Wednesday, March 19, 2014

Generation of Theory from Qualitative data: Summary of Chapter 8: The Flexible Use of Data

The 2nd part of the book named The Flexible Use of Data consists of Chapter VII and VIII.
Chapter VII:  Theoretical Elaboration of Quantitative Data
Main Theme:  The generation of theory from Quantitative data is discussed in detail.

·         Too much emphasis is given to the verification of quantitative data that it becomes difficult to generate new theories out of it. However, some of the quantitative data has great potential of discovering a theory and it breaks away the limits of ‘verification’ and ‘preconceived conceptual schemes’, to give us a theory. But the researchers are focused on the ‘rhetoric of verification’ that their results are presented as ‘plausible suggestion’.
The generating capacities of these sociologists and the richness of their research are not given the fullest impetus’ pg. 185

·         Discovery made through quantitative data is treated as a ‘byproduct of the main work’. When discovery forces itself on the researcher he writes his ‘induced hypothesis’ as if it was thought up in advance and they seem to fulfill the requirements of verification. In rare examples we come across ‘purposeful generation of grounded theory… in short papers where a single carefully worked out explanation of a hypothesis is offered’.
·         Sociologists need to give themselves freedom to use the quantitative data flexibly or they will not be able to generate a theory out of it. Rules of (sampling, saturation, integration, density of property development, etc.) need to be relaxed not for accuracy and verification but for generating theory from quantitative data that will lead to new avenues of quantitative analysis. Rigorous rules for accuracy and verification can be relaxed at strategic points to facilitate theory generation from quantitative data.
The authors say that new rules of theory generation will be developed, some from the existing ways of data collection and some from new ones.
 They believe that there are many styles of quantitative analysis having their own set of rules and their aim is to describe how these styles can be flexibly adapted to generating theory.

Secondary Analysis of Quantitative data
If the purpose of the sociologist is to generate a theory then he/she is more likely to analyze previously collected data (Secondary Analysis), mainly for two reasons:
One it is easier, responsibility of the researcher is to generate theory. Secondly, the task of description (of the total survey) and theoretical analysis can be conflicting. Therefore, most generation of theory will be based on secondary analysis and sociologists with a ‘theoretical bent’ prefer it. Comparative Analysis requires secondary analysis, example: population from different studies, etc. Trivial data, like a market survey on the consumption of products, can have important theoretical relevance.
Limitations of Secondary Analysis
1-      It is difficult to pin down the accuracy of findings because it is necessarily a secondhand view, population goes through constant change, and data collection procedures may not be known. But we need to keep in mind that the problem of accuracy is not as important for generating theory as it is for describing a particular social unit or verifying hypothesis.
It is the general categories and properties and the general relations between them that emerge from data that are important. These can be applied to many current situations as relevant concepts and hypothesis regardless of whether the specific descriptions are currently accurate for the population. Therefore, secondary analysis is uniquely well suited for the generation of theory but limited for description and verification -------------for which it is mostly used.
2-      Another limitation is representativeness of the population studied. Accuracy is crucial in description and verification so the sample should be carefully chosen by random sampling. Secondary analysis of a random sample may lead to systematic and random biases into the secondary study, making its accuracy questionable. It is indeed difficult to be able to answer all wh- questions regarding the data collected previously however, if theory generation is the purpose than representativeness is not an issue for these two reasons:  The direction of a relationship (used to suggest a hypothesis) is assumed until disproved and theoretical (not statistical) sampling guides the choosing and handling of the data.
3-      Being more important, scope of population can be increased by being less concerned about representativeness. The sociologist takes carefully stratified samples from a larger survey sample and cuts down on scope by omitting contaminating influences.

Concepts and Indices
A lot of work is done in the last decade, on the topic of flexible use of concepts and their empirical indices, in quantitative analysis. Lazarsfeld has describes the process by which concepts are translated into empirical indices. Authors have discussed a few points in this category.
When the main objective of the survey analysis is theory generation then crude indices are sufficient to indicate the concepts of the theory and to establish relationship between them which becomes the basis for suggesting hypothesis.

Discovering Hypotheses
‘In generating theory, preconceived hypotheses are not necessary for correlating or cross-tabulating two variables (runs) with indices of core categories and properties. The rule for generation of theory is to have sensitivity to all possible theoretical relevance. But the reverse is true in preconceived hypothesis for verificational studies. According to verificational rules data should be collected after the formation of hypothesis. For generation of theory data can be collected at any time. Mostly it is collected beforehand.
To explore all possible findings, for suggesting hypothesis the analyst may run the core concepts with all remotely relevant items of the questionnaire. At this point theory of the core indices starts emerging. Clusters of items are discovered as associated with the index. This strategy discovers theory by providing link to be conceptualized and analyzed. A theory is induced from the general relationships analyst has found.
Liberties in Presentation of Data
Quantitative data is presented in tabulated form but this need not to be so in generating of theory. All the relationships on which grounded theory is based are so huge that it would make the report of the theory too bulky. It is also important that peers and laymen are able to understand the theory therefore the analyst can make some deviations while presenting data and describing it. The data should not be changed of course, but all data does not have to be presented and described in detail.
No information is distorted, only enough information is presented to show, in the simplest way, the grounded basis of the theory. Data that is not important for theoretical analysis can be left out
Theoretical Elaboration
Next step is elaboration analysis ----‘to make three or more variable analyses to saturate categories further by developing their properties and thereby achieving a denser theory. So, the discovery of relationship among the indices gives the beginning suggestions for a theory and a theoretical direction and focus for its elaboration.
In ‘elaboration’ structural conditions of two variable associations are specified, their causes and consequences are sought, with possible false factors checked for, and their intervening variables discovered.
Lazarsfeld has given three ways of ordering the variables in an elaboration analysis. They are:
1-      Temporal                             2-structural level of complexity                 3- Conceptual generality
However, Lazarsfeld used only the first type of ordering and the other two were merely suggested.      
Elaboration analysis is stopped right in the beginning because survey data is cross-sectional in time, it’s difficult for analyst to establish clear cut, factual time order and there is too much temporal relationship in survey variables.
Elaboration analysis is stimulating because its findings ‘fit’ the thought patterns of sociological theory. The analyst can literally speak through tables and infer from his indices the conceptual level of his talk.
Theoretical Ordering
The theory emerges from data if the purpose of the analyst is to generate theory because then he is no longer concerned about the temporal ordering, required for verification and description. The analyst then moves ahead to order his variables theoretically, (a new principle of ordering).
Lazarsfeld is not able to develop a general theoretical ordering principle because he does not realize their similarity. He fails to understand that temporal sequence can be handled both theoraticallly and factually.
In generating theory as it emerges,
1-       the analyst discovers two-variable relationships
2-       Discovers their elaboration
3-      Generates possible further elaborations of two-variable relationship within the previous relationship
4-      Goes through data to look for indicators for concepts he thinks are related in theoretical ways to his emerging theory
5-      Arranges elaboration tables to test his hypothesis (for suggestion or discovery, not verification) Here, he is theoretically sampling his data, as directed by his emerging theory.
Consistency and elaboration analysis result in a grounded basis for his theory.
Conclusion: Careful relaxation of rules of quantitative analysis can generate a theory.
Successive Stages of Building Up to Theory from Quantitative Data

1-      Most frequent source of data used for generating theory
2-      Indicating categories and properties with the data
3-      Discovering hypotheses with conceptual indices
4-      Theoretical elaboration of hypotheses










The Interrelated Processes of Data Collection
Data Ordering, and Data Analysis to Build Grounded Theory

Data Analysis (4)


Theory Development (5)

Data Ordering (3)



Theory Saturation ?
Data Collection (2)




Theoretical Sampling (1)

Within this general framework, data analysis for each case involved generating concepts through the process of coding which, ... represents the operations by which data are broken down, conceptualised, and put back together in new ways. It is the central process by which theories are built from data. (Strauss and Corbin, 1990, p. 57.)
‘Five analytic (and not strictly sequential) phases of grounded theory building were identified: research design, data collection, data ordering, data analysis and literature comparison. Within these phases, nine procedures or steps were followed.





Table 1: The Process of Building Grounded Theory


Step 1
Review of technical
Definition of research question
Definition of a priori constructs
Constrains irrelevant variation and sharpens external validity
Step 2
Selecting cases
Theoretical, not random, sampling
Focuses efforts on theoretically useful cases (e.g., those that test and/or extend theory)

Step 3
Develop rigorous data collection protocol
Create case study
Employ multiple
data collection

Qualitative and quantitative data
Increases reliability Increases construct validity
Strengthens grounding of theory by triangulation of evidence. Enhances internal validity
Synergistic view of evidence
Step 4
Entering the field
Overlap data
and analysis
Flexible and opportunistic data collection methods
Speeds analysis and reveals
helpful adjustments to data collection
Allows investigators to take advantage of emergent themes and unique case features

Step 5
Data ordering
Arraying events chronologically
Facilitates easier data analysis. Allows examination of processes

Step 6
data relating to
the first case
Use open
Use axial

Use selective

Develop concepts, categories and properties
Develop connections between a category and its sub-categories
Integrate categories to build theoretical framework
All forms of coding enhance internal validity
Step 7
Theoretical sampling
Literal and theoretical replication across cases
(go to step 2 until theoretical saturation)
Confirms, extends, and sharpens theoretical framework
Step 8
Reaching closure
Theoretical saturation when possible
Ends process when marginal improvement becomes small

Step 9
Compare emergent theory with extant literature
Comparisons with conflicting frameworks
Comparisons with similar frameworks
Improves construct definitions, and therefore internal validity
Also improves external validity by establishing the domain to which the study's findings can be generalised

Source: The Creation of Theory: A Recent Application of the Grounded Theory Method
by Naresh R. Pandit

No comments: