20.02.2013 Views

Qualitative_data_analysis

Qualitative_data_analysis

Qualitative_data_analysis

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

188 QUALITATIVE DATA ANALYSIS<br />

same case. We may want to record which of the categories have been assigned to<br />

which cases. If we are satisfied that these subcategories are conceptually distinct,<br />

then we could regard them as values of the underlying variable ‘suffering’ and assign<br />

these values accordingly. We could then discriminate between cases in terms of the<br />

variable ‘suffering’ and relate this variable to others in our <strong>analysis</strong>. Of course, the<br />

values for our variable must be exhaustive as well as exclusive, so we might include a<br />

value such as ‘other’ (or ‘awkward’ might do as well) for any cases, such as our first<br />

letter, where the categories are assigned in combination.<br />

The identification of variables with exclusive and exhaustive categories might<br />

itself be regarded as a major achievement of categorization. From a conceptual point<br />

of view, this requires a clear distinction between the boundaries of individual<br />

categories which can be grouped under an overarching category. The categories<br />

must not only be ‘exclusive’; they must also relate to and express the concept embodied<br />

in the overarching category. This is a task which I alluded to in discussing the<br />

problems of ‘splicing’ categories. The computer can support this conceptual task by<br />

providing quick access to category definitions and the results of assignment<br />

decisions. From an empirical point of view, we must check that one and only one<br />

value can be—or has been—assigned to each case. The computer can help by<br />

allowing us to look for and deal with ‘overlaps’ where more than one value which<br />

we want to regard as exclusive has been assigned to a case. For example, it could<br />

locate for us those <strong>data</strong>bits from the first and any other letters where more than one<br />

of the values (‘discomfort’, ‘disfigurement’ and ‘disability’) has been assigned to the<br />

case for the variable ‘suffering’. We can then check whether our initial assignment was<br />

reasonable, and if so assign this a residual value such as ‘other’. Providing there are<br />

not too many ‘others’ our variable may still prove a useful way of discriminating<br />

between cases.<br />

A simpler but less conceptually rewarding method of generating values is to note<br />

the number of times a category has been assigned to a case. In Table 12.2 we<br />

assumed our variables would have two values, either assigned or not assigned. But<br />

what if we have assigned a category to several <strong>data</strong>bits for each case? For any<br />

category, we can treat the number of assignations as a value for each case, and then<br />

use this as a basis for our cross-tabulations. The computer can easily identify these<br />

values for us, and provide information about the frequencies with which categories<br />

have been assigned to cases as well as the basis for cross-tabulating variables.<br />

Some qualitative analysts may feel very uncomfortable with some of the<br />

procedures we have just discussed. There is a strong aversion to numbers in some<br />

quarters, and a reluctance to accept that numerical considerations influence<br />

qualitative judgements. Nevertheless, it is difficult to see how, in practice, it is<br />

possible to identify associations between categories or assess the strength of<br />

relationships without recourse to a numerical evaluation. If we are looking for

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!