A research design is the framework which researchers plan for use as a guide in collecting, analyzing, and completing their research. The function of a research design is to ensure that the evidence obtained enables us to answer the research questions in a convincing way. When little is known about the research setting, as in the case of Egypt, where the present study is set, exploratory research can be seen as the foundation for a good study. This kind of research can help, for example, to determine the practical problems of carrying out the study, establish priorities for the following stage, operationalize the study constructs, and develop hypotheses for more precise investigation. Using data from 305 questionnaires as well as 36 personal interviews concerning the antecedents and outcomes of strategic decision-making, I will address in this case the role of the two-stage multi-method design in deciding several methodological issues. These issues concern both sampling design (namely, the target population, type of sample, and relevant respondents) and questionnaire development (namely, the degree of disguise, questionnaire length, translation, order of questions, the number of Likert-type scale items, and administering the questionnaire). The conclusion of this case is discussed and the forthcoming cases are highlighted.
By the end of this case, students should be able to
- Understand the role of the two-stage multi-method design
- Use the two-stage multi-method design to better understand new research contexts
- Comprehend how the two-stage multi-method design can be used to make important methodological decisions
- Identify the challenges associated with conducting research in new settings
Project Overview and Context
Research design addresses a number of issues such as methods of data collection, the time dimension (whether a cross-sectional or longitudinal study), the topic scope (breadth and depth) of the study, sampling issues, time and cost constraints, and the degree to which the research questions have been crystallized (whether an exploratory or a formal study). Research designs can be classified according to which of two possible approaches is used to gather primary data. These are (a) observation of conditions, behavior, events, people, or processes and (b) communication with people about a variety of issues (Cooper & Schindler, 2013). Although strategic decision-making may be investigated on the basis of observation techniques and organizational records, my study mainly followed the communication approach through both questionnaires and personal interviews; while observation technique, was relatively little used.
The present study can be classified as a multi-method field study. The data for it were collected using a range of techniques (i.e., interview, questionnaire, and observation), through two stages. The data collection in the first stage lasted approximately 3 months from mid-October 2001 to mid-January 2002. In this stage, 128 questionnaires were collected: 66 questionnaires from the private business sector and 62 from the public business sector. In addition, 36 semi-structured interviews were conducted. These may be divided into 21 interviews in private business sector companies and 15 in public business sector companies. Finally, using observation technique, I attended a single 4-hr meeting of the board of directors of a construction private company.
To collect data for quantitative analysis in the second stage of this study, a survey instrument was dropped off for completion by the executives of a sample of 400 manufacturing privately owned companies, all employing more than 100 people in greater Cairo. The data collection lasted approximately 2 months from mid-July 2002 to mid-September 2002. From the 400 dropped-off questionnaires, 206 replies were collected (an initial response rate of 52%). Of these 206 questionnaires, 37 were disqualified for being blank or largely incomplete, submitted by ineligible respondents, or collected from state-owned companies. The remaining 169 usable questionnaires (at a final response rate of 42%) were included in the final analysis.
Justification for Using the Two-Stage Multi-Method Design
The present study adopted the two-stage multi-method design. With this approach, exploratory study becomes a separate first stage. The reason why the research adopted a two-stage design is as follows. As the nature of the study depends on the level of knowledge in the research area, when not much is known about the situation in hand, an exploratory study should be made before the rest of the research (Iacobucci & Churchill, 2015). In such cases, extensive preliminary work needs to be done to get sufficient understanding of the research problem before we can proceed to the stage of testing hypotheses (Sekaran & Bougie, 2016). In this preliminary stage, as we explore in the new setting, the potential and limitations of such techniques as interviews, questionnaire, and observation (which may vary from those in well-established research settings such as the United States) can further help to enhance our understanding of research concepts and settings.
Overall, given the contradictory conclusions of previous research on strategic decision-making, the effect of context and the absence or paucity of reported investigations of strategic decision-making in an Egyptian setting hitherto, it seemed that an exploratory use of data collection techniques was needed as the foundation for hypothesis testing. Following the results of the exploratory study (i.e., the first stage), the investigation was extended to the testing of hypotheses in the second stage. More specifically, it was decided to use the first stage for the three following purposes:
- Determining the practical methodological problems of carrying out the first stage (sampling and developing a questionnaire to design the second stage better)
- Clarifying concepts and developing measures; in other words, using the knowledge I got from the first stage to develop an appropriate operationalization for the study constructs in the second stage
- Contributing to the development of hypotheses, enhancing the interpretation of findings, and enriching the discussion of the research limitations and directions for future research.
The present case discusses the first of the above purposes, namely, sampling and questionnaire development; it is hoped that the two remaining purposes will be addressed as future case studies. Next, I discuss how the first stage, along with its multi-method data techniques, contributed in the second stage to make several important decisions on sampling and questionnaire development.
Selecting a sample is a fundamental aspect of study design. Some issues which the first stage of this project helped me to resolve on before conducting the second stage were the target population, sample type, and relevant respondents. These are discussed in turn below.
The Target Population
The first stage was of a great help in defining the appropriate target population for this study as “the Egyptian private manufacturing companies working in greater Cairo and employing more than 100 people.” As explained below, two aspects of this population were informed by taking the results of the first stage into account, namely, the targeted sector and region.
The Targeted Sector
I excluded state-owned manufacturing companies from the second stage because my interviews in the first stage revealed that top management teams in these companies were in many cases forbidden to take strategic decisions without getting approval from their holding companies. In other cases, top management teams were taking some strategic decisions merely in response to governmental policies. For example, as an important part of the privatization program early 2000, Egypt has used privatization proceeds to offer early retirement to workers in some state-owned companies to facilitate the process of their privatization; consequently, the Ministry of Public Enterprise and holding companies took a very strong grip on these companies to ease this process. The following statement made by the Minister of Public Enterprise may clarify this conclusion.
In my capacity as Minister of Public Enterprise, it is my duty to strive to achieve the objectives of the Ministry, namely, to contribute to the economic reform process by shifting the ownership of State-Owned-Enterprises (SOEs), which we call Public Enterprises, to the private sector … Our portfolio of companies is one of the most challenging and exciting to manage, and encompasses 190 affiliates operating in 42 industry and service sectors organized under 10 Holding Companies. (http://www.mpe-egypt.gov.eg/ministers_message.asp)
Given the above, the decision to offer early retirement, which was one of the most frequently taken decisions in the sample of public companies that was found in the first stage, was only an implementation of the governmental policies related to the privatization program.
The Targeted Region
In the second stage, the study limited its population to companies working in one region of Egypt, namely, greater Cairo. There were two reasons behind this decision. First, greater Cairo is the center of business and economic activity in Egypt. Second, this decision helped to reduce costs in money and time.
Type of Sample
Although probability sampling has some advantages over non-probability, I chose the latter, for some practical reasons which I discovered during the field study in the first stage. One important reason is that a random sample would have required published data on companies to constitute a sampling frame, but such data were not available in Egypt. One of the factors which increased this difficulty in the present study was the problems faced by the Egyptian economy during the field study period, particularly after 11 September 2001. Therefore, it was necessary to use a non-probability sample in the present study (Elbanna, Abdelzaher, & Ramadan, 2017; Zahra, 2011). More specifically, I used purposive/judgmental sampling by strictly following several criteria in picking companies to be included in the study such as company size, location, type of activity, and so on. Although this helped to approach the characteristics of random sampling, the study sample could easily be biased, not truly random. In other words, compared with random sampling, the study sample does not ensure that the estimates derived will be representative of the entire population. However, it is still a better option than a purely convenience sample.
In view of what I learned in the first stage and read in related research (e.g., Amason, 1996; Bernard, 2017), I decided to select respondents in the second stage for their competence rather than for their positions alone. This is because it became clear during the course of the first stage that good respondents were managers who actively participated in making the strategic decisions in question, who understood the information I needed, whom I could talk to, and who were glad to provide relevant information.
Although the use of multiple informants can help to minimize the effect of any individual perspective, the difficulty of multiple-respondent surveys in the first stage obliged me to use a single respondent for each decision in the second stage. The first stage showed that it was impractical to collect data from two respondents in each company, even though doing so is highly recommended for minimizing response bias. For example, in the first stage, I was able to collect data from two respondents in 11 cases only, while in more than 100 cases, I succeeded in collecting data from one respondent only. Hence, because of response rate considerations, I decided to use a single respondent from each company in the course of the second stage of this study.
The Questionnaire Development
The first stage of this research project helped me to decide on several aspects of developing the questionnaire for use in the second stage. These include the degree of disguise, questionnaire length, translation, order of questions, and administering the questionnaire. I will take these issues in turn.
Degree of Disguise
Disguise concerns the amount of knowledge about a study, for example, the purpose and sponsor of the study, which is communicated to the respondents. Researchers do not disclose the sponsor and the objective of the study if they believe that the respondents would respond differently if both or either were known (Cooper & Schindler, 2013). For this reason, I used an undisguised questionnaire in the first stage, whereas in the second stage, the questionnaire was partially undisguised; that is, the purpose of the study was obvious to respondents, but the name of the university affiliation mentioned was changed from the University of Birmingham (a British university) to Cairo University (an Egyptian university). The reason for this change was that introducing the University of Birmingham as my affiliation and using its formal document as a cover letter would have led to serious negative consequences.
For example, many respondents were sensitive to the purpose of the study, to the extent that some of them thought the study threatened the national security of Egypt and others went so far as to look very doubtfully at me. These concerns were clearer in public sector organizations than in firms in the private sector. As an outcome of this problem, several respondents refused to cooperate with me, while others gave no answer to several questions, especially those which could have helped to identify the respondents (e.g., position, age) or the organizations (e.g., number of employees, type of activity). Similarly, several respondents were reluctant to answer the questions assessing their perceptions of several aspects of the political, economic, and business environment in Egypt. This may help to explain the increase in the percentage of questionnaires with missing data in the first stage. Given the above, I decided to use a partially disguised questionnaire in the second stage to encourage respondents to answer candidly and be more cooperative.
This section shows how the experience gained from the first stage helped to overcome one of the practical problems of conducting this study, that is, the length of the questionnaire.
In the course of the first stage, many respondents complained about the length of the questionnaire (12 pages) and consequently the time they needed to complete it (approximately 45 min). This problem led to a decreasing response rate and increased incidence of missing data, especially in the last sections of the questionnaire. Bryman (2012) explains this by arguing that long questionnaires are rarely feasible because of the possibility of respondent fatigue. This being so, because I had also to think about cost in time and money and the method intended for collecting data during the second stage (dropping-off and picking-up), I excluded several variables from the conceptual model of the study in the second stage. As explained below, these variables included, for example, some demographic data (e.g., education qualification, education degree, age, and work experience), external control, slack in resources, and conflict.
Although demographic data can in Western settings be easily obtained through archival sources or through an easy-to-complete questionnaire, the findings of the first stage showed that the data on upper echelons in society are not publicly available in the Egyptian setting. Furthermore, Egyptian respondents were not happy to provide information on their demographic characteristics. This may be due to the impact of the culture and the economic difficulties in Egypt during the period of this study. Overall, one can argue that people in Egypt appear to be sensitive about supplying researchers with demographic data. Given the above in conjunction with the problem of the length of the questionnaire, I eliminated the upper echelon perspective from the study model in the second stage. This was a serious limitation of the study that I was unable to avoid.
Another variable which was excluded from the second stage was external control. External control in this study was operationalized by assessing the effect of several external parties on the strategic decision-making process dimensions. These were governmental agencies, suppliers, customers, labor unions, competitors, banks, and finance agencies. The correlation analysis in the first stage showed no significant association between external control and the strategic decision-making process dimensions: multiple regression analysis confirmed this result. Moreover, this variable did not moderate the relationship between the strategic decision-making process dimensions and decision effectiveness. Overall, the analysis of the role of each of the external control parties in strategic decision-making provided further support for the previous results. Given the insignificant role of external control as shown by the above results together with the need to reduce the questionnaire length, I decided to remove external control from the study model and questionnaire used in the second stage.
Concerning the slack in resources, it was found to be highly correlated with performance (r = 0.50). One can argue that the higher the performance, the higher the slack in resources. In other words, performance can be used as a proxy of slack in resources. Finally, the first stage showed that Egyptian managers were unable to differentiate between the two aspects of conflict included in the initial conceptual model, namely, constructive and personal conflict, and hence this variable could also be used to reduce the questionnaire length. Given the above, these two variables (namely, slack resources and conflict) were removed from the study model used in the second stage.
Other tactics were used to tackle the problem of a long questionnaire, such as using a structured questionnaire, the drop-off and pick-up approach, and packaging questions whenever possible. Moreover, the scale of environmental uncertainty was shortened from 35 items in the first stage to 21 items in the second stage to further contribute to reducing the questionnaire length (Elbanna & Gherib, 2012). The above actions helped to reduce the length of the questionnaire from 12 pages in the first stage to 8 pages in the second stage. Removing the above variables from the conceptual model in the second stage does not mean that they are of no significance in the strategic decision-making process. Instead, it implies that the practical problems of carrying out this study in the first stage led to establishing the overall priorities for the research in its second stage for hypotheses testing purposes.
The issues of research design and their importance differ from one study to another depending on its nature. For example, as in this study, translation equivalence is an important part of the design because the study is conducted in a different language (i.e., Arabic) from that in which the questionnaire was originally developed (i.e., English).
The first stage of this study showed the importance of accurate translation, taking into account Arab culture and its impact on the way in which respondents understood the study questions. In some cases, this understanding was different from that of Western managers which sometimes led to major concerns. For example, the experience of the first stage suggested that the translation of questions measuring intuition, which were originally developed by Khatri and Ng (2000) in the U.S. setting, was a cause for concern. Some interviewees, for instance, saw the question on the use of past experience in making selected decisions as if it was meant to assess the use of past experience in making strategic decisions in general. Others saw the question as a test of whether they had enough experience or not. Bearing in mind the above, I gave close attention to the translation of the questionnaire in the second stage as explained below.
I originally drafted the questionnaire in English and later translated it into Arabic. Five academic staff bilingual in Arabic and English reviewed both the Arabic and English versions of the questionnaire to ensure that the translation was equivalent. During these reviews, I tried to work closely with reviewers, to ensure that they fully understood the questionnaire items. The resulting Arabic questionnaire could be called as close in meaning to the original English version as possible, while still using Arabic that was as natural as possible and as intelligible as possible to the target respondents.
Order of Questions
The sequence of questions in the questionnaire is an important point in its design. Here, there are no rigid principles but only “rules-of-thumb” to guide the researchers. Taking into account the feedback I received from the first stage, I did two things to help respondents to move smoothly from one question to another. First, to prevent the informants from getting bored and to help them complete the questionnaire quickly, I packaged the questions whenever possible. Second, to help respondents move from question to another in a logical sequence, I divided the questionnaire into 11 sections and grouped similar questions in sequences. The first seven questions dealt with themes related to decisions (i.e., decision process dimensions, decision-specific characteristics, and decision effectiveness). Next, the 8th question handles company performance. The 9th and 10th questions address environmental characteristics (i.e., environmental uncertainty and hostility), and the questionnaire ends with such sensitive questions as the respondent’s position and the number of employees in the organization.
Number of Likert-Type Scale Items
I had included both 5- and 7-point Likert-type scale questions in the first stage, but this confused some respondents. Therefore, in the second stage, I unified the number of scale items, using a 7-point scale throughout the questionnaire. Moreover, the greater number of options seemed better because it increases the opportunity to get a sensitive measurement and extract variance (Cooper & Schindler, 2013).
Method of Administering the Questionnaire
The main methods of administering a questionnaire are mail, phone, and personal interview, none of which is universally suitable. Given the exploratory nature of the first stage, I used different data methods to administer questionnaires. For example, personal interview was used in the private business sector, whereas personal interview, drop-off and pick-up, and giving questionnaires to a group together were used in the public business sector. In the course of the second stage, the questionnaires were collected by means of the drop-off and pick-up technique. The reasons behind this choice can be listed as follows.
- The drop-off and pick-up technique has been used successfully in Egypt (e.g., Elbanna, 2008). For example, the response rate when taking a convenience sample from 250 Egyptian owner-directors of small business companies using this technique was almost 62% (Abou Aish, 2001).
- This approach may increase readers’ confidence that the proper people are completing the questionnaires, which is particularly needed in developing countries such as Egypt.
- Because the questionnaire is long, requiring much time and effort from informants to complete, the personal delivery and pick-up technique seems to be the most appropriate method of collecting this kind of questionnaires (Cooper & Schindler, 2013). For example, my experience in the first stage showed that with personal delivery, I could increase the response rate by encouraging respondents to complete the questionnaire, stressing the value of their participation and the ease of completion, and then arranging a return visit to collect the completed questionnaires.
This case sought to show researchers, particularly novice ones, how the two-stage multi-method design was used to enhance the study design of a research project. The case explained how and why I designed a study that would be divided into two stages and would use different data collection methods. The first stage with its multi data collection methods helped in the second stage to make several methodological advances concerning questionnaire development and sampling design, by resolving several practical methodological problems raised in the first stage.
Having addressed the above issues in the current case, I will discuss other important issues in forthcoming cases. These include, for example, how the two-stage multi-method design greatly contributed to clarifying concepts, operationalizing the study variables, developing hypotheses, enhancing the interpretation of the findings, and enriching the discussion of study limitations and directions for future research.
To end, the two-stage multi-method design has played a significant role in enhancing both the essence and methodology of this research project, to the extent that it led to publication in several international journals, including top-tier journals such as Journal of Management Studies and Strategic Management Journal (i.e., Elbanna & Child, 2007a, 2007b).
Exercises and Discussion Questions
- Why does the two-stage multi-method design matter in the new settings being studied?
- What do you think are the challenges of using the two-stage multi-method design, and what might this mean for researchers?
- Identify three papers which have some serious methodological and conceptual limitations and show how the two-stage multi-method design might help the researchers to overcome such limitations if they had the chance to conduct their research again.
- Suppose that you have been asked to study how managers in Palestine make important decisions. As you know little about this subject in this less researched setting, namely, Palestine, how will you design your study to answer your research questions? If you conduct the same study in the United Kingdom or China, will you design your research differently? How? Why?
- If you are selected to get involved in a new research project examining the determinants of strategic decision-making success in Egyptian manufacturing and tourism firms, will you adopt the two-stage multi-method design in this research project? Why? Is it possible to adopt different research designs in both industries, namely, manufacturing and tourism? Why?