Chapter 3 - Research Method

Excerpt taken from my research proposal, pages 119-133.

*Minor adjustments have been made for formatting and readibilty purposes.

IRB # 01-09-26-1195955.

---

Introduction

The purpose of this quantitative study is to examine the possible relationship between socioeconomic status, perceived motivations to higher education, and perceived barriers to higher education, on the academic achievements of women seeking a postsecondary degree or certification in the Freely Associated States.

In this research, I aim to illuminate the structural factors that either encourage or impede women to achieve academic success when pursuing higher education in the FAS. In the following pages, I will explain in greater detail the research design and rationale, target population, and sampling and recruitment procedures for the study. I will also review the variables of the study and further describe the data collection and analysis processes. Finally, I will explain any potential threats to the study’s validity along with the steps I will take to ensure all ethical standards and requirements are met throughout the research and analysis process.

Research Design and Rationale

The study’s variables and research design were carefully selected to support alignment between the purpose of the study and the use of world-systems theory as its theoretical foundation. The first IV, SES, will be determined by the participant’s self-reported household income. While there are different ways to measure SES, household income has been utilized by researchers to understand how various institutional systems and societal practices affect those who live in low-income households and the overall economic development of LMICs (Castro et al., 2024; He et al., 2022; Malnar & Širec, 2023; Rahman et al., 2024).

Perceptions of motivators to higher education (MH) and perceptions of barriers to higher education (BH) are two separate, additional IVs that will be included in the study. It is necessary to measure these perceptions to determine how the same systems and practices that impact SES also shape how women view educational opportunities in the FAS. The DV, academic achievement (AA), will be measured by using the respondents’ self-reported GPA. As I have shown in Chapter 2, GPA is a common and valid way to measure the AA of students who attend postsecondary institutions (Akram & Suneel, 2023; Mohamed et al., 2023; Rentz et al., 2021; Zhou et al., 2022). These variables are consistent with the study’s theoretical foundation and will illuminate how broader socioeconomic systems are related to the AA of women who attend college in the FAS.

I have chosen to utilize a cross-sectional survey design because it is the most accurate and efficient way to collect data at a specific point in time. Furthermore, the self-reporting online questionnaire allows me to gather respondent data anonymously, even from remote locations. This is especially important for respondents in the FAS as they are both geographically isolated from other nations yet still very culturally diverse amongst their own individual populations. In addition, a multiple regression analysis is an effective application for examining how the three separate IVs relate to the single DV. Moreover, each IV can also be isolated during analysis, which will reveal how much the individual IV effects the DV, independently (Laerd Statistics, 2013).

As an example, Richards and Specker (2020) used a cross-sectional design and a self-reporting questionnaire with a multiple regression analysis to examine how sleep quality, food security, and physical activity were associated with college weight gain. When all of these points are considered, the selected design for my study is the most suitable choice to determine if there is a statistically significant relationship between SES, MH, and BH, on the AA of women who attend postsecondary institutions in the FAS.

Given the study’s target population, there are time and resource constraints that a cross-sectional survey design will mitigate. As women who attend postsecondary institutions in the FAS attend classes across multiple local campuses in each of the three countries, reaching them in person presents a geographical challenge. Moreover, as was previously explained, there is a significant amount of physical distance between the three colleges in the FAS, and ultimately, the students I will recruit for this study. Postsecondary students in the FAS attend institutions on various islands situated about two thousand, five-hundred miles from Hawaii, in an oceanic region that is larger than the continental U.S. (Lum & Tupuola, 2024).

Furthermore, access to the FAS is limited and some islands with satellite campuses can only be reached by boat (Tupuola, 2024). Due to the remoteness of the FAS and the distance between the schools and their campuses, it is necessary to employ the use of an online questionnaire to reach as many potential respondents as possible, in the shortest period of time, and with the least amount of expenses incurred.

To further the body of knowledge in the fields of gender equality and higher education, it is necessary to investigate what women think about the social and economic systems they live in. This provides insight into how structural enablers and barriers impact how women perceive academic and professional opportunities for themselves.

When women are educated, they are empowered and can ultimately contribute to the overall economic development of their communities (Ashraf et al., 2024; Girón et al., 2024; Pervaiz et al., 2023; Wani et al., 2024). Therefore, gaining insight into what enables or inhibits their AA is also important for promoting large-scale financial growth and positive social change. It is also necessary to examine the relationships between women’s perceptions and their actual achievement so as to provide evidence of any significant predictors. These predictors could potentially be further explored, or the data used to inform policies related to women’s education in the FAS.

A cross-sectional survey design is the most effective way to collect sufficient and relevant information that can be used to further the knowledge of human services practitioners, advocates of gender equality and women’s empowerment, and those who work in the field of academics.

Methodology

Population

The target population for this study is women who are attending a postsecondary institution in the FAS either full-time or part-time, and who are seeking an academic degree or professional certification. Postsecondary institutions in the FAS include PCC, the COM, and the CMI. Respondents can be nonresidents however, given the statistical demographics of the student populations at PCC, the COM, and the CMI, it is more likely participants will be native to one of the three countries in the FAS.

This population was selected because women often face structural barriers which impede their ability to pursue higher education. Furthermore, as many islanders migrate to the U.S. or one of its territories, women who choose to stay in the FAS for college represent a specific group of adult students. The size of the target population is approximately one thousand between the three institutions.

Sampling and Sampling Procedures

To meet the sample size requirements and to gather the most relevant data, I will use a combination of convenience and purposive sampling to identify and recruit potential participants. Due to the limited population of women seeking degrees or certifications at higher education institutions in the FAS, I will seek participants who are currently attending PCC, the COM, or the CMI. More specifically, I will recruit participants using an electronic flyer that PCC, the COM, and the CMI have agreed to share on their social media accounts and in classrooms. This strategy will give me access to participants who are available and who also meet the particular criteria of the study.

Convenience sampling requires minimal time and resources to identify participants however, there is a higher risk of bias as the selected group does not necessarily represent the greater population. Purposive sampling is necessary for this study as respondents must have a current GPA at a postsecondary institution in the FAS to qualify. While generalizability in this case is limited, the results will provide a greater understanding of the specific subgroup of women receiving higher education in the FAS.

The instrument will be administered as an online survey to students who voluntarily, with consent, choose to participate. To be considered a valid participant, the individual must be a woman seeking an academic degree or professional certification in the FAS. All others will be excluded from the study. Using G*Power software, a total sample size of 119 was determined by selecting ‘F-test’ and ‘linear multiple regression’: Fixed model,  increase with ‘A priori’ type of power analysis, effect size (  = .15), err prob (α = 0.05), power (1-β err prob = 0.95), and three tested predictors.

Procedures for Recruitment, Participation, and Data Collection

To begin the recruitment process, I first received permission from PCC, the COM, and the CMI to post an electronic flyer (see Appendix A) to promote the questionnaire (see Appendix B). This occurred by email and acknowledgement of these permissions have been saved for verification purposes. Next, I will submit the necessary paperwork to Walden University’s IRB for approval to conduct the study.

Once I receive IRB approval, I will share the flyer that will have a direct link and QR code to the survey with the academic institutions mentioned above. Both the direct link and a QR code included on the flyer will take the respondents to the beginning of the survey. The first page of the survey is the informed consent. This form explains the participant’s role as well as how privacy will be protected and how the data will be stored.

The informed consent page also provides contact information for those who may have questions or concerns related to the study. By clicking the ‘Next’ button at the bottom of this page, the participant will have provided consent and will be subsequently directed to the survey on the following page.

The survey consists of a total of thirty-five questions. The first section is titled ‘Demographics’ and asks the respondents to report their age; marital status as either single, married, divorced, or widowed; and employment status as either full-time, part-time, work study, or not employed. The second section is titled ‘Socioeconomic Status’ and includes one question with multiple choice answers to identify the participant’s total annual household income in U.S. dollars. For this question, household income is divided into five options including $0-$24,000; $25,000-$49,999; $50,000-$74,999; $75,000-$99,999; and $100,000+. The third section is titled ‘Motivators to Higher Education’ and includes fifteen five-point, Likert-style questions. The fourth section is titled ‘Barriers to Higher Education’ and also includes fifteen, five-point, Likert-style questions.

The final section is titled ‘Academic Achievement’ where respondents will self-report their current GPA. Once the participant has answered all of the questions, they will click ‘Next’ to exit the survey. On the final page, the participants will be thanked for completing the study and will click ‘Done’ to complete the process and submit their answers. Completed surveys will remain anonymous and will be automatically forwarded to my student email address.

Instrumentation and Operationalization of Constructs

The IVs, SES, MH, and BH can all be found in three different sections of the Motivations and Barriers to Higher Education for Online Learners Questionnaire (Kimmel et al., 2012).  For this reason, this questionnaire is an appropriate instrument for my study because the questions directly address various socioeconomic, personal, and cultural factors representing my variables which may affect the AA of women in the FAS. The instrument is designed to measure the responses of students in higher education which is also representative of my study’s target population.

This questionnaire was originally used by Kimmel and McNeese (2006) to examine how motivations and barriers of adult students differed by race and gender or ethnicity. In another example, Kimmel et al. (2012) used the questionnaire to specifically study nontraditional learners enrolled in institutions that provided academic programs for working adults. The questionnaire was also used by Alshehry (2016) in a study to determine the motivations and barriers for nurses in Saudi Arabia to pursue a doctoral degree. In these studies, higher numbers associated with the Likert-scale questions signified a greater amount of the predictor being measured. In other words, the questions were scored in a positive direction making it easier to interpret the results.

The first IV, SES, is a categorical variable that will be represented by the participant’s household income, as indicated by Question 10 of the Motivations and Barriers to Higher Education for Online Learners Questionnaire (Kimmel et al., 2012). The second IV, perceived motivations to higher education (MH), will be collected on a Likert scale (Strongly Disagree – 1, Disagree – 2, Agree – 3, Strongly Agree – 4, Not applicable – 5) using all of the fifteen questions from Section 2 of the Motivations and Barriers to Higher Education for Online Learners Questionnaire. These items will be added together for a continuous score. Likewise, the third IV, perceived barriers to higher education (BH), will be measured with the same Likert scale using all of the fifteen questions from Section 3 of the Kimmel et al. (2012) survey. These items will also be added together to generate a continuous score. And finally, the DV, academic achievement (AA) is a continuous variable that will be measured by the respondent’s self-reported GPA.

To ensure the quality and rigor of the instrument, I have examined both its reliability and validity as a measurement tool. The questionnaire was developed by Kimmel and McNeese (2006) to identify differences in motivations and barriers of adult students based on gender and race or ethnicity. The authors incorporated significant items from the literature review along with responses from open-ended questions given to groups of adult students on two different campuses in Mississippi, USA to create the questionnaire. Both undergraduate and graduate students were involved in the instrument development groups.

As a result, the authors created a fifty-one-item questionnaire that was then administered to six-hundred, forty-six students in six separate higher education institutions. While this questionnaire has been used to study students attending classes online, it was originally designed for in-person instruction. As I have discussed, the questionnaire I have selected is a sufficient instrument to utilize to determine if SES and perceived motivations and barriers to higher education are related to the AA of women who attend postsecondary institutions in the FAS, given its components and the tests I will conduct to confirm its reliability and validity.

Data Analysis Plan

To properly analyze the participants’ responses, I will use SPSS to organize and prepare the data, test the assumptions of multiple regression, and finally, to perform the multiple regression analysis. Prior to conducting the analysis, the raw data will be imported into SPSS and reviewed to ensure accuracy and suitability for analysis. Missing data will be evaluated for both the extent and the pattern of missingness for each variable using frequency tables and descriptive statistics. I will document any decisions I make to manage missing or incomplete data and provide the rationale for any actions taken in the results chapter.

I will take these steps to prepare the data for analysis to answer the following research question and hypotheses:

RQ: Is there a statistically significant relationship between socioeconomic status, perceived motivations to higher education, and perceived barriers to higher education on the academic achievement of women in postsecondary institutions in the FAS?

H0: There is not a statistically significant relationship between socioeconomic status, perceived motivations to higher education, and perceived barriers to higher education on the academic achievement for women in a postsecondary institution in the FAS.

H1: There is a statistically significant relationship between socioeconomic status, perceived motivations to higher education, and perceived barriers to higher education on the academic achievement of women in postsecondary institutions in the FAS.

Before proceeding, I will carefully evaluate each assumption to ensure the results are valid and reliable. It is important to test the assumptions prior to completing the analysis to provide information on the accuracy of the study’s predictions and to show how well the regression model matches the data (Laerd Statistics, 2013). If these assumptions are violated, corrections or alternative statistical testing may be required. While there are eight assumptions which must be satisfied, the first two are determined by the study’s design.

To meet these two initial requirements, the study must include a continuous DV as well as two or more continuous or categorical IVs, which are both present in my study (Laerd Statistics, 2013). The remaining six assumptions of multiple regression are related to the nature of the data and discussed in further detail below.

Independence of Errors

The first assumption I will test for is independent observations. In a multiple regression analysis, observations cannot be related so testing for this will show whether errors are independent of each other (Laerd Statistics, 2013). Given the possibility that observations could be related, the independence of errors will be checked using the Durbin-Watson statistic which is provided as a number between zero and four. If the residuals are in fact independent, the Durbin-Watson value will be two, or very close to two. If the errors are found to be correlated, I will continue with the analysis but account for the violation in the discussion of the results.

Linearity

The second assumption I will test for is linearity. It is necessary for the IVs to both independently and collectively demonstrate a linear relationship with the DV (Laerd Statistics, 2013). This can be determined using a scatterplot of the standardized residuals against the predicted variables to show collective linearity and a partial regression plot between each IV and the DV to show individual linearity. To meet this assumption, the relationships between the IVs and the DV must follow a straight line. If linearity does not exist, I will transform the data depending on the results or continue with the analysis and make sure to account for the violation in the discussion.

Homoscedasticity

Next, I will test for homoscedasticity, or that the variance of residuals is constant across all levels of the IVs. To check for homoscedasticity, I will visually assess the same scatterplot that was created to check for linearity to identify whether a pattern is present (Laerd Statistics, 2013). If homoscedasticity is not met and heteroscedasticity exists, the points on the plot will show a funnel of fan-shaped pattern instead of being randomly and evenly scattered throughout. In the case of heteroscedasticity, I can run a weighted least squares regression equation, run a robust regression, run a regression with robust standard errors, or transform the variables to correct the violation. Actions to remedy heteroscedasticity will be determined based on the data and whether linearity is evident.

No Multicollinearity

The fourth assumption that I will test for is multicollinearity. Multicollinearity occurs when two or more IVs are highly related to each other (Laerd Statistics, 2013). This can cause technical issues when performing a multiple regression analysis and also makes it difficult to identify how each variable contributes to any variances.  To check for multicollinearity, I will examine the correlation coefficients to make sure the IVs do not have correlations greater than 0.7. I will also check the Coefficients table to make sure the Tolerance value for each variable is greater than 0.1. If these tests show that multicollinearity exists, I will decide at that time to either exclude one of the variables and run the analysis again or continue with the analysis as it is and account for the violation in the discussion.

No Significant Outliers

The next step will be to check for any unusual data points. Outliers can occur when a data point does not follow the usual pattern of points and is observed to be a significant distance from the predicted value (Laerd Statistics, 2013). To detect any outliers, I will instruct SPSS to identify any cases with a standardized residual that is greater than 3 standard deviations. If outliers are present, I will review the casewise diagnostics table to identify which cases will need to be removed and filter them out of the data set and rerun the analysis. If there are any unusual data points that are not removed for any reason, I will thoroughly document the decision to retain those cases.

Normality of Residuals

The final assumption I will check for is normality. This means that the errors in prediction, or the residuals, must be normally distributed (Laerd Statistics, 2013). One method I will use to check for normality is to visually assess the histogram produced by the Linear Regression: Plots dialogue box to see if the data follows the normal bell curve pattern. To further validate my findings, I will also review the P-Plot that will be produced from the same Linear Regression: Plots dialogue box. To verify normality, I will check to see if the points on the P-Plot are aligned along the diagonal line of the chart. Should normality be violated, I will consider transforming the variables depending on the results or I will continue with the analysis despite the violation and be sure to address it in the study’s final discussion.

Threats to Validity

 As with any cross-sectional survey design, there are inherent threats to validity which may impact the credibility of the results. To begin, data will only be collected at a single point in time which limits the internal validity of the study. Additionally, self-reporting measures may include response biases or unintentional inaccuracies which can affect construct validity. And furthermore, sampling constraints could occur which reduces external validity by limiting the generalizability of the results to a wider student population. It is important to acknowledge these potential validity threats as a contribution to the transparency of the results and to provide clarification on the limitations of the study.

Ethical Procedures

 To remain ethical during the research process and maintain participant protection and responsibility with any type of research data, I have received the approval of Walden University’s IRB prior to recruiting participants and collecting data. I have contacted each postsecondary institution by email to acquire proper protocols and permissions for distributing the flyer and promoting the study. I will not provide any incentives to encourage respondents to participate and the study will not be conducted within my own professional environment. Therefore, there will be no existing power differentials between me as the researcher and the participants.

To guarantee participant protection, I will provide a consent form that explains all the relevant factors of the study to the participants so they can make an autonomous, well-informed decision to voluntarily partake in the research. Informed consent must be provided to and understood by the potential participants and will include information related to the purpose of the research and a description of what the participant will be expected to do. The questionnaire will be completed anonymously using Survey Monkey. To ensure privacy, I will not collect any participant names or contact information. The data that is collected during the research process, whether electronic, textual, or numerical, will be stored in password-protected files on my own computer, with an additional screen lock that only I will have the password to. And finally, any and all data files will be destroyed five years after the conclusion of the study, as per Walden University’s IRB requirements.

Summary

In this chapter, I have outlined the methodological framework I will use to examine whether SES, perceived motivators and perceived barriers to higher education are related to the AA of women attending postsecondary institutions in the FAS. I selected a quantitative, cross-sectional design to gather data efficiently from geographically dispersed colleges. In addition, this design supports a multiple regression analysis, allowing me to examine the effects of the three IVs both independently and collectively. Furthermore, I defined the study’s population, the rationale for using convenience and purposive sampling, and explained the participant recruitment procedures. I also thoroughly discussed my data analysis plan and any threats to the validity of the study. And finally, I have acknowledged the ethical procedures I will follow to ensure I protect the privacy of the participants and any data that will be collected and analyzed. In Chapter 4, I will present my findings, address the research question, and test my hypotheses.

---

References

Akram, B., & Suneel, I. (2023). The Relationship between optimism and academic achievement among university students in Pakistan. Pakistan Journal of Social Sciences (PJSS), 43(2), 283–294.

Alshehry, A. (2016). Motivations and barriers for Saudi nurses to pursue a doctoral degree. Doctoral Dissertations. 548. https://doi.org/10.7275/7900379.0

Ashraf, M.S., Ahmed, F., Kousar, S., Ferreira, P.J.S., & Fortes de Almeida, D.M. (2024). People category of UN SDGs 2030 and sustainable economic growth in Asia and the Pacific Region. Sustainability (Switzerland), 16(18).https://doi.org/10.3390/su16187950

Castro, C., Pavez, C., Contreras, F., & Morales, F. (2024). Academic performance of adult women in higher education: A case study at University of the Américas, Chile. International Journal of Interdisciplinary Educational Studies, 19(2), 249–272. https://doi.org/10.18848/2327-011X/CGP/v19i02/249-272

Girón, A., Cicchiello, A.F., Ferilli, G.B., Kazemikhasragh, A., & Kazemi, Z. (2024). Gender equality and countries’ financial and economic well-being: New evidence from emerging economies. Journal of Economic Issues, 58(3), 1035-1049. https://doi.org/10.1080/00213624.2024.2382052

He, Q., Boterman, W., Musterd, S., & Wang, Y. (2022). Perceived social distance, socioeconomic status and adaptive residential mobility in urban China. Habitat International, 120, 102500. https://doi.org/10.1016/j.habitatint.2021.102500

Kimmel, S.B., Gaylor, K.P., Grubbs, M.R., & Hayes, J.B. (2012). Motivations and Barriers to Higher Education for Online Learners Questionnaire [Database record]. https://dx.doi.org/10.1037/t28333-000

Kimmel, S.B., & McNeese, M.N. (2006). Barriers to business education: Motivating adult learners. Institute of Behavioral and Applied Management.

Laerd Statistics. (2013). Multiple regression SPSS statistics. Lund Research, Ltd. https://statistics.laerd.com/premium/spss/mr/multiple-regression-in-spss.php

Lum, T., & Tupuola, J. G. (2024). The Freely Associated States and issues for Congress. Congressional Research Service: Report, 1–18.

Malnar, A., & Širec, K. (2023). Entrepreneurial activity in Slovenia: The role of gender and household income. Advances in Business-Related Scientific Research Journal, 14(1), 1-14.

Mohamed, E.Y., Sami, W., Emad Almhmd, A., Homdi K Alenazy, S., Ghayeb Alrashidi, A., Mashhi Aldhafeeri, B., & Nasser Binmuhareb, A. (2023). The association between body mass index and lifestyle with academic performance of college of medicine students, Majmaah University, Saudi Arabia. Advances in Human Biology, 13(1), 118–123. https://doi.org/10.4103/aihb.aihb_79_22

Pervaiz, Z., Akram, S., Ahmad Jan, S., & Chaudhary, A.R. (2023). Is gender equality conducive to economic growth of developing countries? Cogent Social Sciences, 9(2), 2243713. https://doi.org/10.1080/23311886.2023.2243713

Rahman, M.N.A., Tang, D., & Chew, F.P. (2024). Does socioeconomic status moderate the relationship between parental involvement and young children's literacy development? International Journal of Instruction, 17(1), 559-578. https://doi.org/10.29333/iji.2024.17129a

Rentz, B., Holquist, S.E., Arens, S.A., Stuit, D., Rhinesmith, E., Nicotera, A., & Plotz, M. (2021). Using high school data to predict college success in Palau. [PDF]. Regional Educational Laboratory Pacific. https://files.eric.ed.gov/fulltext/ED610714.pdf

Richards, A.L., & Specker, B. (2020). Evaluating hours of sleep and perceived stress on dietary cognitive restraint in a survey of college students. Journal of American College Health, 68(8), 824-831.https://doi.org/10.1080/07448481.2019.1618312

Tupuola, J.G. (2024). The Freely Associated States and issues for Congress. [PDF]. Congressional Research Service. https://www.congress.gov/crs-product/R48311

Wani, I.U., Alharthi, M., Khanday, I.N., Subhan, M., & Al-Faryan, M.A. (2024). Exploring the complementary interaction between financial inclusion and gender equality on economic growth: Fresh evidence from developing countries. Cogent Economics & Finance, 12(1), 2365585. https://doi.org/10.1080/23322039.2024.2365585

Zhou, Y., Sackett, P.R., & Brothen, T. (2022). Personality aspects and the underprediction of women’s academic performance. Applied Measurement in Education, 35(4), 287–299. https://doi.org/10.1080/08957347.2022.2155652