|
Bojan Korenini - Zenel Batagelj Segmenting the market according to consumer's benefits and value orientations - the application of structurally determined laddering method |
|
|
Bojan Korenini - Zenel Batagelj Segmenting the market according to consumer's benefits and value orientations - the application of structurally determined laddering method |
As marketing practice accepted the viewpoint that company does
best when directing its activities according to the needs and
desires of customers in chosen target markets, the necessity for
the research methods that would increase the depth of understanding
of the consumers emerged.
The laddering research method counts as one of new approaches
that are trying to fulfill this request. Although there are different
approaches to laddering, the method originates in the field of
qualitative methodology. Theoretically it draws from means-end
theory that treats objects as means to achieving certain valued
states, or ends. By using a series of directed probes, typified
by the question "Why is this important to you" it identifies
the specific linkages between product attributes, consumer benefits
and consumer's value orientations. The data create a hierarchical
value map. Such map is treated as a directed graph, which shows
a unique way in which a product is linked to consumer's personality.
There are a lot of different quantitative and qualitative research
methods that are aimed at exploring the relationship between the
product and consumer's value orientations. In the case of quantitative
research methods, there is a lack of information, how specifically
different characteristics of a product fit into the consumer's
life. On the other hand qualitative methods are better at revealing
such information, but because of unstructured nature of such methods
they are usually conducted on smaller samples. This fact qualifies
them to explore just the respondents under examination rather
then to be representative for the whole population of consumers.
Laddering overcomes such problems. The method provides information
about specific individual-based linkages between consumer product
characteristics and consumer's value orientations. Because of
structured gathering of the data, method can be conducted on larger
samples.
There are different approaches to laddering: soft laddering, hard
laddering and structurally determined laddering method. The later
was developed by the authors of this paper. Although one of the
advantages of soft laddering method is that it can be conducted
on larger samples, in practice sample size usually doesn't exceed
50 or 100 respondents, because soft laddering interviews are difficult
and require specially trained interviewers. Besides identification
of common structures in complex qualitative data is very laborious.
Most frequently the type of marketing decisions where laddering
method proves supportive require statistically grounded conclusions.
The new approach - structurally determined laddering can be conducted
on several hundred or more respondents (it can be also conducted
over the phone). In this way conclusions of the laddering research
are no longer limited to the sample of respondents under the scope.
Beside that structurally determined laddering opens a lot of new
possibilities, in the sense of analysis and practical application
of research results, which depart significantly from the possibilities
that soft laddering could offer.
Authors of the paper wish to discuss different approaches to laddering
method in the context of quantitative and qualitative research
methods, statistical analyses of such network data and benefits
that follow from the application in marketing, mainly in the field
of market segmentation.
|
|
Simona
Korenjak- erne - Vladimir Batagelj - University
of Ljubljana Clustering Personal Networks as Symbolic Objects |
Personal or ego-centered networks are frequently encountered
in the social science research. A unit of the analysis is a respondent
(ego) with his/her personal network (alters). Each unit is described
by selected variables, which are usually measured in different
scales.
Several (other) variables are also measured on alters, that describe
relationship among ego and alters and properties of alters. Personal
networks can be very large.
In the paper we focus on two problems:
- Presentation or description of a personal network.
- Reduction of a personal network's size.
For the descriptions of personal networks symbolic objects are
used.
Using variable's distributions (instead of the value of an appropriate
statistics - e.g. mean value used in the "standard"
approach), the symbolic objects provide a more detailed description
of personal networks. Based on this descriptions the adapted version
of the leaders method was developed, which is a variant of the
dynamic clustering method.
The results of the proposed approach will be presented on the social support networks in Ljubljana 2000 data set, collected at the Faculty of social sciences of University of Ljubljana.
|
|
Mark
Sh. Levin - Ben-Gurion University of Negev Towards Multilayer Network and Combinatorial Models |
We examine multilayer network, multilayer cellular automata,
and their combinations. Applications are oriented to complex composite
systems (social networks, organizational systems, sociotechnical
systems, etc.).
Multilayer network consists of the following: (1) basic elements
for each layer: a set of elements (persons/individuals, tasks,
groups/teams, rules, rights, rooms, goods, etc.) and some structures
on them (binary relations, hierarchy, etc. as horizontal correspondence);
(2) binary relations between structures of different layers (vertical
correspondence). k-layer cellular automata is the following: (a)
a set of elements at each layer; (b) each element has several
types of basic states and four kinds of influences (logical or
probabilistic functions, optimization models) as follows: selfinfluence,
neighbor influence, influence of the higher layer, and influence
of the lower layer; (c) layers can be composite ones. The following
combinatorial problems are considered: (i) assignment/allocation;
(ii) clustering (e.g., grouping, skeleton clustering); (iii) routing;
(iv) approximation / covering; and (v) multiple matching.
A list of applied examples involves the following: (a) allocation
of personnel; (b) allocation of information access and / or decision
making functions into a set of specialists; etc.
SVG (Scalable Vector Graphics) is a language for describing
two-dimensional vector graphics based on XML (eXtensible Markup
Language). Three different types of graphic objects are supported:
paths consisting of straight lines and curves, images and text.
SVG provides all transformations which are usual in standard vector
graphics packages. Additionally, animation can be applied to SVG
pictures. Pictures in SVG can be examined using Web browsers using
a special plug-in. Objects in SVG DOM (Document Object Model)
can be accessed using JavaScript language and in this way parts
of the picture can be manipulated dynamically from the Web browser.
In the paper some approaches to visualization of social networks implemented in Pajek as options for exporting layouts to SVG will be discussed and illustrated by some typical examples.
|
|
Wolfgang
Sodeur - University of Essen Patterns of Institutional Affiliations of Household Members and Children's Social Contacts |
The aim of this paper is to analyse the impact of properties of the local household environment on structural aspects of children's social relations.
The structure of social relations of children is described
by some aspects like the relative amount of social relations at
different locations (home, street, school, sport clubs), multiplexity
of contacts (same persons at different locations), more or less
restricted size of social circles (number of persons met simultaneously),
and so on which seem to be important for socialisation processes
under a theoretical perspective. On this basis children are clustered
into a few classes. Afterwards these classes are partly reconstructed
by means of patterns of relations of other (mainly older) members
of the households.
The empirical data were generated in 40 systematically selected
small areas ("extreme cases") with 3000 to 15000 inhabitants
each. Within each regional unit, a random sample of about 25 children
age 8-9 and 13-14 was drawn. The children and one parent were
interviewed, mainly about their personal relations and the institutions
they were affiliated with.
