Biostatistics Primer: Part 2

Abstract

Biostatistics is the application of statistics to biologic data. This article is the second part of a 2-part series on the application of statistics in nutrition science. The first article, published in the December 2007 issue, reviewed descriptive statistics. Inferential statistics, to be discussed in this article, can be used to make predictions based on a sample obtained from a population or some large body of information. It is these inferences that are used to test specific research hypotheses. This article focuses on inferential statistics and their application in the nutrition and biomedical literature. Additionally, this review will outline some of the most commonly used statistical tests found in the biomedical literature.

An Investigation into Equations for Estimating Water Requirements and the Development of New Equations for Predicting Total Water Intake

“…The criteria for agreement were twofold: 1) if the pair of variables demonstrated a strong (r >0.7) or medium (0.4 < r < 0.7) degree of association (Overholser & Sowinski, 2008), and 2) if the regression line produced by the pair of variables was not significantly different from the line of equality with the CI of the slope and intercept including one and zero, respectively (Volaufova, 2005). Criteria two would only be performed on those variables meeting criteria one; they demonstrated a strong (r >0.7) or medium (0.4 < r < 0.7) degree of association.…”

“…The amount of shared variance found for each of the comparisons was detected using r 2 and the associated pvalue. A correlation coefficient matrix compared Pearson's correlation and the associated p-value to determine the degree of association between each pair of equations (Overholser & Sowinski, 2008). If the pair of variables demonstrated a large shared variance and a strong or medium degree of association as described above, criteria one would be satisfied.…”

“…CI of the slope and intercept had to include one and zero, respectively (Volaufova, 2005). If the pair of variables demonstrated a weak degree of association (r <0.3), this would not meet the criteria for agreement and further testing would not occur (Overholser & Sowinski, 2008). In this case the two variables would not be in agreement with one another.…”

“…If the pair of variables demonstrated a large shared variance and a strong or medium degree of association as described above, criteria one would be satisfied. If the pair of variables demonstrated a weak degree of association (r <0.3), this would not meet the criteria for agreement and further testing would not occur (Overholser & Sowinski, 2008).…”

“…Correlation coefficients of 0.4 -0.7 were considered medium associations. Correlation coefficients >0.7 were considered strong associations (Overholser & Sowinski, 2008).…”