NURS 589 Review of the statistical methodology used in a research paper
Paper must have a title page, your name, page numbering, references.
Length: 6-12 pages (not to exceed 12)
Use this discussion forum for any questions you have regarding this assignment.
1. select a research article from a medical/nursing research publication
(no magazine articles, no pilot studies)
2. Write a paper discussing this article, using what we learned in NURS 589
3. General structure:
a. Title page
b. Brief intro: describe the study; What is the "theory" of the study? What are they trying to show or investigate?
c. Brief variable background: what are they, how collected?
d. What decisions did the researchers make (if any) regarding the variables? Why did they include/exclude some variables (if any)?
e. What statistical tests were used? Why? Null Hypothesis? Results? What are they actually measuring? Talk about this in actual terms. Show me what you learned this semester.NURS 589 Review of the statistical methodology used in a research paper
h. improvements suggested
Review of the statistical methodology used in a research paper
Saint Joseph College
This paper examines the statistical tools used to analyze data in a research done by Wu, Prosser,
and Taylor (2010). The purpose of their study was to examine the relationships between
perceived social support and depressive symptoms on blood pressure. They conducted a crosssectional
study of 159 African American women from various Detroit urban areas. Blood
pressure was the dependent variable while depression and social support were the two main
independent variables. They used three statistical tests to analyze their data, namely: Pearson’s
product moment correlation analysis, multiple linear regression analysis, and logistic regression
analysis. They concluded that depression had a direct effect on hypertension in these women,
while social support did not.
Pearson’s product moment correlation coefficient is the first of three statistical tools used by Wu,
Prosser, and Taylor (2010). (Marston, 2010; Pallant, 2007) Correlation is a statistical technique
for investigating the relationship between two quantitative, continuous variables. Correlation
refers to any of a broad class of statistical relationships involving dependence. Pearson product
moment correlation coefficient (PPMCC) is the most widely used correlation technique. It is
also known as Pearson's r, or Pearson's correlation. It is a measure of the strength and direction
of the linear association between the two variables. It is the go to parametric test to determine
the strength of linear dependence between two variables, as it is sensitive only to a linear
relationship between two variables. In this case the researchers wanted to know the relationship
between depressive symptoms and perceived social support on blood pressure in African
American women. They wanted to know if there were any connections between depression and
social support on one hand and blood pressure on the other among these women. Depression and
social support were defined so that they could be quantified using the center for epidemiological
studies depression (CES-D) scale, and multidimensional scale of perceived social support
(MSPSS) respectively. This allowed the researchers to have all continuous variables, which is a
requirement for using Pearson’s correlation. Pearson’s correlation can also be used to test the
relationship between one continuous and one dichotomous variable. Hence in the case of
hypothesis II (page 697, subsection “data analysis”) the researchers had one continuous variable
(blood pressure) and one dichotomous (depressive symptoms which is either higher – a score of
16 or more – or lower). NURS 589 Review of the statistical methodology used in a research paper
Once it is determined that the variables are continuous, the researchers likely proceeded by
making sure that the assumptions of for using Pearson product moment correlation coefficient
were met. (Marston, 2010; Pallant, 2007) One of the most important assumptions is ensuring
that the relationship between the variables is linear. They would have likely drawn a scatterplot
with blood pressure on the y-axis and CES-D scores on the x-axis and another with blood
pressure again on the y-axis and MSPSS scores on the x-axis. For correlation only purposes, it
does not really matter on which axis the variables are plotted. However, conventionally, the
independent (or explanatory) variable is plotted on the x-axis (horizontally) and the dependent
(or response) variable is plotted on the y-axis (vertically). The dots on these graphs should look
roughly linear. The correlation coefficient should not be calculated if the relationship is not
Another assumption that the researchers would have had to check is normality. (Pallant, 2007)
Each variable, that is blood pressure depressive symptoms and perceived social support, must be
normally distributed. This means that for each variable, most responses should fall around the
mean so that a rough bell curve could be drawn if a histogram was depicted for each variable.
Looking at table 2 on page 699, this seems to be the case looking at the mean and the median for
each variable. They are not far from each other. There are other assumptions such as
independence of observations. This means that each participant should have only one set of data.
Each observation or measurement should not be influenced by any other observation or
measurement. We can assume that none of the assumptions for using Pearson’s product moment
correlation coefficient were violated, as the researchers reported valid results and did not have to
fall back on Spearman rho which is the non-parametric equivalent.
In looking at the results of the correlation analyses on page 698 (and table 3 on page 699) the
researchers state that “CES-D scores were significantly correlated with both systolic…and
diastolic blood pressures.” Putting it this way can be a little misleading. In interpreting results,
the r value is examined, as it tells the strength and direction of the relationship between the
variables. In this case .20 for systolic blood pressure, and .18 for diastolic blood pressure.
Pearson's correlation coefficient (r) for continuous (interval level) data ranges from -1 to +1,
where r = -1 suggests that the data lie on a perfect straight line with a negative slope; r = 0
suggests that no linear relationship between the variables; and r = +1 indicate that the data lie on
a perfect straight line with a positive slope. Positive correlation indicates that both variables
increase or decrease together, whereas negative correlation indicates that as one variable
increases, so the other decreases, and vice versa.NURS 589 Review of the statistical methodology used in a research paper
To determine the strength of the correlation Pallant (2007) suggests that r=.10 to .29 is a little
strength; r= .30 to .49 is moderate strength; and r= .50 to 1.0 is very strong. (Pallant, 2007) For
example the strength of correlation of r=.5 and r=-.5 is the same, it is only in a different
direction. Going back to the results of the paper the correlation between blood pressure and
symptoms of depression is of little strength. This is however a positive correlation, suggesting
that if symptoms of depression increases, blood pressure increases, and vice versa. The
correlation between depressive symptoms and social support was the strongest one – r= -.44 –
showing moderate strength. This is however a negative correlation indicating that as social
support decreases depressive symptoms increases. With p ranging from less than .001 to less
than .05, we can be very confident with these results. As shown in table 3 there is a weak
negative correlation between blood pressure and social support (r=-.10 for systolic blood
pressure and r=-.09 for diastolic blood pressure) but we cannot be at least 95 percent confident
that these results are valid because p is greater than .05.
The researchers did not talk about the shared variance in their interpretation of the results of
Pearson’s product moment correlation coefficient analysis. (Pallant, 2007) According to Pallant
(2007) this is important because it gives an impression of how much variance the two variables
share. This is done by calculating the coefficient of determination. The squared correlation
coefficient (r2NURS 589 Review of the statistical methodology used in a research paper
) is the proportion of variance in Y that can be accounted for by knowing X.
Conversely, it is the proportion of variance in X that can be accounted for by knowing Y. In the
case of depression and systolic blood pressure (r=.20) they share only 4 percent of their variance.
There is not much overlap between the two variables. With depression and social support
however, nearly 20 percent shared variance. This means that perceived social support helps to
explain nearly 20 percent of the women’s score on the CES-D scale. This is important.
Multiple linear regression analysis is the second statistical tool employed by Wu, Prosser, and
Taylor (2010). (Marston, 2010; Pallant, 2007) It makes sense that the researchers also used this
statistical technique in their analysis as it is based on correlation but allows for a more
sophisticated interrelationship among the variables. Correlation and regression analysis are
related in the sense that both deal with relationships among variables, but regression goes beyond
correlation by adding prediction capabilities. Simple regression is used to examine the
relationship between one dependent and one independent variable, while multiple linear
regression is used there are two or more independent variables and one dependent variable.
After performing an analysis, the regression statistics can be used to predict the dependent
variable when the independent variable is known. In this case the researchers, likely based on the
empirical evidence in the literature review, concluded that there were other variables besides
depressive symptoms that affected blood pressure in African American women. These other
variables – termed confounding variables – were identified as age, body mass index, income,
education antihypertensive medications, and social support. They are primarily interested in the
relationship between depressive symptoms and blood pressure, but they have to adjust for these
other variables. They want to find out how all these variables by themselves or together cause
variations in blood pressure in African American women. There is an assumption that all these
confounding variables also have a linear relationship with blood pressure. The researchers
therefore have to look at all these linear relationships and find the one that best predicts the Y
variable (dependent variable in this case blood pressure) as a linear function of the X variables
(all the independent variables mentioned above). The researchers are hoping that the best fit is
going to be the relationship between depressive symptoms score and blood pressure.
Furthermore the researchers want to determine if they know a particular African American
woman’s CES-D score, can they predict her blood pressure. They are attempting to estimate or
predict the value of one variable based on the knowledge of other variables. This is why they
used multiple linear regression analysis.NURS 589 Review of the statistical methodology used in a research paper
Again the researchers displayed valid results so we can assume that none of the assumptions of
multiple linear regressions were violated. (Marston, 2010) Marston (2010) explains the
assumptions as: each independent variable must have a linear relationship with the dependent
variable; for each value of the independent variable, the dependent variable should be normally
distributed, and the standard deviation should be the same; like correlation analysis each data
value should be independent of each other. Additionally, the dependent variable – blood
pressure- is a continuous variable, which is a requirement for doing multiple regression analysis.
The results are displayed in tables 4 and 5 (page700). Like the researchers point out there is a
significant relationship between symptoms of depression and systolic blood pressure, but no
association with diastolic blood pressure. It would appear however, that age seemed to be the
best predictor of increased systolic blood pressure. BMI has just as much of an effect as
symptoms of depression. This was mentioned briefly by the researchers. In this case the results
displayed in a coefficients table. If the researchers wanted to, they could now use this
information to formulate a prediction equation for how these variables (the ones with a
significant effect on blood pressure) can be used to predict blood pressure.
Logistic regression analysis is the third statistical test used by the researchers. (Marston, 2010;
Pallant, 2007) It is used explore the relationship between an independent variables and a
dichotomous dependent variable. It allows the researcher to assess how well independent
variables predict or explain the categorical dependent variable. It assesses ‘goodness of fit’. In
this case the researchers are assessing whether higher depressive symptoms scores will lead to
high blood pressure or not in African American women. Here blood pressure is the dependent
dichotomous variable and the categories are having high blood pressure and not having high
blood pressure.NURS 589 Review of the statistical methodology used in a research paper
The results are displayed in table six on page 701. These results were very significant and very
predictive hypertension in African American women who had depression. According to the
results, we can be at least 95 percent confident that African American women with a high
depressive symptoms score is 3.7 times more likely to have high blood pressure than women low
scores. Additionally even women with moderate depression were 3 times more likely to have
high blood pressure than those who did not have depression. The results also indicated that older
women were 1.1 times more likely to have hypertension than younger women.
In conclusion, the study is successful because it used the appropriate statistical tests to analyze its
findings. It achieved its purpose of uncovering the relationship that depression and perceived
social support have on blood pressure in African American women. It is important for patients
and clinicians to know that depression contributes to hypertension. This is an important
component to consider when formulating a treatment plan. The researchers also thought that
perceived social support affects blood pressure. Their research seems to support an indirect
relationship. Social support had a significant negative effect on depression, which in turn affects
Marston, L. (2010). Introductory statistics for health and nursing using SPSS. Thousand Oaks:
Sage Publications Ltd.
Pallant, J. (2007). SPSS survival manual: A step by step guide to data analysis using SPSS for
Windows. Maidenhead: Open University Press.
Wu, C., Prosser, R., & Taylor, J. (2010). Association of depressive symptoms and social support
on blood pressure among urban African American women and girls. Journal Of The
American Academy Of Nurse Practitioners, 22(12), 694-704. doi:10.1111/j.1745-
NURS 589 Review of the statistical methodology used in a research paper