Measuring and Coding Complex Items: (Semi-) Automated Solutions 1 |
|
| Chair | Dr Eric Harrison (City University London ) |
Personal social networks shape various aspects of peoples’ lives. Individuals can gain a range of benefits and support out of them. Given the significance of the social network construct for both science and policy, elaborate tools are required to measure them in social surveys. Yet, social networks and especially changes to them over time are not easy to determine. This presentation aims to introduce the approach developed by the Survey of Health, Ageing and Retirement (SHARE) to address this challenge.
SHARE developed a social networks (SN) module that does not only measure the current social network but also traces changes over time. The module applies a name generating mechanism in which respondents identify those people who are important to them and then add information on each person named. In addition, changes in respondents' social networks over time as well as the reasons for these changes are detected.
The SN module is part of the main SHARE questionnaire which participants complete through a computer-assisted personal interview (CAPI). The module starts with the interviewer asking the respondent to name up to six people with whom he or she most often discusses important things. Then, the respondent is given the possibility to mention one additional person who is important to him or her "for some other reason", resulting in a personal social network of up to seven confidants. The interviewer then asks for additional information about each of the individuals listed in the network roster, including relation to the respondent (e.g. spouse, sibling, friend, etc.), gender, residential proximity, frequency of contact, emotional closeness, year of birth, as well as occupational and relationship status. Finally, respondents are asked to rank their overall satisfaction with their social network.
In addition, the longitudinal version of the SN module allows for the linkage of currently collected social network members to those that were mentioned in the previous interview; After the identification of the current list of confidants, respondents are shown a list of their previous social network and are asked to link the people they just named to the network members of their previous interview. If a person that had been previously named was not mentioned again, respondents are asked why this person was not referred to again in the current interview.
Furthermore, the list of social network members is linked to other questionnaire modules, such as financial transfers (FT) or social support (SP). This allows for the distinction between exchanges of monetary or social support with members of the social network and with other persons.
In summary, the SN module generates data on the interpersonal environment of the survey respondents, both in its current state as well as in its changes over time. The longitudinal approach of the module enables researchers to gain insight into significant dynamics of social networks, such as changes in their compositions, specifics about the single network members, or information on newly added or lost confidants.
Educational variables are important socio-demographic variables which are covered in almost all surveys. Measuring education in a proper and all-embracing way includes measuring this construct on at least two dimensions. First, the stratification dimension including the hierarchical element of educational qualifications on the vertical axis, and second the dimension of the different fields of studies and training on the horizontal axis. Latter dimension, unfortunately, is often not included in surveys although it is a quite important variable. For example, the subject of academic studies or vocational training is often related to gender patterns. These patterns consequently lead to gender segregation in occupation and job market where we then identify a large gender wage gap. As another example, research identified discrepancies in values, attitudes, and beliefs which correlate with the subject of study.
For measuring the subjects of academic studies or vocational training several classifications exists: e.g., Fields of Education and Training by Eurostat, OECD’s Fields of Science and Technology, and some national documents as the Australian Standard Classification of Education. All of them are modified and adapted versions of the International Standard Classification of Education of 1997 (ISCED-97) by UNESCO. In 2013, the new version of ISCED fields of education and training classification (ISCED-F) was agreed on. This classification is the new standard for measuring and coding subjects of academic studies or vocational training. It also will be implemented in the survey tool on measuring key socio-demographic variables, developed within the SERISS project, as part of work package 8.
What is most important with regard to new standard classification of ISCED-F, is its independence from the other ISCED classifications on educational attainment (ISCED-A) and education programmes (ISCED-P). Therefore each field of education and training can be matched to several ISCED levels of both ISCED-A and ISCED-P. This claims to have a classification which covers both subjects of study of tertiary education, and subjects of vocational trainings of secondary education in all-embracing way. Whether this is the case and whether the categories are understandable for the respondents is in question. Latter one becomes quite interesting with regard to the interdisciplinary academic studies which increase a lot in the last years.
With regard to the social sciences surveys which included the question on the subject so far are the early rounds of the ESS and PIAAC. The instruments used in these surveys differ also because of their differences in the underlying concept. Both survey instruments and the final variables will be evaluated and compared with regard to concept, implementation and the observed distributions. The raised issues as well as the classification of ISCED-F, its conceptual background, issues around survey measurement and implementation into a cross-national survey tool will be discussed in the presentation in more detail.
Illegal activities, such as the corruption of public officials and the collusive tendering in publicly funded works, are relevant social phenomena. However, it is problematic to monitor them over time and space and to quantify their economic effects. For this purpose, this paper aims at using alternative sources of information for gathering data, such as newspaper and magistracy sentences, and to define a methodology to automatically identify relevant elements for studying facts related to illegal economy. By way of example, we present a study on 4632 judgments of Italian Magistracy issued between 2011 and 2016 on two types of crimes: corruption and collusive tendering. Using TalTac 2.0 for running the textual analysis, it has been possible to compare a huge amount of words (14 million) in a small amount of time, identifying common subjects among them and extracting all the relevant information and figures. The results in this study were made possible using text mining analysis. From extracted words and figures, we defined meaningful statistics, matching them with places, actors, organizations. Thus, it is suggested here a direct link between corruption and collusive tendering activities in the country and economic and social damages, establishing micro-data as useful information for the analyses of illegal activities. In particular, methodology constitutes a good starting point for studying phenomena like the ones presented in this work. Deriving quantitative data from words, spatial studies on the phenomena in question could be implemented. For example, evidence of spatial autocorrelation in corruption and illegal activities could lead to anti-corruption and anti-drugs regional policies. Furthermore, data could be evaluated considering the implications of spatial clustered illegal activities on socio-economic scenarios and could potentially improve research on social capital and institutional economy, focused on identifying what is commonly unidentified. Words, in the era of big data, can become a reliable source of information, especially in those fields in which figures are limited.