How to Read Statistical Analysis

Comparing Studies: What are the Data Saying?
Types of Studies
The study by Ong, Chua and Ng (2014) entitled “Barriers and facilitators to self-monitoring of blood glucose in people with type 2 diabetes using insulin: a qualitative study” was a qualitative study that used interviews to collect data from diabetes patients. The study by Stevens, Shi, Vane, Nie and Peters (2015) entitled “Primary care medical
home experience and health-related quality of life among adult Medicaid patients with type 2 diabetes” was a quantitative study that used surveys to collect data. The study by Wildeboer, du Pon, Schuling, Haaijer-Ruskamp and Denig (2018) was a mixed-methods study that used interviews and self-reported data to collect information on care providers’ views towards using a patient-oriented treatment decision aid when treating type 2 diabetes patients.
Statistical Tests
Statistical tests were not conducted in the qualitative study by Ong et al. (2014). Instead, thematic analysis was conducted using the comparison technique by taking the transcriptions of the interviews and identifying emergent common themes across the two dozen transcripts.
In the study by Stevens et al. (2015), a 3-point Likert-scale was used to measure responses from participants in the survey. These scores were then analyzed using descriptive statistical analysis. T-tests were conducted to establish significance between the Likert scale scores and the EQ-5D index that was also used to measure variables assessed in the study. F-statistics were also used to compare averages. Chi-squared tests were conducted to test whether there was a relationship between demographics and health measures. Finally, multivariable linear regression analysis was conducted.
In the study by Wildeboer et al. (2018), qualitative content analysis was conducted with interviews being transcribed and coded. Pearson point-biserial correlation was conducted to test associations between views of care providers and the use of the decision aid.
Applicability and Why the Statistical Test was Chosen
In the study by Ong et al. (2014) no statistical test was chosen because the data collected was entirely qualitative and the researchers were looking only to identify themes that could help explain the barriers and facilitators of self-management for diabetes patients.
In the study by Stevens et al. (2015), the t-test, F-statistical analysis and chi-squared tests were applicable because the researchers wanted to establish whether or not there was significance with respect to the scores they recorded, and they also wanted to test to see if there was a significant relationship between the health measures used and the demographics of the sample. These tests were chosen because they fit with the data collected and the design of the study.
In the study by Wildeboer et al. (2018), the Pearson point-biserial correlation test was conducted to test for correlation between the viewpoints provided by the care providers interviewed for the study and their intention to use the decision aid. The Pearson test was chosen because it is an appropriate test for correlation among two variables.
Differences between Parametric and Nonparametric Tests
The parametric test is a hypothesis test that is used to obtain general information that can then be used to explain the mean of the sample. A typical parametric test is the t-test. Pearson’s correlation test is another typical parametric test. Essentially, whenever an assumption about the parameters of a sample population is utilized, the parametric test is the statistical test that the researcher will engage. The measurement level of the parametric test is the interval or ratio level and the mean is the measure of central tendency.
The nonparametric test is a hypothesis test that has no assumptions built into it. It is a distribution-free approach to analyzing data. The variables that are measured here are nominal or ordinal. When the independent variables are not metric-based, the nonparametric test is used.
The main differences between the two are that when a test statistic is based on distribution, a parametric test is performed. When the test statistic is arbitrary, a nonparametric test is performed. When a parametric test is conducted, the variables are measured at the interval or ratio levels. When the nonparametric test is conducted, the variables are measured at the nominal or ordinal level. Parametric tests measure the mean; nonparametric tests measure the median. Parametric tests apply to variables; nonparametric tests apply to variables and attributes. Parametric tests use Pearson’s correlation test when measuring relationships between two variables; nonparametric tests use Spearman’s correlation test (Surbhi, 2016).
How They Were Applied in the Articles
In the study by Ong, neither parametric nor nonparametric tests were conducted because the study was qualitative and the researchers did not perform statistical analysis of variables. In the study by Stevens et al. (2015), the researches conducted both parametric tests (t-test) and nonparametric tests (chi-squared tests) to test their hypotheses of the different variables at different measurement levels. In the study by Wildeboer et al. (2018), the researchers conducted parametric tests (Pearson’s correlation test) to test the relationship between the two variables of the study.
How the Factors of Reliability and Validity are Accounted for in the Articles
In the study by Ong (2014) validity was ensured by the researchers by testing their interview questions with a pilot study using two participants to make sure the questions were relevant and were suitable in obtaining the kind of information the researchers were seeking to obtain. The issue of reliability was not addressed. Likewise in the study by Stevens et al. (2015) and in the study by Wildeboer et al. (2018), the issue of reliability and the issue of validity was not discussed. No mention of pilot studies was conducted to ensure validity and no mention of reliability was given either. Nonetheless, the design and data collection methods were thoroughly described so that other researchers could test them for validity and reliability if they were so inclined. The methods were not hidden by the researchers—only no discussion of these factors was provided.
How the Studies Could be Applied in the Context of My Practice
The study by Ong et al. (2014) could be applied in the context of my practice because it identifies barriers and facilitators to self-monitoring of blood sugar levels among a specific population, indicating that cultural factors will play an important role in how patients confront their diabetes. In my practice, having awareness of cultural background can thus be helpful in assisting the patient in overcoming obstacles that stem from the patient’s own cultural, lifestyle and environmental background.
The study by Wildeboer et al. (2018) could be applied in the context of my practice as it shows how care providers look to decision aids as being helpful and the factors that cause them to embrace this patient-centered practice and the factors that cause them to reject it. By promoting factors that embrace it in my practice, I can help care providers to be more patient-centered in their care approach.
The study by Stevens et al. (2015) can be applied in the context of my practice as it showed how patients view home care more favorably than in-patient care. This applies because it reinforces the idea that patients want to be put first: they want their concerns and well-being to come before the care providers’. Reminding care providers to always put patients first in every regard can help to improve quality of care in my practice and help health care professionals to always be working to improving their own care approach towards patients.

References
Ong, W. M., Chua, S. S., & Ng, C. J. (2014). Barriers and facilitators to self-monitoring
of blood glucose in people with type 2 diabetes using insulin: a qualitative study. Patient Preference and Adherence, 8, 237.
Stevens, G. D., Shi, L., Vane, C., Nie, X., & Peters, A. L. (2015). Primary care medical
home experience and health-related quality of life among adult medicaid patients with type 2 diabetes. Journal of General Internal Medicine, 30(2), 161-168.
Surbhi, S. (2016). Difference between parametric and nonparametric tests. Retrieved
from https://keydifferences.com/difference-between-parametric-and-nonparametric-test.html
Wildeboer, A., du Pon, E., Schuling, J., Haaijer?Ruskamp, F. M., & Denig, P. (2018).
Views of general practice staff about the use of a patient?oriented treatment decision aid in shared decision making for patients with type 2 diabetes: A mixed?methods study. Health Expectations, 21(1), 64-74.


Appendix
Comparison Table.
Type of article
Title of article
Type of analysis
Applicability of test
Reliability and validity

Qualitative
Barriers and facilitators to self-monitoring
of blood glucose in people with type 2 diabetes using insulin: a qualitative study
Thematic analysis
Correlation analysis could be conducted if quantitative approach was desired
Pilot study conducted to test validity of interviews. Reliability not addressed but methods in which data was coded was provided.

Quantitative
Primary care medical
home experience and health-related quality of life among adult medicaid patients with type 2 diabetes
t-test, F-statistical analysis and chi-squared tests
z-test and Mann-Whtiney could also be applicable here
Reliability and validity were not discussed by the researchers but the methods were thoroughly described so reliability and validity could be tested later on

Mixed-Methods
Views of general practice staff about the use of a patient?oriented treatment decision aid in shared decision making for patients with type 2 diabetes: A mixed?methods study.
Pearson point-biserial correlation test
Thematic analysis could be applicable to bear out the qualitative aspect of the study
Reliability and validity were not discussed by the researchers but the methods were thoroughly described so reliability and validity could be tested later on