmo Weber at Padua University and numerous changes and cuts were decided to reduce the burdensomeness of the modules and improve the conversational flow of the modules. To give but one example, we will be piloting a new automated coding of occupational categories based on the latest research by ISCO experts. In practice, respondents will state their occupation in their own words and an autocomplete mechanism will suggest a number of standardized occupational categories upon the interviewer’s entering of the first few letters, much like the google auto-complete function. The interviewer will then confirm with the respondent which suggested category makes for the best match. If successful in the pilot runs, this will be a very efficient way of coding occupational categories in congruence with an internationally harmonized standard.
After receiving translation verification of all 27 languages used in the fifth wave of SHARE we used this as input for the translation of the wave 6 questionnaire. Feedback from the country teams was very positive. The input provided by the translation verification served as one of the two independent translations that are stipulated by the TRAP-D process which is considered state-of-art in multinational and multi-cultural studies such as SHARE.
Further improvements were accomplished in devising a module for gathering Dried Blood Spots in wave 6. Country-specific adaptations had to be created to allow for different procedures in several countries, mostly due to legal requirements about the process. Several versions had been implemented and they have been shown to work quite well in the pretest.
Finally, we have increased our efforts in harmonizing the assessment of respondents’ willingness to have their survey data linked to administrative data of their countries. We had successfully done this in Germany with data of the German Pension Fund in waves 4 and 5 and extend this in wave 6 in other countries. As access rules and laws governing the usage of administrative data are highly heterogeneous between the European countries, it was a key challenge to create a unified survey instrument able to handle that heterogeneity.
(b) Central data base management, dissemination, and archiving:
To prepare the internal release 0 of wave 5 as well as the data collection of wave 6, data from all previous waves has been checked against the new wave 5 data. This includes consistency checks between waves and modules, e.g. correct linkage of panel respondents. At the same time the data base management team wrote programs for the first scientific release of wave 5 data. These programs are mainly designed to make the data more user friendly. They include for example reformatting of many variables to facilitate analyses, converting non-Euro currencies to Euro values, coding of open answers as well as labelling of variables and values. Information was collected to produce generated modules, e.g. with regional information or ISCED codes for education and last but not least the team worked on the documentation of the data.
In addition testcases files were prepared for the wave 6 pilot studies of SHARE countries. for the Pretest as well as for the Main Study so called preload data files were prepared for all countries. The preload data was also checked for inconsistencies with information from the all survey agencies to insure that interviewers address the right households. These preload files including information on respondents from previous waves are used to make the interview shorter and less repetitive for panel respondents. In addition refreshment samples from various countries were checked and prepared for the main study of wave 6.
Not only because interviews in SHARE are conducted in 25 languages, checking cleaning and dissemination of the data can only be done in close cooperation with the country teams. A lot of communication and coordination between the central data base management and the country teams, as well as special training, took place, e.g. at the SHARE operators meetings in March and July, to ensure the high quality standards and the unique ex-ante harmonisation of the SHARE data
(c) Central training for interviewers, user outreach, training, and feedback:
A prominent example of our success in user outreach and user feedback during the reporting period was the user conference in the fall of 2013 in Liège, Belgium: http://www.share-project.org/home0/news/article/share-user-conference.html. This conference, building on our experiences of earlier user conferences, brought together designers and users of SHARE. More than 100 international researchers from various disciplines presented and discussed their work. It provided a very fruitful opportunity for the scientific teams dealing with the revision of the actual questionnaire to learn from the scientific data users what was actually used. It is rewarding for all survey designers to see how the output of their efforts helped in advancing the understanding of the population ageing process. In addition, the release of a new training data set was celebrated during that meeting – and is expected to considerably push the use of SHARE data in university courses.
Another important accomplishment during the reporting period was holding the full circle of three intensive training sessions to prepare the fieldwork of the sixth wave (fieldwork start in January 2015) for national head interviewers, the people who later train the fieldwork interviewers that conduct the SHARE interviews. All technical, logistical and managerial aspects of successful fieldwork were conveyed and, even more importantly, trained with hands-on experience. SHARE puts great emphasis on multi-modal teaching methods that ensure that standardised best practices of interviewing (such as active listening and being prepared to deal with respondent reluctance) are being trained at the national level. In addition, we assessed the correct implementation in the SHARE countries with a standardised form and in some instances visited national training sessions to observe national trainings first-hand and compare the curriculum with the template provided through the central TTTs. All software innovations were trained, especially new protocols to update information on deceased households and difficult situations for the interviewer, such as nursing home interviews or refusals.
(d) Software development:
During the reporting period, we finalized all major (and minor) changes in all software tools. The online platform used to translate the questionnaire was completely revamped and hence renamed to “Translation Management Tool (TMT)”, formerly the “Language Management Utility”. New features of the TMT (inline editing of questions, multi-language views, full compatibility to manage multiple surveys simultaneously, a richer overview of the history of the questionnaire translations, advanced options to better document the translation process, a better local management dashboard, new exports and the option to implement more advanced translation workflows like TRAP-D) were put to the test during the translation period of wave 6 in 2014. Multiple iteration of translations were conducted by all country teams and we received valuable feedback on future improvements. Overall, the innovations in the translation tool were received very well by the SHARE researchers responsible for translating.
During the reporting period, the pilot and pretest periods served to enhance the sample management system (SMS) and the sample distributor (SD) before the actual start of fieldwork in January 2015. All modifications proved to facilitate the preparation and implementation of fieldwork. Among the major changes were the improvement of the name import and export functions which are important for panel care, the modification of the household state variable to obtain the most recent information about the current state of each household, and the extension of the SD technical export with a set of new variables.
A cutting edge innovation was the first fieldwork using a sequential mixed-mode software to interview a subsample of SHARE panel respondents. This fieldwork was conducted in the Czech Republic as this country was able to run the first in-between-waves data collection ever in SHARE. We provided the survey agency with a sample distributor that could seamlessly migrate households between survey modes (web-based self-completion, CATI interviewing and computer-assisted self-interviewing CASI). The actual questionnaire was based on the Day Reconstruction Method (DRM) Outcomes are being currently evaluated but first evidence suggest good success at administering the survey in different modes.
(e) Coordination and communication:
SHARE is in its essence a team-based effort, not only across scientific disciplines but also across professional boundaries and across different interests and goals of the groups involved. Thus, maintaining a culture of intensive cooperation remained a key challenge we had to deal with during the entire reporting period. This was accomplished by very tight communication between central coordination in Munich and the area coordinators on the one side, and the country teams on the other side. Most of the daily communication uses the Internet. Substantial personnel effort is devoted to the cooperation between the central coordination and the country teams. In addition, we had workshops and conferences about every two-three months.
We continued with the production of “Compliance Profiles” for the performance of fieldwork agencies in preparation, fieldwork and post-fieldwork delivery of contractually demanded deliverables. Like ion wave 4, we followed our proven and well-received quantified internal evaluation model which was based on quality standards outlined in the SHARE model contract. Again, we released this evaluation report on our internal website after presenting a first draft to the Scientific Monitoring Board.
Potential Impact:
The major steps were pilot and pretest in February and June 2014 and the fielding of wave 6 in January 2015 after the end of this project. The data of wave 6 will be unique as they contain information on the time use and well-being as well as dried blood spots from respondents from all participating SHARE countries.
List of Websites:
www.share-project.org/m4