Papers
Inferential logic in media content analysis
Pedro Silva1
psilva@sfsu.edu
19 October 2006
Content analysis is a research methodology that relies on proper application and interpretative restraint by the researcher. It can generate categorical data which, when statistically analyzed, may help visualize emerging trends. However, this is all it can do - to emerge patterns from categorized quantitative data. By itself, it does not correlate variables, much less show causal relationships; any such statements should be taken as conjecture. This does not mean inferring effects, for example, should never be attempted. But it does mean any definitive statements should be the product of additional research beyond the scope of mere content analysis. I present three approaches to media research with content analysis at their core, although not all use it in a methodological sense. Communication in the sovietized state, as demonstrated in Korea [3] tries to infer content from its reception, not unlike describing gravity through its effect: it might well be that something else is in fact responsible for the effect we attribute to gravity. The consumer's stake in radio and television [4] reports on a strict content analysis methodology and is an example of how such research cannot produce insightful discussion on its own. The unique perspective of television and its effect: a pilot study [1], on the other hand, demonstrates how content analysis can help scientifically infer trends-effects correlation, if validated through some qualifying process. 2 This paper attempts to use inferential logic as a tool to debate the new-knowledge generation potential of three different approaches to content analysis. The three following sections are thus named Inductive inference, deductive inference, and abductive inference.1 Inductive inference
Inference is usually considered the act of reasoning by deduction or induction. Deductive inference is reasoning from the general to the particular. Inductive inference is reasoning from the particular to the general. This is where, I believe, [3] situated their study. Logically, inductive reasoning cannot generate new knowledge; it is falacious by nature. One cannot freely generalize from the particular. That is not to say statistics don't have their place, but even statistical theory has strict limitations on its implementation and scope. Obviously, I do not mean to disregard the massive amount of qualitative data gathered: a total of 2975 interviews [3,p. 318] were conducted during the post-occupation of Korea period. The possibilities for in-depth understanding of the reception of the occupation are great. However, to infer the content of the Russian media system from it seems an inadequate methodology. This study is an example, then, of not an improper application of a content analysis methodology but a non-application of it.2 Deductive inference
[4] is half-way though creating new knowledge. Taking the results of his study, he briefly tries an interpretation in light of Wayne Coy's seven points or criteria for the evaluation of a station's performance (p. 324). By using an established theory (Wayne Coy's criteria) as a framework for the analysis of his own data, he is able to deduce some aspects, necessarily in a stricter sense, of his study. In other words, he can particularize Coy's argument to the networks under scrutiny and test the theory, his method, or the networks's performance. In doing so, he will undoubtedly be adding to the existing body of knowledge. Nevertheless, it can be argued this, too, is not equivalent to generating new knowledge, under inferential logic. Nevertheless, [4]'s methodology was appropriate to the case at hand, and its implementation apparently flawless. His motivation for conducting a content analysis study is well described in his statementOne of the more popular sports among what used to be called our ïntelligentsia" is that game I call "Let's Predict the Effects of Television." Everybody seems to be doing it, especially those who are best qualified by virtue of the fact that "they wouldn't have a television set in the house." (p. 319)This reflects an analytical concern, moving away from speculation and approximating scientific certainty. Should [4]'s approach be coupled with [3]'s, the resulting study would likely be both more valid and insightful than the sum of its components could aspire to.
3 Abductive inference
Abduction is the process of reasoning to the best explanation. In this sense, it is the reverse of deduction. Given a set of facts, abductive inference attempts to find the most likely explanation. Originally proposed by the philosopher Charles Sanders Peirce, abduction is shown to be the only logical process which actually creates new knowledge [2], but both deduction and implication (induction) are described as also part of science. What this means is that the process of generating new knowledge is necessarily hollistic. Methodologically, I might offer [1]'s work as an example of such a process. In this study, a set of facts - "landslide effect [of] mass hysteria" - was observed. From this effect, quantitative analysis, through content analysis, was conducted, and an explanation advanced as a result of it. In effect, this study presents new theory to accomodate a previously unobserved, or uncared for, experience. In doing so, the Langs employed the principles of abductive reasoning. It is interesting to note, nevertheless, that this type of inference is actually constituted by deduction and induction, merely altering the causal relationships between them. In other words, their methodological process allows them to derive a as an explanation of b - instead of deriving b as a consequence of a -, using inductive reasoning in the process.4 Conclusion
I hope to have shown how inferential logic can be applied to the analysis of social research methodology. It is a simple process which may help determine whether further study is needed, or what the scope of one's research is, for example. Nevertheless, the system has obvious flaws, in that a reasonable analysis requires some degree of conceptual framing. That is, as in a content analysis, one must define nuclear units and categories, and reduce the subject study to its components. This is required to understand where research logically stands. In doing so, the original researchers's aims become diluted, and the inferential logic test runs the risk of simplifying things too much. With that being said, as a tool, inferential logic takes some of the subjectivity out and substitutes it for a fixed framework of analysis built on mathematics and philosophy. It should thus stand to scrutinity of most schools of thought, be they humanities or social sciences.References
- [1]
- Kurt Lang and Gladys Engel Lang. The unique perspective of television and its effect: a pilot study. In John Durham Peters and Peter Simonson, editors, Mass Communication and American Social Thought: Key Texts, 1919-1968, pages 328-337. Rowman & Littlefield Publishers, Inc., Lanham, 2004.
- [2]
- Charles Sanders Peirce. Description of a notation for the logic of relatives, resulting from an amplification of the conceptions of boole's calculus of logic. In Memoirs of the American Academy of Arts and Sciences, volume 9, pages 317-378. American Academy of Arts and Sciences, 1870.
- [3]
- Wilbur Schramm and John W. Riley. Communication in the sovietized state, as demonstrated in korea. In John Durham Peters and Peter Simonson, editors, Mass Communication and American Social Thought: Key Texts, 1919-1968, pages 310-318. Rowman & Littlefield Publishers, Inc., Lanham, 2004.
- [4]
- Dallas Smythe. The consumer's stake in radio and television. In John Durham Peters and Peter Simonson, editors, Mass Communication and American Social Thought: Key Texts, 1919-1968, pages 318-328. Rowman & Littlefield Publishers, Inc., Lanham, 2004.
Footnotes:
1Broadcast and Electronic Arts Department, San Francisco State University 2This can be another content analysis, since cross-analysis leads to cross-correlationFile translated from TEX by TTH, version 3.77.
On 24 Apr 2007, 23:50.
