STATISTICAL TESTS IN BIOMEDICAL RESEARCH
In medical research, a number of statistical tests are available to be used for different purposes. These purposes are closely linked with your study design, and of course, in the format of your recorded data.
Basically, you can record your data variables in three different format: 1) continuous variables, b) categorical variables, and c) ordinal variables. You need different tests to analyse your data in these three different modes of variables.
Once you have your data recorded in the particular form, you frequently need to manipulate these raw data to make the variables suitable for statistical treatment. It is also usual to create new variables from your data, e.g. to built up a set of categorical variables from your continuous variables.
To do that, you need to know how the statistical software operates. Above all, however, you need specific skills to plan, which type of statistical tests are suitable for analysis of your particular data, and which are unsuitable.
Here is a simple list (by no means complete) of statistical tests that the commonly used statistical software will perform, once properly programmed.
- Types of data
- Descriptive statistics
- Chi-square test
- Fisher's exact test
- Student's t-test
- ANOVA
- Confidence intervals
- Dummy variables
- Linear regression
- Multiple regression
- Parametric and non-parametric tests
- Equivalence of t-tests, ANOVA, and regression
- The Wilcoxon rank sum test
- The Mann–Whitney U test
- The Wilcoxon signed rank test
To help your orientation, a short piece of basic information
of these frequently used tests was made. You can download this
PDF document here:
In addition to these "conventional" type of data analysis from your study results, a variety of sophisticated cost-benefit and cost-effectiveness analysis have become popular in evaluating the impact of some medical intervention on health economics. This is certainly true with the new drugs; to be safe and effective is not enough, it must be cost-effective as well.
A powerful software is TreeAge 14.0 Pro which is capable in making a wide variety of modeling (Markov-, Monte Carlo-, etc.) and decision analysis (Sensitivity-, Cost-Effectiveness) needed by health economists. The most relevant of these tests are:
- Sensitivity analysis
- Markov cohort analysis
- Cost-effectiveness analysis
- Threshold analysis
- Monte Carlo Simulation
You can read more of these powerful tools by making a click here.
