Prospects and Limitations of Partial Ranked Data Analysis in Social Science Research – Investigations of Social Networks and Support

Abstract

Demands for care are met not only by professional care services but as well by family members and friends, especially by partners and children. These personal resources are not equally available to everyone. Moreover, there may be differences in preferences when choosing between alternative care arrangements. In particular, the often intimate situations arising in nursing care will lead many to prefer close relatives or friends over professional care services. We consider preference orderings between potentially supporting persons or professional care services. The orderings were obtained by asking respondents to name the most preferred groups of persons they would ask for support in a hypothetical case of being in the need of care. This question is part of the standard panel questionnaire program of the Socio-Economic Panel Study (SOEP) and is thus regularly posed. Similar questions to state preferences among a given set of potential alternatives can be found in many social surveys. The analysis of such data requires suitable methods that take their special features into account. We present a method to visualize partial rankings. The method is then used to present rankings over partial preferences, where the set of alternatives depends on the availability of networks of relatives, partners and friends. The method visualizes rank data using the skeleton of polytopes, by which the complexity of given preferences can faithfully be represented. Using the method for responses to the question about nursing care clearly demonstrates a shortcoming of this type of questions in general survey questionnaires.