BUACC5931 Research and Statistical Methods for Business

Response to Question 1

The given population has 69000 respondents and the chosen sample size has 15000 respondents. The sample size for the prescribed study is 21% of the entire population. The sample size should be chosen in such a way that the sample should represent the population properly and it should give a significant statistical analysis of the population of interest. The question requires to give support in favor of or against of the chosen sample size. To get a statistically sound result from a sample, one needs to take into account standard deviation, margin of error, and the confidence level. The sample size is based on the amount of error that is allowed while performing the calculation. Thus, increase of the sample size suggest the decrease of the margin of error (Stegmueller, 2013). Besides, the increase in the margin of error occurs due to the increase of the variation in the population. The relationship among the proportion of population that is chose for sample size, the Z-score (which is obtained from the Z-table), and margin of error can be represented by the equation below,

Where, Z = z-score which is 1.96 in the Z-table at 95% confidence level, the sample proportion, c = margin of error (Stegmueller, 2013). Here the chosen sample proportion is 21% that is, 0.21.

In the course of thus answer, the margin of error will be calculated using the above mentioned formula and then the conclusion will be drawn about the sample size on the basis of the margin of error.

Therefore, the margin error, c =   = 0.006518289 ≈ 0.007.

Clearly it is seen that choosing the 21% as the sample proportion ultimately reduces the value of the margin of error. Usually, the margin of error is taken as 0.05. However, the value is 0.007 here. Therefore, the sample size is large here due to such a small margin of error value. The large sample indicates increase in the statistical power (Stegmueller, 2013). Moreover, large sample size provides more accurate statistical measures like measures of central tendency and measures of dispersion. Therefore it is advantageous to have a sample size of 15000 for 69000 population size (Perrin, 2015).

Response to Question 2

The sampling method used in the course of the research study is Simple Random Sampling (SRS) method as the data has been collected from the population at random and no constraints of gender, age, or employee level have been taken into account. As per the information provided, the sampling method is simple random sampling without replacement and the mode of collected data is primary data using survey questionnaire method. In other words, it can be called self-administered surveys. The entire data collection method is inexpensive and reduces time consumption. However, there may be low response rate due to the lack of interest of the respondents to follow-up the study. Moreover, there may be misunderstanding and wrong interpretation of some questions which leads to increase in the incorrect and inappropriate data collection.

• Advantages of SRS method:

1. The selection method being random, there is no chance of bias. It allows unbiased statistical estimates of the dataset
2. This sampling method id particularly useful for evaluating inferential statistics.

• Every member of the population has equal probability of being selected in the sample. Thus, the outputs of the statistical measures will be significant and systematic error will be reduced (Perrin, 2015).

1. A truly selected random sample is a true representative of the entire population and thus the sample statistics will define the population efficiently.
2. This sampling method is more convenient than the other sampling methods in use. The random numbers are mostly used to collect the sample. Thus, the data collection method is very easy. It removes all sorts of classification error and thus it is the simplest data collection method.
3. This method allows faster formation of sampling groups.

• Disadvantages of SRS method:

1. The method being random, no other feature or characteristics is taken into account. Thus, no further knowledge is considered while collecting the data for the sample (Perrin, 2015).
2. If the units of the sample contain heterogeneity, then the SRS method cannot be used. Moreover, if the sample units are widely dispersed, then also this method is not efficient enough to provide error free statistical results.

• The absolute value of the error is much larger than that of the stratified random sampling.

1. Sample size needs to be large to get a better statistical result in his method.
2. The researchers are needed to be highly skilled for conducting the sampling procedure successfully. If the researcher does not have high expertise then this method will not be efficient to collect the data.
3. Effective population grouping is required in SRS method.

• The collected data may not be reflective of the population or community of interest. If the sample units are forming any group then the response of any respondent may get influenced by the response of the others.
• If the requirement is to collect the data using any personal interview or questionnaire method then there will extra monetary expense.

Response to Question 3

The measures used to study the collected data are two types for measuring quantitative and qualitative data. There are five variables which are Qualitative job security, Quantitative job security, Job satisfaction, Psychological distress, and Control variables. All the variables are measured on Likert scale. Likert scale are mainly used to measure nominal data. In the course of the study, the reliability and variability of the Likert scale will be described. This reliability can be defined as the measure having no error and inconsistency. It is mainly the consistency of any measurement. A measurement is said to be reliable when the test procedure is consistent in evaluating the test score or result. On the other hand, if a measure measures what it is supposed to measure then the measure has validity for measuring that attribute or variable. The Likert-scale has efficiency in capturing variances and self-reported behaviors. The measuring scale used to measure the job insecurity in the course of this study, meets the validity and reliability criteria. The reliability can be measured using the Cronbach’s alpha coefficient. Technically, this coefficient measures the internal consistency. Thus, it measures reliability and the coefficient is a coefficient of reliability (Eisinga, Te & Pelzer, 2013). Higher the value of the coefficient, better the existence of consistency or reliability. The alpha value is mostly accepted by the science researchers if the value is 0.70 or higher (Eisinga, Te & Pelzer, 2013). The formula used to calculate Cronbach’s Alpha is

  ; where N= number of observations, = average value of the covariance, = average variance.

In the course of the study, the Cronbach’s alpha is 0.89 for measuring the reliability of Qualitative job insecurity. It clearly shows that there is high internal reliability. For the Quantitative job insecurity, this Cronbach’s alpha results to 0.87 which also indicates a very high level of internal consistency that is, high reliability. In the results of the analysis of the Psychological distress, the alpha value is 0.89. Hence, it is also an evidence of high consistency.

The variability of any research study depends on several attributes which are criterion validity, predictive validity, construct validity, and content validity.

Response to Question 4

To evaluate the relationship between the qualitative and quantitative job insecurity, the control variables gender, age, and education level are needed to be considered as the control variables enhance the outcome of the research. During the occurrence of the experiment, these variables are held constant to evaluate the relative relationship and they are kept unaltered throughout the experiment. In the course of this proposed study, the gender, age, and education level need to be fixed while assessing the association between the q8ualit5sative job insecurity and quantitative job insecurity. The presence of these control variables is necessary for minimizing the effects of the variables other than the explanatory variable. This leads to increase in the reliability of the outcome of the experiment. In other words, the control variables are another essential factor of the research study. If the control variables are not present and if they are not kept fixed, then the research study will be ruined. Moreover, the research study may give skewed results if there no presence of the control variables or control group. Control variables are mainly the independent variables which are kept fixed. In particular, if the age or the gender variables are not kept constant while comparing the output of the respondents in the Likert-scale, then there will be change in these values and it will ultimately provide insufficient results of the research study. In an experiment, when there are many variables affecting the result, it is convenient to construct a control group to control the influence of the variables. Moreover, it will also help to account for the variables which are originally affecting the result.

Response to Question 5

Research design can be defined as a systematic plan for studying a scientific scenario or measuring the association among quantitative variables or qualitative variables of interest (Baltes, Reese & Nesselroade, 2014). There are four types of research designs that are used in the research study which are-

• Exploratory research design which helps to interpret the research at the beginning of the research procedure
• Descriptive research design which provides an in-depth and detailed research
• Evaluation research design which measures the effectiveness of the research study
• Explanatory research that offers an extensive research approach with all the explanation

  In broader sense, the research designs are mainly 3 types which are Descriptive, Exploratory, and explanatory. Descriptive design of research takes into account the analytical approach. In the course of this research study, the statistical analysis has been done along with calculation of the Cronbach’s alpha coefficient. Therefore, the study has proceeded with an analytical view of the collected data.  Thus, the research design used in the course of this study can be considered Analytical research design or Descriptive research design (Baltes, Reese & Nesselroade, 2014). The other research designs cannot be used here. It is the most widely used design as this design explains the characteristics of the data. In particular, the research design can also be termed as the quantitative research design. The data has been collected using the survey questionnaire method and the interpretation has been taken from the evaluation of the collected data (Baltes, Reese & Nesselroade, 2014). These are the steps that are included in the Quantitative research method. This research design helps to examine the relationship or association of two variables and the requirement of this course of the study is to calculate the association of the qualitative job insecurity and quantitative job insecurity with the help of measuring the Cronbach’s alpha and statistical interpretation. Thus, it can be concluded that this research method is Descriptive quantitative research design method. Any hypothesis is not required in this type of research design. Moreover, the research design involves the statistical analysis of the existing data collected from the survey (Baltes, Reese & Nesselroade, 2014). In addition to this, the used research design is, to some great extent, an experimental research design as it takes into account the statistical analysis coming from the experiment.

References

Baltes, P. B., Reese, H. W., & Nesselroade, J. R. (2014). Life-span developmental psychology: Introduction to research methods. Psychology Press.

Eisinga, R., Te Grotenhuis, M., & Pelzer, B. (2013). The reliability of a two-item scale: Pearson, Cronbach, or Spearman-Brown?. International journal of public health, 58(4), 637-642.

Perrin, A. (2015). Social media usage. Pew research center, 52-68.

Stegmueller, D. (2013). How many countries for multilevel modeling? A comparison of frequentist and Bayesian approaches. American Journal of Political Science, 57(3), 748-761.