When a business wants to understand the functionality of its system and structure, it researches to gather information using statistical procedures. “Research is a discerning pursuit of truth” Hair, Wolfinbarger, Money, Samouel, and Page (2015). The research onion developed by Saunders in 2007 is an essential tool for describing the research method (Saunders, Lewis, & Thornhill, 2015). Business research has several interrelated components that seek to predict and explain phenomena in business by gathering, analyzing, interpreting, and reporting information to inform decision making. A functional business research studies a wide range of factors including people, systems, and how they interact. Business research can be formal or informal, but it should be replicable, and the benefits should outweigh the cost (Hair et al., 2015). Depending on the factors motivating business research, the research can either be applied or primary business research. As such, this paper investigates the methodology used in business research with a case study of De Witte et al. (2010) who investigated the association of employee’s perception of quantitative and qualitative job insecurity with job satisfaction, and psychological distress in the Belgium banking sector.
De Witte et al. (2010) sampled fifteen thousand employees from a population of 69,000 bank employees in Belgium representing nearly 21% of the total employees. A sample size that is reliable and valid should allow the researcher to generalize the findings of research from the sample of the population being examined (Sekaram & Bougie, 2016). Therefore, the sample size should be a reliable estimate that closely reflects the population parameters with minimal error. Mostly, no sample size can be larger than the sample population, regardless of the probability sampling technique. Typically, the sample size is a function of the variability in the population, the precision needed, confidence level, and the sampling technique used.
The extent of precision and confidence desired by the research determine the sample size. However, a population size that is too large or too small is a problem and may lead to Type II errors implying that the researcher may accept the findings of the study, when in fact the outcome should be rejected (Sekaram & Bougie, 2016). That is, a sample size that is too large may reach significance levels leading the researchers to believe that the significant relationship in the sample is true of the population when the might not be in reality. Therefore, there is no sample size too large or too small that helps a research project. Efficiency is achieved when a sample size can be reduced or increase for a given level of precision.
Sekaram and Bougie (2016) indicate that the rule of thumb as developed by Roscoe in 1975 proposes that:
- A sample size that is larger than 30 and less than 500 are appropriate for most studies.
- Samples that are subdivided require a minimum of 30 for each of the subsamples (senior/junior, female/male, et cetera).
- Multivariate research requires a sample size that is several times as large as the number of variables.
- Simple experimental research with controls can achieve success with a sample size that is only 10 to 20 in size.
Other factors influencing the choice of the sample size include the absolute or relative sample size, time and cost, non-response, heterogeneity of population and kind of analysis (Bryman & Bell, 2015). In research, it is the absolute size of a sample that is important and not the relative size. Typically, the larger the samples size, the higher the precision up to a sample size of nearly 1000. Beyond the 1000 mark, the level of precision slows down and plateaus which makes time and cost a matter of less concern. Of importance to our case study is the issue of heterogeneity of the population. Heterogeneous samples are highly varied as compared to homogenous samples; therefore, “the greater the heterogeneity of a population, the large the sample will need to be” Bryman and Bell (2015).
In consideration of the factors discussed thus far, the sample size of our case study (15000) was necessary. The cross-sectional study sampling 63 banks across the country were largely heterogeneous thus requiring a large sample. The research is also multivariate and therefore requires a large sample according to the thumb rule. Although the large sample size may affect precision, the findings can be reliable and can be generalized.
The respondents in De Witte et al. (2010) were selected randomly without any particular stratum. In other words, the study used a simple random sampling technique. A common example is drawing raffle tickets from a container. If all the tickets have the same size and texture, stirring the tickets in the container completely gives each raffle ticket an equal chance at getting drawn. Therefore, if a sample size of fifty is needed, then the process of selection must be repeated 49 times after the first withdrawal. Random sampling is the simplest sampling method because there is only one stage in selecting the sample (Zikmund, Babin, Carr, & Griffin, 2013). This type of sampling design best applies to the generalizability of findings of an entire population (Sekaram & Bougie, 2016). However, Simple random sampling may not be the best if the research budget is tight and the resources are limited while the number of subjects is vast or is dispersed geographically; this would make it expensive. The issue of cost and generalizability are of importance to the consideration simple random sampling.
The process of random sampling is almost rid of human bias in research. Selection of people to interview for a job posting is not made on the merits of looking friendly or approachable; the selection is completely mechanical (Bryman & Bell, 2015). Moreover, the employees do not have to be available in the workplace for them to be interviewed the process is not dependent on their presence and can be conducted away from the interview. Selection is made without the interviewees’ knowledge since they only become aware of their selection when they are contacted with the news.
Measure of Variables
The merits of measure can be examined using some ways that are derived to represent concepts of social science. Nonetheless, discussions on validity and reliability of measures have the potential to mislead, and it would think that all new measures of the concept are going through rigorous scrutiny to ensure that validity and reliability of the measure. Most measurements are typically asserted, straightforward but with little testing to ensure reliability and validity. Such as investigating face validity and internal reliability when multiple-indicator measures have been derived. However, many cases of concept measurement make no further testing yet validity and reliability are related to the fact that validity precedes reliability implying that an unreliable measure is also not valid. The measure should not fluctuate; if it does then, it can be reliable and valid because it might contain different variables on different occasions. A lack of internal reliability implies that a multiple indicator measures cannot be valid because it measures two or more distinct elements. Furthermore, if the internal observation is not consistent, it means that the observers are unable to agree on their observation which makes the measure invalid.
The case study of De Witte et al. (2010) measured four variables, namely: quantitative job insecurity, qualitative job insecurity, job satisfaction, psychological distress, and the control variables. Each variable was measured using an approach unique to the variable. The validity and reliability of the measures are dependent on the ability of the approaches to providing answers to the relative research question. A pilot study is conducted to the test the validity and reliability of a questionnaire before it can be trusted in conducting the survey. The design of the questionnaire should state a clear introduction and survey purpose (Hair, Wolfinbarger, Money, Samouel, & Page, 2015). The measures in the case study had been tested and used in other related studies. For instance, the four items measure for quantitative job security variable had been used by De Witte in 2010; the qualitative job security measure had been developed by in 1989 by Ashford, Lee, and Bobko (Hair et al., 2015). Therefore, the study can be replicated and generalized.
Collection of Data on Social Demographics
The collection of demographic data in the survey is an important exercise that indicates important personal information about the respondent that helps the researcher to categorize the sample size according to the age difference, sex, educational level, income status, and occupational position. Demographics are characteristics of the population. The purpose if the demographic in the survey design is to allow the researcher to assess who should be included in the survey and how to delineate the survey response data into comprehensive participant groups (Bryman & Bell, 2015).
The decision on who should be surveyed is influenced by the main topic of the study. For instance, in the case study, the researcher may determine that only respondents within a particular age limit or income status will be surveyed. Or the researcher may decide to narrow down to respondents with a particular level of education or occupation position. The demographic data gives a clear-cut direction to determine who will participate in the survey.
Upon completion of the survey, the data can be divided into categories of data regarding the demographic information (Saunders, Lewis, & Thornhill, 2015). Again, using the example of the case study, the researcher can decide to cluster the responses from an individual with secondary education, or those who have a tertiary level of education. De Witte and his colleagues can also decide to analyze data in the cross-tabulation form to compare and contrast the survey data across demographics.
Though the researcher may be tempted to ask multiple demographic questions, too many may not auger well with the respondents. The respondents may feel aggravated creating concern about the collected data. Moreover, the participants may feel that the demographic questions are invasive on their privacy and confidentiality. It is important for a researcher to decide which demographics to include and which to omit. The choice of the demographic questions provides meaningful results to the study that may assist in decision-making. On the contrary, if the respondent feels threatened, they may become antagonistic and give inaccurate information in the survey.
The research design describes the research process. The research design is a framework ha describes the considerations that were made in deciding the appropriate methodology for the study, how the research participants were selected, and the process of data analysis (Bryman & Bell, 2015). Several research designs exist such as descriptive, exploratory, and explanatory. De Witte et al. (2010) used a descriptive research design and subtype cross-sectional survey. The descriptive research design functions to present the experiences of the respondents (Saunders, Lewis, & Thornhill, 2015). According to Bryman and Bell (2015), the descriptive design relates closely to ethnographic study, but in the descriptive design a quantitative framework is facilitated; for instance, the demographics of the sample population are reported. An explanatory design focuses on explaining the demographics of the participants effectively enabling the researcher to establish the influence of variables. An exploratory design, on the other hand, explores the concerns of the study before the survey is conducted and is used to inform the areas of further research.
The research method describes the approaches were taken by the researcher to collect and analyze data for discussion and interpretation. The type of research method to be employed in h study depends on the topic and objectives of the research. First, the researcher identifies the population to be studied and uses sampling techniques to select a sample size that will represent the entire population. A crossectional study typically requires a large sample size which is necessary if the research is to be generalized or replicated. The variable measure, data collection tools are developed by the researcher to answer the research questions. In this study, the stages in the methodology have been described with the help of the research onion.
Bryman, A., & Bell, E. (2015). Business research methods. USA: Oxford University Press.
De Witte, H., De Cuyper, N., Handaja, Y., Sverke, M., Naswall, K., & Hellgren, J. (2010). Associations between quantitative and qualitative job insecurity and well-being: A test in Belgian banks. International Studies of Management & Organization, 40(1), 40-56.
Hair, J. F., Wolfinbarger, M., Money, A. H., Samouel, P., & Page, M. J. (2015). Essentials of Business Research Methods (2nd ed.). New York: Routledge.
Saunders, M., Lewis, P., & Thornhill, A. (2015). Resach Methods for Students (7 ed.). Pearson.
Sekaram , U., & Bougie, R. (2016). Research methods for business: Askill building approach. United Kingdom: JOhn Wiley & Sons Ltd.
Zikmund, W. G., Babin, B. J., Carr, J. C., & Griffin, M. (2013). Business Research Methods. Mason, OH: South-Western Cengage Learning.