Ana-Maria Simundić | PubMed Health

PubMed2006

[Types of variables and distributions].

Variable presents any measurable quality of an individual or a system. Data are collected by measuring variables. There are four scales of measurement: nominal, ordinal, interval and ratio. There are qualitative and quantitative variables. Categorical variables are measured with nominal and ordinal, while numerical variables are measured with interval and ratio scale of measurement. Sample data are usually grouped around some central location value. At the same time there is always a higher or greater variability present in the sample. Measures of central tendency and variability are ways of summarizing data. Distribution characteristics are of utmost importance because they form a basis for the choice of descriptive statistics and significance tests. Normally and not-normally distributed data are tested with parametric and nonparametric tests, respectively. Equality of variance is also a criterion for the choice of significance test. Computer programs offer a full range of statistical procedures and calculations. In order to use them it is necessary to be familiar with some basics in biostatistics. This lecture is an introduction to the types of data, levels of measurement, ways of summarizing data and types of distributions.

Analysis of Variance↗

PubMed2006

[Diagnostic accuracy].

Clinical decision making while making a diagnosis is based on pre-test probability evaluation, a decision whether further tests are needed, and post-test probability evaluation based on test results. The add value of the test is the increase in probability, which depends on its diagnostic accuracy. Some measures of diagnostic accuracy are sensitivity, specificity, positive predictive value, negative predictive value, likelihood ratio, area under the curve, Youden index, and diagnostic odds ratio. ROC curve is the most efficient way of graphic description of the relationship between sensitivity and specificity. Diagnostic accuracy of a test is evaluated in a cross-sectional study on a group of individuals subjected simultaneously to the index test and the gold standard. All parameters of diagnostic accuracy are very sensitive to the study design, therefore it is of utmost importance to perform and write a study to adhere to the proposed STARD standards. This lecture gives an insight into the basics of biostatistical terminology from the field of diagnostic accuracy.

Biometry↗