Over my career – whether working in corporate finance, marketing management, or in patient advocacy and public health education – I have probably relied on the analysis of quantitative data using descriptive, inferential, and multivariate statistics more than any other single skill set.
Something as simple as frequency distributions can render clarity and understanding to data because some degree of order is imposed. The shape of the distribution of values and variability can thus be observed, by a systematic arrangement of values from lowest to highest together with a count of the number of times each value was obtained. This allows analysis of percentages of occurrence, in which the sum of all percentages equals 100. Graphs, such as histograms, can be constructed to instantly communicate patterns of intervals, while frequency polygons are graphs enabling us a measurement of the mean. From this, we can assess variability and calculate standard deviation, so critical to quality management.
Descriptive statistics (percentages, means, standard deviations, and so forth) are used in a wide variety of applications. In health services research, for example, they are used most often to summarize sample characteristics, describe key research variables, and document methodological features such as the response rate by a population to an intervention. In business, a company’s employee attrition rate may be calculated routinely to determine the percentage of employees that left the business over a specified period of time, usually one year. Attrition includes all employees who leave the company, whether the leaving was voluntarily and involuntarily. Such a statistic is useful to determining whether a turnover problem is growing worse or better.
A common clinical research situation in which quantitative data are analyzed using inferential statistics involves comparing two groups of subjects on a dependent variable. An example could be the comparison of an experimental group to a control group of patients on a physiologic measure such as blood pressure or heart rate. Using t-tests, the statistical significance of differences between the two group means can be determined. Inferential statistics, based on the laws of probability, allow us to make inferences about a population based on data from a sample and thus help with sample size requirements to insure that one does not fail to detect an effect that is present, or Type II error, also known as a ‘false negative.’ In the judicial system, in which a person is considered innocent until proven guilty, a Type II error would be akin to allowing a guilty person to be set free. In addition to applications in medical testing, there are broad applications of inferential statistics in computer security including spam filtering and character detection. Security screening applies such statistical analysis in metal and explosive detection, for example, with sensitivity settings sufficiently high to prevent a would-be terrorist from slipping through as an ordinary tourist, at the expense of inconveniencing a number of passengers being tested.
Multivariate statistical procedures help us untangle complex relationships among three or more variables, allowing us to make predictions about one variable based on the values of a second variable or on the basis of two or more independent variables. Factor analysis is a data reduction technique in which a researcher reduces a large number of variables to a smaller, more manageable, number of factors. Factor analysis uncovers patterns among variables and then clusters highly interrelated variables, allowing a market researcher, for example, to drill down through a lot of information about population to identify key characteristics. Logistic regression is useful in predicting the risk of an outcome occurring given one condition, versus the risk of it occurring given a different condition. One of the most commonly used statistical techniques, logistic regression is used by lenders in credit scoring, by product managers in predicting product revenues, and by sports journalists in predicting the outcome of a soccer match.
Here in the Lowcountry, we are fortunate to have the strength of statistical wizardry in the form of College of Charleston professors Martin Jones and Jim Young, whose expertise and graduate course offerings are new at the Lowcountry Graduate Center. We welcome them as a pathway for enhancing your knowledge base and job advancement across a breadth of applications. Imagine what a diverse group of professions one of their classes could represent!
Nancy Muller, PhD
Director and Associate Dean