ECREA16-Abstract

391

Friday, November 11

1 6 : 0 0 – 1 7 : 3 0

identified by a systematic online key word search and we only examined stories, which consisted of at least two different multimedia elements except from

text (N = 115 stories published by 49 editorial offices). The stories treat topics relevant for shared cultural events: A quarter focuses on politics and society,

18% on history, and 10% on culture. Storytelling and usability are innovative: On average, the stories contain eleven segments, combining three different

multimedia elements (which are mostly text 96%, photo 97%, and video 96%; only 40% use audio elements). The majority also integrates complex data

visualization. Furthermore, although the stories experiment with new ways of navigation, most of them still indicate a rather linear reception process: Only

37% offer different ways through the story and only 19% allow a selective reception in any predefined sequence.

PP 376

The Virtual Doorstep: How Ordinary Citizens Experience Use of Their Material by Mainstream Media in Constructing Collective

Memories

G. Cooper

City University London, Journalism, London, United Kingdom

The relationship between journalism and collective memory is a complex one; in the past, no main theorist of collective memory included ‘newsmaking’as

an important component of the field. Yet for most of the public, journalism is a primary source of information about the past and shared understandings

of it (Kitch, 2008). Rapidly changing technologies and an increasingly fragmented media may require us to find newmodels for understanding what Neiger

et al (2011) call “media memory”, particularly as the internet also allows the possibility of memories being formed by immediate public documentation

via social media (Hoskins, 2009). But what consequences are there for the ordinary citizen whose personal memories via tweets, posts or pictures are used

by the media – often without the creators’ knowledge or consent - to form collective memories of crisis events? This paper draws on 25 semi-structured

qualitative interviews with those caught up in the Great East Japan earthquake (March 2011), the Oklahoma tornado (2013) and the Vauxhall helicopter

crash (2013) and whose images/words were used by two major UK media players, the BBC and the Guardian, in liveblogs, often without their knowledge

or consent. These users were then questioned about the consequences for them of their material being used, with the aim of answering the following

questions: • How are voices of citizens in crises being used to create newmedia memories? •What were their feelings and actions on having private memory

become part of a public act of memory. • What issues does the use of this content raise around contextual integrity of privacy and how these norms can be

further transgressed by the mainstream media?

PP 377

Beyond Statistical Reasoning on Reliability in Standardized Content Analysis: A Constructivist Reflection About the Social Aspect

of Methodology

J. Völker

, A. Scholl

WWU Münster, Department of Communication, Münster, Germany

As reliability is a major quality criterion of standardized content analysis (in addition to validity and objectivity) a lot of effort has been put in developing

reliability coefficients (e.g. Krippendorff 2004) which aim to measure and evaluate the reliability of the same or of different coders’coding performance, to

find the weaknesses of the category scheme and to improve both the category scheme and the coding performance. We do not challenge this aim. How‑

ever, we want to focus on the complexity of textual structures with the help of reliability coefficients. A medium-sized reliability coefficient not only tells

us that the coding scheme and coding rules or the coders’performances are below the desirable scientific standard but also that the text is complex and its

semantics are ambiguous. However explicit and algorithm-like the coding rules are defined, the coding procedure itself relies on common sense knowledge.

Otherwise a computer could replace the human coder (cf. Scharkow 2013: 299–302).The more content analytical categories not only are syntactic or formal

but touch the semantic or pragmatic level of meaning (“projective variables”, cf. Potter/Levine-Donnerstein 1999: 259), the more the relationship between

the coder and the text is complex. Instead of a dyadic relationship between coder and coding rules we prefer a triadic relationship between the text’s

complexity, the coding rules’ unambiguousness and the coder’s comprehension of the text. From a constructivist perspective the text neither has a single

objective meaning nor is its reception idiosyncratic. Thus, reliability should not be considered the result of the correct coding of the text’s real meaning but

a mixture of explicit and implicit concurrences of the coders’coding practice. Beyond the straightforward perspective of evaluating the quality of the mea‑

surement we go into more detail to get further (meta-) information about the text’s semantics from the reliability score. We discuss three potential reasons

for a reduced reliability: a) If the coding rule instructs the coder to identify not only the general theme of the text but more specifically different facets

of the theme within the text, coders will probably identify a different number of thematic facets. b) If two values of a variable are semantically close to

each other and cannot be strictly separated because of overlapping connotations, coders will probably not reach a consensus. c) If coding of a text passage

heavily depends on the context (e.g. whether a person is prominent or not), the coding result will probably vary according to the coder’s world knowledge.

Eventually the researcher has to decide whether to accept a lower reliability or to cancel the problematic categories. As a consequence, reliability and va‑

lidity appear to be antagonistic rather than to be based on each other: The easier it is to gain acceptable reliability, the less valid are the categories as they

do not measure semantic or pragmatic meanings of the text. To take this antagonistic relationship serious, we argue for a combined reflection on reliability

coefficients and validity related issues in research studies rather than reporting reliability coefficients only.