The theoretical sampling

The study was developed considering the principles of theoretical sampling, as described in Strauss and Corbin (2008). The theoretical sampling is an essential step in the grounded theory process. It consists in the way taken by the researcher in the search for data to build the theory. The sampling is configured during the research process, in other words, the researcher does not determine previously which will be the sample groups, because the data are the ones that will reveal the way to be followed.

The researcher should develop some sensitive abilities in order to support some decisions, for example, the choice about the best cases considering the emergent data. The process of data collection and analysis, in this sense, is cumulative and presents funnel characteristics, to the extent that sample groups are becoming more selective, according to the appearance of evidences of theoretical saturation.

Therefore, the procedures of data collection and analysis were carried out in alternating sequences, and five sample groups were formed until the moment of theoretical saturation. In order to present the companies and managers that composed the theoretical sample letters and numbers were used. It was important to protect the participants’ real names. The general aspects of theoretical sample and sample groups are shown in Fig. 1. It is possible to notice the identification of the participants through letters and numbers.

[Figure ID: f1] Fig. 1.

Theoretical sampling.

  —Source: the authors (2015).

In the first sample group or open sampling, that was the first step of empirical research; six interviews were performed with managers of three companies located in Western Santa Catarina State. In this step, in-depth and non-structured interviews were conducted. The objective and question of the research were used as guide in this initial process. During the interviews, in several moments, the researcher intervened and asked questions as Why?, How?, Who? and What happened?. According to Strauss and Corbin (2008), questions like these ones are important in order to find quality information that will be essential in the moment of definition of categories, properties and dimensions.

In the second phase of the theoretical sampling, seven interviews were performed with managers of two companies located in Great Florianópolis region. The choice of these cases was influenced by the results of data analysis of the first sample group. It was noticed the necessity of exploring a large company and interviewing somebody related to the financial area and also interviewing an external consultant.

The third sample group was formed by four companies located in Great Florianópolis and South regions of Santa Catarina state. The results from the previous sample group helped in the cases selection and in the interviews. It is important to highlight, however, that in some situations the choice of “ideal case” was not possible, because of several reasons, such as difficult access to the company.

In the fourth step of the theoretical sampling, the process of selective sampling got started, with the objective of integrating the categories and identifying the relationships between them from the analysis of properties and dimensions.

The cases that formed the fourth sample group are located in the Great Florianópolis and Vale do Itajaí regions. The researcher decided to return to company J due to the relevance of the case and the interest of the manager in following the research.

The theoretical saturation of the model started becoming more evident with the interviews performed with the fourth sample group. In this moment, only some aspects, mainly related to the dimensional variations and name of categories, needed to be better clarified.

It is important to notice that is the researcher who realizes that theoretical saturation is happening. The evidences that emerge from interviews only reinforce this perception. In other words, the participants of the research do not say directly that the saturation happened, but they reinforce this perception. The researcher, through sensibility, should gather information and conclude if the research is ready to be finished.

Thus, it was decided to finish the research with a fifth sample group, returning to the first group, in order to do a selective sampling and also performing one of the research validation steps. The initial intention was interviewing all managers that had previously participated, however, for several reasons, this possibility was ruled out.

The data collection and analysis

In Strauss and Corbin's (2008) approach, the term grounded theory is used in order to refer that the theory was derived from systematically grouped and analyzed through a research process data. The data collection and analysis should constitute an interlaced process with the analysis, conducting to new sampling data (Gray, 2012). The analysis starts as soon as the first interview is finished, and the results conduct to the collection next phase and so on, until the theoretical saturation, in other words, until the moment when data do not indicate relevant things, aside from what is already known (Strauss & Corbin, 2008).

In the research presented here, the interview was the main way used for data collection. The preference was for in-depth and unstructured interview in the beginning and semi-structured interview during the theoretical sampling. In this way, the flexibility during data collection was respected, and conditions for the researcher to explore both sensitive and creativity were offered. It is important to notice that other techniques, such as observation and technical literature, were used on a smaller scale during the building of the model.

In the data analysis, procedures such as microanalysis, open codification, axial codification and selective codification were used (Strauss & Corbin, 2008). However, the usage of these steps does not necessarily follow a sequence, because in for the most part of research these steps are simultaneous (Schröeder, 2009).

In this case, the data collection process started with the selection of the companies that were part of the first sample group. The selection criteria were based in the research's both objective and question. Basically, the study respected three criteria: (1) the company needs to be experienced in implementation; (2) must have been in the market for at least eighteen years; and (3) be located in Santa Catarina state. Regarding the first criterion, the researcher found three different scenarios: companies that had failure in the implementation; companies that had success in the implementation; and companies that were experiencing the implementation process. Some information about the first sample group are showed in Fig. 2.

[Figure ID: f2] Fig. 2.

First sample group.

  —Source: the authos (2015).

The initial contact with the companies that formed the first sample group was made by telephone and e-mail. The contact with the companies was the most difficult point during the realization of the research. In many cases there was no acceptances from managers in participating. In other cases, managers simply did not return the contact. These events are in line with what Strauss and Corbin (2008) have written. Because of this, in several situations it is not possible to investigate the “ideal case”.

In the moment of the interview, it is important to start with an informal talk about general aspects related to activities of the company. This point is important and recommended in order to create an enabling environment for the interview. After this informal period, it is relevant to ask if it is possible to start and record the interview.

After the conclusion of the interview, it is important to initiate the transcription process. The transcription of interviews is normally a hard phase and it demands a lot of time and attention. In this study all the interviews were transcribed by the researcher; what is considered important in order to preserve the proximity with data and to encourage both creativity and sensibility.

Whenever possible, the data analysis should be done between the interviews because the analysis results should support the next steps. In the present study this process was followed. Also, the software of qualitative data analysis NVivo was used in order to facilitate handling the data. From coded sections or analysis units, codes were created. Preliminary codes originated categories and subcategories.

Due to its characteristics, NVivo can be considered a software used to create theory, because it aims to support both process and procedures of grounded theory (Lage, 2011). The objective of NVivo is to offer the researcher an agile way to the data analysis process, in order to allow more reliability in the results and support the decision-making process (QSR International, 2012).

NVivo is considered an interesting tool to improve the data analysis process in grounded theory, because it allows the codification of material using several techniques, as codification of paragraphs or codification In Vivo. Similarly, with NVivo, it is easier to explain the connection between data to create categories and the relations between categories and subcategories. The functionalities offered by NVivo are considerable. However, it is important to emphasize that the researcher should be aware that the software is a support tool and the main decisions about the research will be made by the researcher.

Normally, open and axial codification are done in a simultaneous way. Firstly, codes are created through open codification and, after, they are grouped by similarity of meaning, originating the first categories. These seminal categories will support the build of the interview guide that will be used in the next research steps.

In the end of first step, it is possible to notice some needs and orientations, for example, which cases should be used, which questions should be added or removed in the interview guide. It is important to observe what data are indicating in order to avoid building a theory disconnected from the empirical evidences. In Fig. 3, some information about the second sample group is shown.

[Figure ID: f3] Fig. 3.

Second sample group.

  —Source: the authors (2015).

The third step of data analysis of this study was very hard. Several attempts of scheduling were done without success. So many managers seemed to be upset because of the large number of requests coming from the academic environment. In this sense, one interesting alternative can be trying to contact companies that are known as “open doors” to academicians. Some information about the third step of this study is shown in Fig. 4.

[Figure ID: f4] Fig. 4.

Third sample group.

  —Source: the authors (2015).

During the study some setbacks appeared, but some events increased the researcher's self-esteem. In several moments the managers congratulated the researcher for the work and declared that the theme was very interesting and was a real problem faced and hard to solve. In other moments, the managers told the researcher they would like to be informed about the progress of the research and its results.

During the codification process and data analysis the researcher developed some memos and diagrams. These analytical tools support the information record and in the research process in a general way. Memos, according to Strauss and Corbin (2008), are written records of analysis that can vary in type and format. Memos can be like codification notes, theoretical notes or operational notes. In this research, the memos were developed and stored in NVivo Software.

In the fourth step of data collection and analysis three companies that have a successful history in the market were selected. The interview guide used at this moment presented the theoretical model with its categories, properties and dimensions. It is important to highlight that the model was presented in the final part of the interviews, in order to avoid possible influences. The interviews were addressed to some model aspects that needed analytical deepening, for example, some properties, dimensions and names of categories. Some information about the fourth research step is shown in Fig. 5.

[Figure ID: f5] Fig. 5.

Fourth sample group.

  —Source: the authors (2015).

After the data analysis of the fourth research step, some modification was done in order to adjust names of categories, properties and dimensions. At this moment, the researcher started the rereading of all interviews, in order to validate the theoretical model that had just been built. The rereading of interviews constitutes one step of research validation. As it will be presented in sequence, there are other ways to make the validation.

The fifth, and last, step of data collection occurred with the main intention of performing another step of validation. It is important to observe that the validation process occurs during all steps of theoretical sampling, however, the final validation is important to ensure that the information presented in the model are reflecting what is specified on the data. As mentioned before, in this study the final validation was made through the rereading of interviews and interviewing again the managers that formed the first sample group. Some information about the fifth research step is shown in Fig. 6.

[Figure ID: f6] Fig. 6.

Fifth sample group.

  —Source: the authors (2015).

Therefore, the fifth research step was formed by two companies that had participated of the first research step. Thus, at this moment, three validation interviews were performed. There were no significant disagreements regarding the results presented. Only two aspects related to dimensions suffered some adjustments. After this, the theoretical model was considered satisfactory and the research field was completed.

With the theoretical saturation, the substantive theory “A Phase of Change and Learning” emerged. As noted in the substantive reality investigated, the implementation of performance improvement systems can be understood from a perspective of changing and learning. Both structural and behavioral changes are required and, thus, the organization goes through a phase of development and learning, in which it acquires competencies to deal with a dynamic and evolutionary environment. The steps of grounded theory are summarized in Fig. 7.

[Figure ID: f7] Fig. 7.

Steps of grounded theory based on Strauss.

  —Source: Based on Strauss and Corbin (2008).

In Fig. 8 it is illustrated the model that represents the substantive theory “A Phase of Change and Learning”. The model is formed by one central category, which is also named as substantive theory, four conditions (Knowing the reality of the company; Planning; Knowledge about how to do; and Top management support), three actions/interactions (Communication; Peoples’ involvement; and Control of results), and two consequences (Competitive ability and Cultural change).

[Figure ID: f8] Fig. 8.

Substantive theory “A Phase of Change and Learning”.

  —Source: the authors (2015).

Change and learning take place when the company is successful in implementation, but also when the initiatives fail. In case of success, the company will be more developed and in a higher evolutionary level, characterized by the acquisition of skills and competences to deal with change. In case of failure, there was a change to a lower stage, because the company spent resources and returns were not observed. In this case, the resistance for change became stronger and it will hamper new initiatives to implement improvement systems.

Thus, the reality of the company is a condition that creates the situations that conduct to the implementation process. A company that has a good planning, with action plans, indicators and well-defined tasks, will have a favorable condition. Similarly, the degree of knowledge about the model that will be implemented, linked to aspects related to the project management will impact the change program. The support of senior management, in turn, is a condition that can change how the implementation takes place. For example, if there is change in senior management, the support for the implementation can disappear.

The strategy of involving people aims to maintain the support of senior management and motivate people through training and benefits from the change program. Similarly, maintaining communication is a necessary action to preserve a systemic view of the process and to keep people informed about what is happening. The action of control, in turn, aims to preserve what was predicted in the planning and keep company in the right way.

The consequences of the strategies performed are the culture change and competitive ability. According to the efficacy of actions such as involving people, maintaining communication and control, the consequences will be more or less satisfactory. Moreover, the way how conditions are structured will influence the results.

Each category, in its turn, was structured from subcategories, properties and dimensions. In this way, the description of each category was structured in order to enable the reader “enter” the concept and realize how it is grounded. In this way, from the knowledge provided by empirical reality, which is emphasized through the interviews, it was possible to understand how the implementation process occurs. From this comprehension, it is possible to make some reflections about the main aspects that organizations faced on when trying to improve their competitiveness conditions through the implementation of performance improvement systems.

The theoretical scheme validation

The validation of the theoretical scheme built from procedures provided by grounded theory should not be understood as a quantitative test. The main concern in this step is to verify if the elements that formed the theory effectively reflect what was said in the interviews, in other words, if the abstractions conceived by the researcher are in accordance with the raw data.

According to Strauss and Corbin (2008), there are several ways to proceed with the validation of the theoretical scheme. One of the ways is returning to the raw data in order to compare to the elements of the scheme. In this validation modality, it is necessary to perform a rigorous comparative analysis in order to verify if the theoretical scheme is able to explain the reality of the cases researched.

Another way of validation mentioned by Strauss and Corbin (2008) is hearing the research participants again in order to present the scheme built, its categories, properties, dimensions and relationships. The participants should see themselves inside the scheme. In a general way, they should be able to identify themselves with the information that are in the scheme.

As mentioned previously, the theoretical scheme built in this study was validated using both ways mentioned by Strauss and Corbin (2008). After the conclusion of the fourth theoretical sampling step, the rereading of all interviews performed was started, in order to conduct a high-level comparative analysis. This is a step that demands a lot of time and attention.

The other validation step of the scheme was made returning to the first cases. Because of some difficulties in scheduling interviews, company B did not participate. In this way, the validation was performed with cases A and C. The final theoretical scheme was presented and the researcher “told the story” of the research. This validation phase was considered satisfactory. It was noticed that managers were able to see themselves inside the substantive theory. Some considerations were made, mainly, regarding to the dimensions, however, this can be considered normal, since the model is a reduction of reality and it would be impossible to present some aspects in a more detailed way.

Research limitations

The study undertaken cannot be considered a complete theoretical approach, since the empirical research was focused on a substantive reality. As qualitative research characteristics, studies that use grounded theory do not present generalizable results. The results presented will not necessarily reflect the reality of other contexts.

Related to the data collection process, it is important to emphasize that there were situations in which the “ideal cases” could not be investigated, mainly because of the difficulty to access the companies. Due the fact that the cases are defined according to data, the researcher did not have time to convince managers to participate. The time is a very important factor, and the researcher should deal with it in order to avoid extending the time planned to finish the research.

The use of interview as a main source of data can be considered a limitation, because in some occasions the speech of the interviewee can be in disagreement with the reality. In this particular point, it is important to notice that the researcher behavior may be crucial to obtain reliable data. The researcher should concern about establishing a reliable relation with the interviewee and never influencing on the answers.