GT was developed, like other scientific advances that are prefaced by disenchantment with the prevailing orthodoxy, as a reaction to the then over zealous preoccupation with verification of theory. GT’s basic and central theme is generating theory from data that are systematically obtained from social research; consequently, GT is an inductive process (Glaser & Strauss 1967). It is a method for inducing and developing theory that should provide clear enough categories and hypotheses to explain and aid understanding of the basic (psycho)social process being studied. The theory evolves continuously from the data during the process of the research in that, unlike many other methods, the researcher does not commence with existing literature but begins with an identified area of study. Then, as data are collected, the process of constant comparative analysis occurs (a central feature of GT) whereby each item or label of data is compared with every other item or label. Glaser and Strauss (1967, p. 32) stress that this strategy of comparative analysis within the induction of grounded theory puts a high emphasis on:

GT usually occurs when there is little or no research into the subject or area. Consequently, research questions in GT (if present at all) are markedly different to research questions postulated at the start of a deductive study. Indeed, Glaser 1978, Glaser 1992, Glaser 1995a and Glaser 1995) repeatedly purports that a ‘true’ GT begins only with a ‘general wonderment’. The key research question, Glaser predicates, will emerge as the key (psycho)social problem facing the population being studied. Though it should be noted that a more contemporary view outlines the difficulties that such an approach creates and how it may be prudent, for a number of reasons, to include a non-specific question (Cutcliffe 2005).

A crucial difference between GT and other approaches is its emphasis on theory generation and development (Glaser 1967, Glaser 1998 and Glaser 2001). It is a way of thinking about and conceptualising data, to induce theory that is grounded in the reality of the research participants (Smith & Biley 1997). As a result, the theory can be seen to originate from the ‘ground level’ from the (psycho)social world from where the data originate. As Stern (1985), who was among the first nurses to use GT in a nursing focused study suggests, GT scientists construct theory from the data rather than applying a theory constructed by someone else from another data source. As a result, a GT should therefore ‘fit’ the situation being researched. To sum up:

As stated above, a principal feature of GT is its emphasis on theory generation, not conceptual description, nor theory verification. That is not to say that description or verification have no place within GT; these processes, however, are subsumed within the overall process of theory generation. A generated GT then can be presented in many forms:

GT produces two basic kinds of theory, substantive or formal theory. Grounded substantive theory is theory that has been developed for a substantive or empirical area of (psycho)social enquiry, for example client care. Grounded formal theory is theory that has been developed for a formal or conceptual area of (psycho)sociological enquiry, for example deviant behaviour. Glaser and Strauss (1967, p. 34) asserted that substantive theories are usually induced from the data and formulated first and then these substantive theories are followed by formal theories. Furthermore, they pointed out that it is the induced substantive theories that then enable new formal theories to be generated. As such:

When referring to grounded substantive and grounded formal theory, Glaser and Strauss (1967, p. 32) indicated:

Glaser and Strauss (1967) indicated that a GT will have theoretical elements that are generated by comparative analysis; these are conceptual categories and their properties, and hypotheses or generalised relations among the categories. Each of these elements is defined as:

As categories begin to emerge from the data, the researcher’s attention is fixed on a category (Glaser 1992). This attention then helps the researcher discover emergent properties about the category. At this point in the theory generation though, such categories and their properties only conceptualise what the researcher can see (Glaser 1978), and the researcher needs the theoretical coding, which forms the connections between them which yield hypotheses. In turn, these hypotheses suggest how the incidents and categories may be related to each other (Glaser 1978).

However, Glaser and Strauss (1967) point out that these hypotheses have at first the status of suggested, not tested, relations amongst categories and their properties. The generation of such tentative hypotheses does not require an extensive compilation of evidence to establish proof, just evidence enough to establish a suggestion. In the early stages of the research, these hypotheses may seem unrelated, but through the ongoing processes of data collection, constant comparison, and the subsequent emergence of categories and properties, and their development in abstraction and interrelations, they begin to form an integrated central theoretical framework (Glaser & Strauss 1967).

Glaser and Strauss (1967, p. 45) pointed out that theoretical sampling involves:

Whether the researcher is concerned with inducing a substantive or a formal GT, the process of data collection is thus guided and controlled by the emerging theory. Glaser (1978, p. 36) embellished this initial description when he added that theoretical sampling is concerned with eliciting initial codes from the data from the start of the data collection through constant comparative analysis, and then:

Glaser and Strauss posit that this method of sampling is a radical shift away from the methods used when attempting to generate theory, whereby the limits, scope and origins of the sample are predetermined. A ‘GTist’, however, still needs a starting point, and Glaser and Strauss described this starting point as the researcher beginning the research with a partial framework of local concepts, which designate a few principal or gross features of the structure and processes under study. They provide a useful example of such local concepts and principal/gross features. If I am going to study hospitals, then I am aware that there will be doctors, nurses, patients, wards, treatment rooms, admission and discharge procedures. However, the researcher will not have any a priori frameworks of how these concepts interact and relate.

Theoretical sampling also involves the researcher’s sensitivity (Glaser 1967, Glaser 1978 and Glaser 1992). To enable the theory to emerge from the data, the researcher needs to be sufficiently theoretically sensitive to be open to the emergence of whichever codes emerge. Glaser (1992, p. 27) added that:

Glaser (1992) argued that theoretical sampling within GT can be simple and direct. Once the decisions regarding the collection of initial data have been made, plans for the collection of further data cannot be made in advance of the emerging theory. As the emerging theory points the way towards the next source of data, the researcher does not know which ‘way’ to go until he has collected, coded and analysed the initial data. Future sources of data are thus dictated by the answers given to earlier questions. Glaser (1992) suggests that this process allows researchers to reach the limits of their data and data collection resources, and thus reaches the relevant theory expediently. Similarly, a GTist using theoretical sampling is unable to cite the total number, the nature, or type of groups from which the data were collected until the research is completed (Glaser 1978).