Introduction/Purpose of the Research
There has always been question of whether employees of a company are paid on an equal pay scale regardless of their gender. The problems that are going to be researched will be broken into two parts. These problems will be: gender and years of education. After watching President Obama talk about the economy and the lack of so many Americans without work, it became clear that it is important to understand and evaluate how wages affect multiple companies and the employees that are employed with these companies. This research will be important to show why it is increasingly difficult to be able to afford living day to day for some employees and others find it easier to live.
The questions that are going to be answered are the following: What is the difference between the pay of men vs. women and how the education variable affects his or her pay?
The next step following the purpose of the research and problem definition is the research hypothesis. Researchers must consider and decide on the possible outcomes for the agreed research topic. Researchers must be aware of the many stages of the project such as ethical considerations, data collection, analysis and completion. The success of a project may depend on his or her ability to plan the activities, combined with some luck, judgment and determination (Serrant-Green, 2008). The researchers must identify the different variables to be consider, how these variables are going to be measure and define the measurement scale for each variable. Prior to the development of the hypothesis a conceptual model known as the theoretical framework on how the several factors already identified as important for the research work together making logical sense must be accomplished. Some of the factors to be considered are the prior documentation on the problem area, management’s beliefs, published research, boundaries and constraints of the particular situation to mention some. One of the main concepts to keep in mind is the problem variable which by definition is anything that can take different or varying values on the research problem. The importance of identifying the appropriate variables and their relationships nature and direction are evident for developing the correct hypothesis which in turn will be subsequently test the hypothesis, with an end result identifying possible proven solution to our problem.
Hypothesis development is nothing more than testing whether the relationships theorized hold true or not. Testing the variables scientifically allows the researcher to obtain reliable information. The formulation of testable statements is known as hypothesis development. “A hypothesis can test the relationships or differences of the research problem i.e. employees who are healthy will take less sick leave” (Sekaran, 2003). Using this model for our specific research, men make more money than women and the gap gets wider as years of education add up.
Another manner to test the hypothesis is the directional or non-directional manner i.e. the greater the stress in the job the less job satisfaction for employees. Using this hypothesis mode for our research problem; payroll goes up as more men joined the organization and last payroll increases as the number of years with the organization go up.
The null and alternate hypothesis method is a proposition that states exactly the lack of relationship between two variables. The mean of the population correlation or that the difference in the mean of two groups in the population is equal to zero. In our research problem the null and alternate hypothesis method means that there is no difference between the population and the sample selected for the research. This model of hypothesis does not apply to our particular problem.
One more possible hypothesis method to consider on our research is the negative case analysis. According to this model the researchers look for data that will prove the opposite of the hypothesis is true. In our research situation it will show that men do not make more money than women and pay does not go up as the number of years of education increase.
1) Explain three possible outcomes that can result from the research
In sum, three possible outcomes of the research hypothesis could be that men make more money than women due to the higher number of men in the workforce. Second, employees pay increases with years of education as employees gain efficiency skills. A factor all researchers must acknowledge as he or she proceed according to Gommesson is to “recognize that interpretive elements are influential and present in all types of research and see them as an asset rather than a cross to bear” (2003).
2) Define Operational Definitions
a) Identify the Variables
“A variable is anything that can take on differing or varying values,” (Sekaran, 2003). Gender, and years of education will be the variables being compared to determine the difference in pay between men and women. The years of education will be the variable used to determine if there is a steady progression in compensation in relation to years of education.
b) Define Level of Measurement
“A scale is a tool or mechanism by which individuals are distinguished as to how they differ from one another on the variables of interest to our study,” (Sekaran, 2003). There are four basic types of scales: nominal, ordinal, interval, and ratio. In this study we will be applying the nominal scale and ordinal scale. The nominal scale allows the researcher assign code numbers to certain categories or groups. For example male and female can be assigned code numbers 1 and 2 which have no value other than to create two different categories. The other scale that will be used in this study is the ordinal scale. This scale allows the researcher to rank the categories in a meaningful way. For example: years of service, age, and education are categories that can be ranked from smallest to highest or highest to smallest.
c) Define the Measurement Scale
Nominal data lends themselves to dichotomous or category scales and ordinal data to any one of the ranking scales: paired comparison, forced choice, or comparative scales. The dichotomous scale is used to retrieve a Yes or No answer and the category scale uses multiple items to retrieve a single answer. The paired comparison scale is used when respondents are asked to choose between two objects at a time. The forced choice scale enables the respondents to rank objects to establish importance or preference. The comparative scale creates a point of reference to assess the respondents’ attitude toward the objects under study.
Remaining steps in the research project
In the weeks ahead we will need to make sure that the instrument we develop measures the data accurately. The reliability of measures will need to be without error to insure a consistent measurement of all variables. And finally we will need to apply certain validity tests so we can be reasonably certain that we are measuring our hypothesis.
a. Descriptive Statistics
In an ongoing effort by the team trying to determine if the difference in the wages from our sample population of men and women is affected by various levels of education.
In our investigation, the team will hope to convince the audience of the hypothesis chosen by the team through introducing our statement regarding the research issue, performing the five step hypothesis testing procedure on the data, explain the nonparametric test the team chose to analyze the data and why the team chose this particular test. The team will then interpret the results of the test; explain the differences that were observed from the teams week three paper. The team has included the raw data tables and result of this weeks test in Appendix A-D. The key is using the right data at the beginning to make the difference in how the test results will turn out.
The data chosen by the team used in this research paper is the same as what was used for the previous two sample hypothesis running test called Wages and Wage Earners Data Set. To help the team determine the significance between wages earned by men and women of different educational levels, the team needed to convert the data from the tabular format as seen in Appendix A, to a layout of merged data that would assist the team in setting up the nonparametric test as shown in Appendix B. Because the test chosen uses the sum of the rank and the sample size to compare the independent data groups, the team had to format the data into a worksheet of sum ranks as seen in Appendix C. Finally, it was considered necessary of the team to setup the data to be able to run the nonparametric test that the team chose for this paper.
Formulate the Hypothesis
For this week’s research paper, the team chose to use the population median as part of our hypothesis statement. The reason for this is that the type of nonparametric test selected by the team deals with rank instead of a precise statistical assessment. The research question remain the same as it did last week; is there a significant difference in wages earned for a position if the amount of years spent in school and the sexual characteristics change? Looking at the median of the data in the five different groups in our study it still consistently shows a recognizable distinction. The only way to determine if the group medians are the same in our original study is by running the nonparametric test chosen by the team, which is the Kruskal-Wallis test. Perform the Hypothesis Test
In doing our research, the team remembers the importance of performing the five-step hypothesis test on the data. In doing so, the team can be assured that our test is not biased and the results will overwhelmingly state our stance for the hypothesis. Stating the hypothesis that H0: all c population medians are the same and H1: not all the population medians are the same will lead us to our next step of choosing a significant level. Because we used the µ = .05 in the previous One Factor ANOVA test, we will use the same in this test to ensure that no partiality in our research. The next step in the hypothesis test is to state the decision rule, which was determined by using the degree of freedom for the columns of our five different groups. The team used the tabular data from table 1 in the Appendix of this paper and Appendix E of Doane and Seward to determine that ν = c − 1 = 5 – 1 = 4, and at µ = .05 which results in the team rejecting H0 if the Doane and Seward Chi-square Appendix E value was higher than 9.488 (2007). After that, the team had to calculate the test statistic by using the Appendix B worksheet rank sum sample size and rank sum from each group. The following formula gives the results of that worksheet
Our team then made our decision that because the test statistic of 18.51 is far greater then that of the Chi-square value at v = 4, µ = .05 in Appendix E, we must reject the null hypothesis that all c populations are the same. These results can be viewed in Appendix D, where at the significance level of .05, it shows the Chi-square of 25.75, which is still far above the level for the team to reject the null hypothesis that all c population medians are the same. The team continues the research by giving details to the nonparametric test we used and why we opted to use it.
Explain Kruskal-Wallis Test
The use of the nonparametric test teaches all researchers to use smaller sample, normality is not a requirement, and can be used for ordinal (ranked) data. The reason for our team choosing the Kruskal-Wallis test is that; one, the corresponding test to the One Factor ANOVA test that we ran in the previous research paper of week three; two of the team members were able to compare independent samples from the five different groups even though the sample sizes were not the same, but did include the minimum of five observations; three the Kruskal-Wallis test does not require normality within the populations; and finally when the outliers, or unequal grouped variance, the Kruskal-Wallis test performs as well as the One Factor ANOVA test. To help the team understand the findings of this week’s research, the results of the Kruskal-Wallis test will be interpreted and compared against the results of last weeks test to help explain any differences.
By looking at the median of the data in the five different groups in our study we still show a recognizable distinction. The only way to determine whether the group medians are the same or not in our study is by running the test chosen by the team, which is the Kruskal-Wallis test. These details to the nonparametric test we used and we show the reason we opted to use it. In our previous Week 3 testing and findings using the ANOVA test we concluded that we would accept our alternate hypothesis of not all the means are equal, which states that education and gender are factors in the differences of the wages between men and women. The findings in these two studies using the ANOVA test and the Kruskal-Wallis test have demonstrated that educated men and women and years of education is directly related to their salaries.
sample variance 154,893,232.30
sample standard deviation 12,445.61
coefficient of variation (CV) 50.90%
1st quartile 16,373.00
3rd quartile 31,503.00
interquartile range 15,130.00
low extremes 0
low outliers 0
high outliers 0
high extremes 1
suggested interval width 5000
sample variance 340,313,003.72
sample standard deviation 18,447.57
coefficient of variation (CV) 50.55%
1st quartile 22,485.00
3rd quartile 49,898.00
interquartile range 27,413.00
low extremes 0
low outliers 0
high outliers 0
high extremes 0
suggested interval width 10000
Female Years of Education
Sample variance 4.30
Sample standard deviation 2.07
coefficient of variation (CV) 16.60%
1st quartile 12.00
3rd quartile 13.00
interquartile range 1.00
low extremes 3
low outliers 1
high outliers 5
high extremes 2
suggested interval width 1
Male Years of Education
Sample variance 10.94
Sample standard deviation 3.31
coefficient of variation (CV) 25.55%
1st quartile 12.00
3rd quartile 16.00
interquartile range 4.00
low extremes 0
low outliers 2
high outliers 0
high extremes 0
suggested interval width 2
FYE Avg. Wage MYE Avg. Wage
8 20888 4 19981
9 19306 5 46646
11 14617 6 19388
12 22159 7 26795
13 25648 8 29736
14 26536 9 66738
16 29354 10 31013
17 58701 11 23287
This paper discussed scientific ways to answer the question of whether employees are paid equally regardless of gender and years of education. From the above chart a female with eight years of education earns an average of 20,888 while the male counterpart with the same years of education earns 29,736. Our research shows male wages are higher than female wages with the same years of education. In addition the male wages increase at a higher rate than female wages as the years of education increase. Our research showed the hypothesis in reference to males earning more than females with the same years of education is correct but more research is recommended. Another comment that needs mentioning is the gap between male and female wages is getting smaller as years go by. After watching President Obama talk about the economy and the lack of so many Americans without work, it became clear that it is important to understand and evaluate how wages affect employees in reference to gender and years of education.
Doane, D (2007). Applied Statistics in Business and Economics. Burr Ridge, IL: McGraw-Hill.
INDUSTRIAL MARKETING, 18(6/7), 482-492. Retrieved February 4, 2008, from University of Phoenix database.
Sekaran, U. (2003). Research methods for business: a skill building approach. [University of Phoenix Custom Edition e-Text]. Hoboken, NJ: Wiley. Retrieved February 4, 2008, from University of Phoenix, GEN480- Interdisciplinary Capstone Course.
Serrant-Green, Laura. “Managing research is vital for project success. (Editorial). .” Nurse Researcher. 15.3 (March 2008): 3(1). General OneFile. Gale. Apollo Library. 9 Feb. 2009