CHAPTER 4: METHODOLOGY
This section describes the research method, the survey used and details of how data were collected. The methods to determine statistical significance of the results and test the hypotheses are described.
Primary data were collected using a quantitative approach by making use of the modified JDS. The actual data collection method is described below. To supplement the data and aid in the discussion process informal interviews were also held with members of the professions. Secondary data were collected by studying literature. This consisted of popular and academic journal articles, popular newspapers and online articles and textbooks. These data were used to compare with the primary data, to create an understanding and background for the research and to acquire skills necessary to undertake the research.
An existing survey, the Revised Job Diagnostic Survey (see annexure A) was used to gather data from the chosen group. Details of the use of this survey as opposed to the original JDS are discussed in the literature review. The revised JDS was found to provide measurement equivalence across worker populations in research by Idaszak, Bottom and Drasgow (1988). The validity and reliability of the survey and accompanying model were shown by Boonzaier (1994) to be both valid and reliable. Boonzaier (1994), in his 1989 study obtained correlations between MPS and the 3 personal outcomes of between 0.41 and 0.58. In terms of reliability, Boonzaier (1994) refers to his 1989 article which summarises the reliability coefficients of the personal outcomes across various
studies. The results vary between 0.68 and 0.84 indicating satisfactory internal consistency when compared with Nunnally’s (1967) standards.
The scoring procedure is given in annexure B. This is a structured questionnaire which provides quantitative results based on Hackman and Oldham’s theory. Two sections were added to it in order to gather additional demographic information on the respondents. One was added for actuaries and one for engineers. The types of questions are the same for both groups but the answer options available were customized for each of them. These sections were set up to obtain the following information:
The respondent’s age
- The period worked in the profession
- The period spent in the current job
- Whether more or less than half the respondent’s time is spent managing
- Whether the respondent is a member of the relevant professional body
- The area worked in.
- For engineers these areas were:
o Designing materials, components, systems or processes
o Planning the capacity and location of infrastructure
o Investigating, advising and reporting on engineering problems
o Improvement of materials, components, systems or processes
o Managing or operating plant and processes
o Managing implementation or construction projects
o Implementing designs or solutions
o Research, development and commercialization of products
For actuaries these areas were:
o Life insurance
o Short term insurance
In order to confirm the appropriateness of the survey, it was pilot tested by the
- Mechanical engineer Mr M. Fehrsen
- Electrical engineer Mr A. Da Silva
- Actuary Mr R. Rusconi
They found it to be easy to understand and use. Hence, no changes were necessary. The survey provides the following:
• Scores from 1 to 7 for each of the following job characteristics: Skill variety, Task identity, Task significance, Autonomy and Feedback.
• Scores from 1 to 7 for each of the personal outcomes: Internal work motivation, General Job satisfaction and Growth satisfaction.
The job characteristics provide information on the motivating potential of a particular job. This gives insight into the job structure and highlights potential areas which limit motivation and job satisfaction. These can be seen as the independent variables.
The job characteristics results are used to calculate the Motivating Potential Score (MPS) which consolidates the scores and provides a single value for comparison.
The personal outcomes provide information on the current state of satisfaction and motivation of the employee and can thus be seen as the dependent variables. These results are consolidated into a score, average satisfaction.
Only some of the detailed demographic information such as age and area of specialisation was used for this research. However, it may well be useful for further study.
DATA COLLECTION METHOD
Only a written version of the survey existed so it was necessary to re-write an electronic alternative and incorporate features to make it easy to use in order to maximise the return rate. It was written in MS Excel and incorporated dropdown menus so that respondents could fill it out using only their pointing device.
Not visible to the respondents was a second sheet of the survey which automatically calculated required results such as the MPS. The survey took at most 10 minutes to complete. It was sent as an attachment in a covering e-mail which explained how to fill the survey in, save it and then return the e-mail with the survey attached.
The data for actuaries were collected by the actuarial society. They sent out the survey data and collected the results. Approximately 1300 surveys were sent out and useable ones were obtained from 197 of them. This yielded a response rate of 15%. The e-mail addresses for the engineers were obtained from ECSA, the Engineering Council of South Africa. They selected a random sample. The author then sent out the surveys to these addresses and received the results. In all, 830 surveys were sent out to engineers and correctly filled in ones were obtained from 148. This yielded a response rate of 18%. The intention initially was to send out the same number of surveys to engineers and actuaries but as the process progressed it was decided to get as large a sample as possible.
This was because the return rate was very unpredictable and also many of the e-mail addresses obtained were no longer valid. Table 4 shows the breakdown of the various groups.
An MS Excel spreadsheet was drawn up to automatically pull the results of the surveys into a single sheet. It is shown in annexure D. Various filters were added for sorting and this formed the basis for creating the results tables and graphs seen in the remainder of this document.
TESTING THE STATISTICAL SIGNIFICANCE OF THE RESULTS
As with most statistical analyses, the data obtained are a sample of the population. Although differences between the sample results may appear significant it is important to consider whether the assumptions drawn can be applied to their populations. It is thus necessary to evaluate the statistical significance of the differences between the various results.
The t-statistic for determining the significance of the difference between two means, in other words they are the sum of the squared differences of observations and their mean.
T is then compared to the table of t-statistics with m + n – 2 degrees of freedom.
A statistics tool “PH stat”3 was used. An analysis was carried out for each major result, using the highest and lowest scores. The null hypothesis was that the mean of each of the two populations being compared were the same. Thus, if the null hypothesis was rejected then it indicated that there was a significant difference between the two population means for that category. The chosen level of significance was 0.05. A typical set of results are shown in table 5.
In this case the null hypothesis was rejected, indicating a statistically significant difference between the corresponding mean scores. Whether or not there is a significant difference between two results is influenced not only by the difference between the sample mean scores but also by the sample size and standard deviation.
So for example there may be a large difference between two scores that indicates that the difference is significant. However if one or both of the scores have a high standard deviation the test might show that the difference is not significant.
METHOD OF HYPOTHESIS TESTING
Each hypothesis is restated below and then its method of testing discussed.
Motivating potential scores are positively correlated with average satisfaction.
Testing this hypothesis involves comparing the average satisfaction with the motivating potential score and determining whether there was a correlation between the two. The implication of the hypothesis is that the higher the motivating potential score, the higher the average satisfaction, the average score of the personal outcomes, general job satisfaction, internal work motivation and growth satisfaction. The MPS is a consolidation of the job characteristics scores.
The relationship between MPS and average satisfaction was compared using Excel and PHStat2, an excel statistics add-on included with Levine (1999). The method used was to do a simple linear regression and then to study the outcome from a number of viewpoints. This included a visual inspection of the scatter and residual plots and an ANOVA analysis.
TABLE OF CONTENTS
LIST OF TABLES
TABLE OF FIGURES
1.2. RESEARCH QUESTION AND ITS SUB COMPONENTS
1.3. RESEARCH OBJECTIVES
1.4. MOTIVATION AND POTENTIAL BENEFITS
1.6. DEFINITION OF KEY TERMS
2. LITERATURE REVIEW
2.1. HERZBERG’S TWO FACTOR THEOREM
2.2. THE JOB CHARACTERISTICS MODEL OF HACKAM AND OLDHAM
2.3. LITERATURE ON THE JOB SATISFACTION OF ENGINEERS
2.4. LITERATURE REVIEW SUMMARY
4.1. RESEARCH METHOD
4.2. SURVEY INSTRUMENT
4.3. DATA COLLECTION METHOD
4.4. TESTING THE STATISTICAL SIGNIFICANCE OF THE RESULTS
4.5. METHOD OF HYPOTHESIS TESTING
5. RESEARCH RESULTS
5.1. OVERVIEW OF THE RESULTS
5.2. HYPOTHESIS 1 – CORRELATION BETWEEN MPS AND AVERAGE SATISFACTION.
5.3. HYPOTHESIS 2 – COMPARISON OF ENGINEERS AND ACTURIES WITH OTHERS
6. DISCUSSION, CONCLUSIONS AND RECOMMENDATIONS
6.1. ENTIRE GROUP
6.4. CHEMICAL ENGINEERS
6.5. CIVIL ENGINEERS
6.6. ELECTRICAL ENGINEERS
6.7. MECHANICAL ENGINEERS
6.8. IMPLICATIONS FOR FURTHER RESEARCH
GET THE COMPLETE PROJECT
Job satisfaction and motivation of graduate engineers and actuaries