The Essay On Job Characteristics' Impact On Stress & Well-being At Belgian Banks.

Sample size

Job characteristics are believed to have an impact on stress and well-being at work (Karasek & Theorell, 1990). The demands of the job on the one hand and the extent to which you have control over your own activities (decision latitude) on the other, are two factors which together define how stressful a job is.  Those jobs which are high demand, but offer limited control, are considered to be high-strain and carry an increased risk of job dissatisfaction, stress and burnout.

Based on this theoretical framework, the Union of Belgian Banks sent out a research call to several institutions, with a bidding process based on criteria such as quality of the proposal, timing, and – above all – budget. The aim of the research was to carry out quantitative research to investigate the relationship between job characteristics and job satisfaction in all Belgian banks at individual level. But in order to do this effectively, several methodological issues needed to be resolved during the research process.

First of all, a research consortium was selected to conduct the research, or more precisely, the two highest ranked bidders were asked to jointly undertake the research. The two competing research institutes, a private company specialising in stress at work and the Katholieke Universiteit Leuven (Belgium), were required to co-operate and develop a level of trust in order to conduct the research. For example, both research institutes had different ideas as to which scale should be used in the questionnaire.  They could not just combine the scales or include both scales, simply because they are supposed to measure the same concept. Furthermore, this would also make the questionnaire too complex. Therefore, the research institutes had to combine their knowledge, look for compromises and jointly work on a shared vision, which is, to say the least, rather time consuming.

A second obstacle that needed to be overcome was the sample (see Chapter 7).  In total, 69,000 employees work for Belgian banks and it was decided that questioning all employees would be too complicated and too expensive. Therefore the research committee, consisting of representatives of the banks, the unions and the research consortium, opted for a cross-sectional design (p. 45) with a fixed sample of 15,000 employees (roughly 21%; see p. 187 ‘Absolute and relative sample size’).

Not all banks participated in the survey. First, participating banks were selected randomly. Then within each participating bank, the respondents were selected at random with no particular quota for gender, age or employee level. In the postal survey (see p.231 ‘Self-completion questionnaire or postal questionnaire’) several steps were taken to improve the response rate (see p. 234 ‘Steps to improve response rates to postal questionnaire’; see also suggestions by Dillman, 1983). The survey was based on addresses which had been provided by the banks (name, language, address) and each employee randomly selected received a personalized envelope through regular mail, sent to him/her by the employer. The completed questionnaire needed to be returned (free of charge) through the internal post within each bank.

Sampling method

There are various logistical problems.  The researchers had to travel to each bank to collect the completed questionnaires and due to the fact that in Belgium, part of the population speak Flemish (Dutch) and part speak French, two versions of the questionnaire needed to be available and then carefully translated and tested for the accuracy of their translation (see also Tips and skills, p.488: ‘Translating interview data’).  The questionnaires were sent to the respondents’ home addresses, a French version if the respondent lived in the French part of Belgium, a Flemish version when living in Flanders.  This prompted a series of angry calls when Flemish people, living in the French part, or vice versa, received a questionnaire that was not in their native language.  Furthermore, Brussels is officially bi-lingual and, to complicate matters even more, contains many headquarters in which the main language spoken is ... English! In order to minimise attrition, it was important that these respondents received a questionnaire in their preferred language. Another logistical issue was the co-ordination and control of the distributed information.  The Belgian banks, who were the research financers, chose a decentralised way of working, hence organizing a ‘sensibilization campaign’ within each bank whereby the researchers had to visit all the banks to explain the theoretical framework and the outline of the research to representatives of both employers and employees.

Additional initiatives to prompt a higher response rate were taken up by individual banks, or, more precisely, by some of the banks.  The researchers were required to carefully follow-up on those initiatives implemented by the banks, to ensure that these initiatives remained both neutral and valid for the research.  Some of these initiatives proved difficult to deal with due to the selective use of information that had been employed (e.g. letters forcing the employees to participate; or union campaigns to guide certain answers). Hence, the researchers had to be sensitive for the respective organisational cultures, while making sure they kept a neutral position towards all partners involved in the research.  

Once the data collection was completed (response rate of 47.6%), the data handling needed much attention. A comprehensive check and double-check was conducted on wrong entries, filters, missing cells ... just to increase the reliability.  One issue was the major difference in response rate between the banks.  Due to a strong campaign, some banks reached a response rate of over 60%, whilst others barely reached 20% because they did nothing to increase the response rate.

The sample size for this study is fifteen thousand employees selected from a total of sixty-nine thousand bank employees (about 21% of the employees). Discuss the advantages and disadvantages of having a sample of this size. What factors should be considered in decision on sample size?

To find the relationship between job characteristics and job satisfaction at employee individual level, this study randomly selected from employees from each of the participating banks. What is the current method of sampling? What are the advantages and disadvantages of the current sampling method?

One of the shortcomings of the current sampling method is the high cost of data collection. What are your suggestions to improve the sampling methods?

In this study a Self-completed questionnaire was created and send out to the respondents. Discuss some of the problems in the process of data collection and how to address them in future study

In any quantitative study we want to believe that the sample selected is a representative sample of the true population. One of the methods to check representativeness is to use Secondary Data. What secondary dataset can be used to check the representativeness of the sample and how can it be used?

Large sample sizes tend to be appropriate since they give more accurate statistics. (Perneger et al., 2015) Urges that when a large sample size is used the mean values and outliers are easily pointed out. This implies that determining the mean allows the researcher to point out the outliers more easily. The main reason being that as the sample sizes increases, the statistics obtained approaches the population parameters which makes it easier for the researcher to pinpoint the anomalies in the data (Perneger et al., 2015). The outliers are points that strongly differ with the sample point estimate; mean. In particular, large sample sizes improve the confidence level, the margin of error, power and effect sizes. As the sample sizes increases, the information about the population increases, which in turn reduces uncertainty. That is the interval where the true estimate of the population becomes more precise. Thus, in terms of the confidence interval, three main things happen when a large sample size is used; an increase in confidence, precision and a decrease in uncertainty.

The larger the sample size, the larger the power (Perneger et al., 2015). That is, when a large sample is used for inferential purposes, the test is more likely to detect differences. A large sample has a shorter margin of errors, which makes the error bars shorter. Therefore, when testing whether two or more means differs, a large sample is more likely to detect the difference. Thus, the ability of a test to determine whether there is a significance difference is what is considered as the power, and the larger the sample, the higher the power. Also, the sample size plays an important role in the effect size, where to detect the difference between observed groups a small effect is required, a small effect requires a large sample size.

Nonetheless, there are some disadvantages of using a large sample size like the one used in the given research. First, a large sample increases false positive results. That is, it exaggerates the significance values (p-values). This makes the insignificant results significant (Perneger et al., 2015). Therefore, using a large sample size, it might be difficult to detect a difference between the samples. A large sample size is sensitive to very small, the inconsequential departure of data from the distribution. For instance, chi-square goodness of fit may lead to the assumption that the data do not follow a given distribution due to a slight departure, but when a small sample is used, the test shows that the data follow a given distribution (Perneger et al., 2015).

There are three main factors that determine the sample size to be used in a research. First is the margin of error. No sample collected if perfect, since there are sampling errors like non-random sampling errors, too small samples among others. The first factor that helps in determining the sample size is the margin of error or the confidence interval. This indicates that allowance the researcher is willing to take where the sample mean lies between. Second, is the confidence level, which is determined before carrying the research. This is the level of confidence the researcher wants to be with the actual mean interval. Lastly but not least is the standard deviation, which shows the expected variation from the actual parameters.

Sampling method

The research used cluster sampling techniques. First, clusters were formed using the branches of the Belgium banks as the sampling frame. This was a one stage sampling method since all the participants were first categorized in accordance with their geographical locations where the banks are located.  A sample of banks was randomly selected. Then a sample of participants was selected randomly from each cluster. The main objective of cluster sampling is to reduce the cost of sampling and increasing efficiency (Smith et al., 2010). In this case, all the banks make the sampling frame, from which a sample of representative banks is randomly selected. The aspect of randomization is important in research since it helps in the following ways. First, in the experimental research, randomization help in achieving accurate results. This is because the researcher is able to measure the effect size. Also, it reduces cases of lurking or confounding variables, since the assignment is due to chance.

This sampling technique is also referred as area/geographical sampling. This design has various advantages. First, it is time-efficient and cost-efficient relative to large geographical location. It would be economical to select samples in a population, rather than selecting the units randomly from a pool of all workers from all banks (O'Leary, 2013). Therefore, selecting at bank level is the most appropriate approach. The researcher also, using this technique has the freedom of accumulation of large samples. Practically, cluster sampling is easy to apply. The research design enables the researcher to accumulate large samples. For instance, in the case study, a sample of 15,000 employees, is quite a large sample. Cluster sampling increases the precision per individual since there is compensation for by the fact that the researcher selects a large sample at the same cost. The technique enjoys the advantages of stratified sampling as well as those of simple random sampling. Through this, the method is more powerful. Last but not least, the technique allows the collection of information from one or more areas at the same time.

On the other hand, this sampling technique has some limitations or cons. In the clusters the units selected might be similar, leading to large sampling error. This reduces the representativeness of the selected sample (O'Leary, 2013). When different sample sizes are selected from each cluster, there are increased chances of sample bias. The findings obtained using this technique are sometimes impossible to generalize, or applicable to other areas. This is because the sample obtained may fail to show the diversity in the sampling frame. Also, the group level information is required for this technique to be successful.

The cost of data collection should be manageable and significantly lower than carrying out a census. The main reason for taking sampling is to cut the cost down. The research to investigate the association between job characteristics and job satisfaction in all Belgian banks used quite a large sample (n = 15,000). The researcher should narrow down this sample to a more manageable sample, which will cut down the cost significantly. This will reduce the cost of printing the questionnaires, and time factor will reduce. Also, the mailing cost will reduce since fewer mails will be sent. Therefore, to reduce the cost, the researcher need to use a smaller sample size and adopt the suitable sampling technique that will give a representative sample. Since using cluster sampling increases the sampling error, the researcher should adopt another technique. The bank has the list of all employees; thus, this list should be used and a random sample drawn.

Procedure of data collection

Hiring two top competing research institutions was also not necessary. The best institution suitable for this kind of research was the private company specialized in stress at work. They had more knowledge in that area since they exclusively study about stress, which was the main objective of the research. This might be the reason why they had a difference in coming up with the scale of measurement. The private firm has adequate experience on the measures of stress, which will help in just measuring the correct attributes required, using the correct scale. Using one research institute will significantly reduce the cost.

There was no need of printing questionnaires in different languages. All people working at the bank are exemplary learned, and they can understand the two main official languages. Thus, one questionnaire would have been prepared but translation written on the same paper. This would have allowed the respondent choose the language. Also, instead of using the mail, which charges for transportation of mail, the researcher can use other methods of delivering the questionnaire which is cost efficient. Also, this will save the cost of printing the papers.

The Self-completed questionnaire was administered where the bank sent the individual mails to the participants. This was a way of increasing the response rate(O'Leary, 2013). However, there are some problems associated with this kind of data collection. The feedback of such questionnaire is sometimes not representative. Those that feel that they are associated much or concerned about the organization will respond (Rowley, 2014). For instance, those that feel that the organization does not offer much will avoid returning the questionnaire, whereas those that are interested in the firm will return the questionnaire. This means that the views of those that are concerned with the organization will be collected. Therefore, the extreme views of those that are in favor or against the institution will be obtained. Through this, the data gathered are not representative of the actual population.

The time taken to respond to this kind of questionnaire is long. The researcher used postal survey, which is very slow, since it may take days or even weeks to reach to the respondents since most of the people do not regularly check their postal. The researcher needs to specify the deadline for responses. However, there are high chances of some responses coming in late.

A low response rate is another drawback that faces the self-completion questionnaire. The respondents cannot just select “yes/no” answer, rather they have to give a detailed response to the questions. In particular, the response should give in detail full-sentence feedback. This, in most cases, brings problems in analysing the data, for there might be a lot of views from different peoples. The questionnaire also takes longer time to complete, which leads to lower response rate. This also slows down the data analysis process, since it might take longer time to systematically arrange the ideas and code them for analysis.

To rectify this problem in the future study, the researcher should use a closed-ended questionnaire with appropriate scale. The researcher should use structured questions, where the respondents are restrained to a given set of responses. Given that the researcher will use the institution knowledgeable with stress, this will help in correctly making an appropriate answer choice. Also, at this level, the researcher will not be capturing new ideas, which makes structured questionnaire ideal.

The researcher should also use an appropriate channel to deliver the questionnaires more effectively and within a short time. For instance, e-mail will be more appropriate since it will be time-saving and the researcher/bank will not incur the cost of printing the questionnaire. A time frame for returning the questionnaire should be set to ensure that the responses work on completing the questionnaire before the deadline. In accordance with, (Rowley, 2014), a self-completed questionnaire reduces about 10% response rate, but when the structured questionnaire is used, and motivation done to respond, there will be a higher response rate.

Checking representativeness of the sample is an appropriate task since it checks whether the data are consistent with population characteristics. Other studies like that of (Springer, 2011; George & K.A., 2015) data can be used to determine whether the data collected are valid. That is, the research data will be used to determine whether the sample data are consistent with the population. A good example of a government report that can be used is https://www.oecd.org/gov/Belgium.pdf which indicates that in Belgium there is a very high rate of job satisfaction in the banking institution. Thus, the researcher can compare the proportion of job satisfaction using the collected data.  In particular, the government report indicates that there is approximately 80% of the people are satisfied with their jobs. The researcher should expect the sample data for job satisfaction to be around this point estimate.

For instance, the government institution data about banking institution can be used, which shows the proportion of workers in this sector by their demographic background. Therefore, the researcher will compute the distribution and determine whether there are some groups of people that are over represented or under represented (O'Leary, 2013). For instance, the proportion on male and female in this sector should be relatively to those of the employment rate in the Belgian economy.

References

George, E., & K.A., Z. (2015). Job related stress and job satisfaction: A comparative study among bank employees. The Journal of Management Development, 34(3), 316-329. Retrieved from https://search.proquest.com/docview/1664767814?accountid=45049

https://www.oecd.org/gov/Belgium.pdf

O'Leary, Z. (2013). The essential guide to doing your research project. Sage.

Perneger, T. V., Courvoisier, D. S., Hudelson, P. M., & Gayet-ageron, A. (2015). Sample size for pre-tests of questionnaires. Quality of Life Research, 24(1), 147-151. doi:https://dx.doi.org/10.1007/s11136-014-0752-2

Rowley, J. (2014). Designing and using research questionnaires. Management Research Review, 37(3), 308-330. doi:https://dx.doi.org/10.1108/MRR-02-2013-0027

Springer, G. J. (2011). A study of job motivation, satisfaction, and performance among bank employees. Journal of Global Business Issues, 5(1), 29.