Mixed methods |
|
| Chair | Miss Aida Montenegro (University of Bonn ) |
Background: Survey questionnaires commonly include attitudinal response questions on a construct of interest. However, respondents’ values considerably influence behaviours such as their attitudinal responses. When comparing groups that have different and unspecified values these ‘incidental’ differences can overwhelm differences in the constructs of interest. The problems are exacerbated in cross-sectional, cross-cultural and longitudinal research because socio-cultural groups have different values that change over time. Traditional Positivist research, guided by the classic/true-score approach, has responded by costly solutions of lengthening questionnaires and increasing numbers of respondents – or even ignoring values-change, as in pre-post treatment comparisons. Alternatively, to contain costs, obfuscating modelling e.g. partial credit Rasch transforms of Likert scores, has been used to mediate differences due to cross-cultural values and so recover the construct information from the original responses – all to questionable avail.
Aim: To develop and test a computationally simple and intuitively acceptable method of mediating the most influential personal and group value differences affecting attitudinal responses. Hence, to improve sensitivity of comparisons, for both fuzzy (emic) and delimited (etic) attitudinal constructs, in cross-sectional, cross-cultural and longitudinal research.
Method: We followed the Positivist intention of recovering construct information by modelling and transforming the attitudinal response. However, Constructivist (respondent - Q related) rather than Positivist (variable - R related) principles were used. Thus the attitudinal response was deconstructed using the Constructivist assumptions of Cognitive Interviewing rather than the Positivist assumptions of True-Score Theory. Consequently, two constructivist subjective confounded components of ‘judgement’ and ‘expectation’ were proposed, where values are assigned to the expectation component. These corresponded to the two Positivist subjective confounded components of ‘true-score’ and ‘error’, where the values are assigned to error. Positivist assumptions (e.g. errors N(0,2)) allow the error to be removed by averaging across responses. However, these assumptions do not hold across respondents who have different cultural values. Hence, a within-respondent method was used to mediate respondents’ expectation components. Culturometrics is a Constructivist paradigm whose self-norming methods espouse the Humanist sentiment “the measure of man is man”. We therefore used a culturometric self-norming method to transform each response into a Cultural Index (CI) that retained the judgement information and could be reported relative to any chosen set of cultural values.
CI use is reported here in two research contexts. The first is a one-on-one Mall interrupt survey (N=126) to assess the relative attitudinal contributions of Jamaican, African and Anglo-American cultures to the econate identities of native Jamaicans in Kingston, Jamaica. Attitudinal responses indicating cultural components of identity were transformed to CIs by culturometric self-norming. Cross-gender and inter-generation comparisons were made based on these CIs and tested for concurrent and predictive validities. The second study reported here is a longitudinal (pre-post treatment measures), cross-sectional and cross-cultural attitudinal evaluation of Anger Control among prisoners (N=85) in four Trinidadian prisons.
By transparently mediating values the Cultural Index allows moreWRONG COUNT ERROR TO CORRECT
I investigate the students’ proactive behaviours in a university context. The students’ proactive behaviours are being analysed within the concept of agentic engagement. This form of engagement refers to the contributions to the flow of instruction initiated by the student. From this perspective, students find ways of enriching, modifying and personalizing their learning and instruction. Agency as a dimension of engagement needs more research to validate the construct (Sinatra et al., 2015). My study contributes to the understanding of agentic engagement by investigating the students’ proactive behaviours in a real learning context. In doing so, I use a mixed method that combines the following data collection instruments: students' self-reports, researcher's observation forms and interviews with students enrolled in a German university.
Specifically, the observation form is designed in such a way that time, the type of proactive behaviour and the quality of autonomy support are reported. The observation form presents a list of students’ proactive behaviours grouped into three categories named (1) Initial Self-selection, (2) Volunteering Initiative as a Response, and (3) Initiative to Transform a Sequence. The observation is systematically done during 8 weeks before administrating the questionnaire. In order to identify the most appropriate items to evaluate the construct, five items of the scale of agentic engagement (Reeve, 2013) and the results of the observation form are contrasted. For the final design of the questionnaire, the other types of engagement (behavioural, emotional and cognitive) and a set of motivational constructs (goals and perceived autonomy support) are included. After applying the questionnaire, the interviews are conducted with questions related to the behaviours included in the observation form and the items of the questionnaire.
In order to analyse numeric and narrative data, the analytical approach takes into account discourse analysis and statistical analysis. For this presentation, I will describe the process of designing, applying and analysing an observation form on agentic engagement, as well as the selection of items for measuring it, by comparing a previous scale with the analysis of the observation form. I will also present the challenges and reflections on bilingual questionnaires and Likert-type measures for participants in a German university context.
In the context of the recent global financial crisis Greece has been severely affected facing an acute recession echoed in the sharp increase in unemployment, poverty and social exclusion rates. Recession’s tremendous impacts are traced in the rise of citizens facing extreme hardship to deal with daily problems, such as inability to cover basic needs (e.g., food and clothing), lack of housing and no access to healthcare. In response to the crisis social solidarity networks have emerged targeting to assist socio-economically deprived individuals to improve their means of subsistence and health-care. The main rationale of the paper is to present the empirical findings and methodological implications of two quantitative studies exploring social solidarity activities at the local and the national level. The former is conducted in the urban area of Heraklion (Crete) and adopts the quantitative approach of content analysis using data collected from newspaper reports. The unit of analysis is the social solidarity action while the unit of data collection is the newspaper mention of specific events associated with social solidarity activities. The analysis shows the increasing trend of actions as the economic crisis exacerbates, the plurality in forms and activities organized from different social solidarity organizations providing preliminary evidence that the recession has driven the emergence of civil society in the Greek urban community under study.
The second study shifts the focus to the national level using data from an online survey conducted during 2016 with representatives of groups/organizations conducting social solidarity activities. The findings show groups’/organizations’ main type of activities, their major constraints to achieve their goals, their mechanisms and tactics to address citizens’ rights through conventional and non-conventional political actions, their collaborative networks as well as reported changes in different type of actions targeting specific needs since 2010.
The paper sheds light on the different facets of social solidarity activities as captured in the two quantitative studies. Most importantly it unravels the distinct methodological challenges faced in each of these quantitative approaches. In the first study, despite the merits of the method to track social solidarity activities at the local level, one of its major shortcomings involve the exclusion of activities that have not been published in the local press. The most critical limitation of the second study at the national level is associated with the use of retrospective measures, i.e. questions that ask respondents to recall information from the past (e.g. items measuring reported changes since 2010). Its second limitation is associated with the low response rate, which provides indicative rather than representative conclusions for social solidarity activities across the country.
With a goal to initiate a discussion, the presentation addresses the design, process, method and analytical framework of an empiric study, used to generate meta-inferences coming from different streams of research data, e.g. qualitative and mixed-method (quant/qual.) sources. A first stream of data is elicited from the qualitative text, and the second from open-ended reflections on the text and quantitative ratings obtained through a multi-country, multi-group stakeholder survey. Both streams of data are used to reconstruct and match discursive patterns of corporate communication and those of stakeholder values, attitudes and expectations. Both discourses are spatially represented in the form of semantic/concept maps produced using multi-dimensional scaling. Comparison of the two maps, supported by quantitative measures enables meta-level conclusions on semantic and semiotic relevance of the text to the audience. The presentation showcases convergence and complementarity of the mixed-method findings and the issues related to them.