Content analysis

The dimensions must be entirely separated from each other, which means no overlaps. Equally, the categories for each dimension must be mutually exclusive and there should not be any ‘gray’ areas within dimensions which could leave coders uncertain of how to code accurately. These exhortations are not actually any different to instructions that could be given for other forms of structured research methods, like interviewing and observing.
Reference: Bryman: Social Research Methods: 5th Edition Page(s) 294,295

Content analysis is unobtrusive and tends not to suffer from the reactive effect, because the newspapers are not written with any sense that they might be subjected to this kind of analysis at some point in the future. It is flexible and can be kind to researchers with low financial resources. The time required can be considerable but the reward of high potential reliability can offset this. A further advantage is that it may allow us to gather information about social groups that are difficult to access, such as elite sections of society or celebrities, partly because it does not depend on direct observation or interviews with these people.
Reference: Bryman: Social Research Methods: 5th Edition Page(s) 302-305

Social researchers have examined the way in which certain words have been used as part of a wider discourse. Why are some words used more often than others? These can be deeply interesting questions concerning the reportage of events and the creation of a ‘mass-mood’ or feeling towards events and start by counting the individual words. Seale et al (2006, cited on p 291), for instance explored the frequency of particular keywords in breast cancer and prostate cancer posting. His team found men with prostate cancer were more likely to use technical words associated with research, whilst women were more likely to use words associated with feelings.
Reference: Bryman: Social Research Methods: 5th Edition Page(s) 289-292

Bryman (p285) explains that content analysis involves quantifying the content of a text or document according to predetermined categories, which is alleged to be a scientifically rigorous, ‘objective’ strategy. Because of the concentration on quantification of utterances in analysed texts, this is quite obviously a quantitative strategy, so answer (d) cannot be correct. It is not so much a research method, in the sense of data-gathering, as it is an approach to data analysis but this is handled so distinctively that most researchers refer to it as a method.
Reference: Bryman: Social Research Methods: 5th Edition Page(s) 285

Although the main use of content analysis has been an examination of mass-media printed texts, content analysis is not restricted to words. It can also be applied to great effect in analysis of images in magazines, films and animated cartoons. Bryman reports (p284) on its application to radio and television programmes as well as to the lyrics of pop songs. The focus of much content analysis is on communication, including that in published research reports.
Reference: Bryman: Social Research Methods: 5th Edition Page(s) 285

As with any other “method”, there are disadvantages. The base documents may be deficient because they are not representative, because older documents might have been lost or destroyed, for example, or they may have been distorted. Analysis of web pages could suffer severely from this effect. Furthermore, it must be said that with the best will in the world, “it is almost impossible to devise coding manuals that do not entail some interpretation on the part of coders” (Bryman, p305). Critics of this method have pointed out that even when applying ‘objective’ schedules of codes, researchers draw upon their everyday, common sense knowledge to interpret the meaning of categories. This inconsistency can pose a threat to the inter-coder reliability of the data.
Reference: Bryman: Social Research Methods: 5th Edition Page(s) 305,306

The coding manual is a set of instructions that helps coders decide how to assign codes to the textual data. It includes a list of all the possible categories and their corresponding code numbers. A good manual will include all the dimensions of the coding process and give guidance to coders to remove possible ambiguous inferences. We can understand how important it is when we consider some of the things that might go wrong in the coding process, like low inter-coder reliability, which would render an elaborate study fairly useless.
Reference: Bryman: Social Research Methods: 5th Edition Page(s) 293,294

Remembering that this is a quantitative method, it can be interesting to see frequencies of utterances in texts in different time periods. When sampling dates for content analysis, it is advised to make your sample as representative as possible, so that you can generalise your findings to other similar texts. In this case, Warde (1997, cited on page 288-289) did not select a random, probability sample of dates in the year, but he did choose magazines from one month in each of the four seasons. This was to reduce any sample bias that might have resulted from the types of recipes included in the magazines at particular times in the year, such as Christmas.
Reference: Bryman: Social Research Methods: 5th Edition Page(s) 288, 289

SPSS is a computer software package that aids quantitative analysis of numerical data (see chapter 16). It can be used for analysis of data generated by any quantitative strategy, with particular strengths for data derived from probability samples. Assuming the texts examined in content analysis to have been robustly sampled, then all of the numbers generated can be input to SPSS for statistical analysis. This includes the column headings in a coding schedule, the variables; with the rows entered as individual record data. NVivo is a similar type of computer programme for use with qualitative data (see chapter 25) but doesn’t help with this kind of data.
Reference: Bryman: Social Research Methods: 5th Edition Page(s) 293

Units of analysis are the tangible objects or subject matter that are coded as data in content analysis. These might include the people who produced the text as well as the people who figure most prominently in it; the item type, distinguishing between editorial comment and features, for example; the text perspectives and themes; and even actual words, including the frequency of their use. So a unit of analysis means what we study rather than how well we study it.
Reference: Bryman: Social Research Methods: 5th Edition Page(s) 289-292