Biostatistics for the Clinician Hypertext Glossary Section II Glossary

George Oser, Ph.D., Craig W. Johnson, Ph.D. Allan J.Abedor, Ph.D.

Biostatistics for the Clinician Term: Biostatistics for the Clinician Ordinal Variables Ordinal variables, the second level of measurement, and the highest level of qualitative variable, typically are used to order, or rank values of variables in addition to naming the values. Ordinal scales have all the characteristics of nominal variables but in addition, order or rank the data. Consequently, ordinal variables allow "less than" or "greater than" assessments to be made among the values of a variable. Interval Variables Level of Measurement Nominal Variables Qualitative Variable Ratio Variables Types of Variables Variable
Biostatistics for the Clinician Term: Biostatistics for the Clinician Parametric Tests Parametic statistical procedures are inferential statistical methods designed to be used with normally distributed quantitative (interval or ratio) variables. Examples of parametric procedures include the t- Tests, and the Pearson Correlation Coefficient. Chi-square Tests Inferential Statistics Nonparametric Population t-Tests.
Biostatistics for the Clinician Term: Biostatistics for the Clinician Poisson Distribution The family of Poisson distributions is a category of discrete frequency distributions like the binomial distribution showing distributions of events having two possible outcomes, like success or failure. But, unlike the binomial distribution, it requires not two but just one parameter to compute probabilities. It turns out that this means the clinician can quickly use it as a simple method to closely approximate the binomial distribution for distributions of rare events (i.e., distributions where there are very many events, but the probability associated with a specific dichotomous outcome on any one event is small - like begin struck by lightening). Binomial Distribution Gaussian Distribution
Biostatistics for the Clinician Term: Biostatistics for the Clinician Population A population is a collection of objects, events or individuals having characteristics that a researcher may be interested in studying. To study a population, the researcher typically selects a small group, called a sample, from the population. Power Random Sample Sampling Distribution
Biostatistics for the Clinician Term: Biostatistics for the Clinician Power Statistical power, or just "power", is the probability of rejecting the Null Hypothesis when it is false. In other words, when a relationship exists between the independent and dependent variables in the Population, statistical power measures the probability of detecting that relationship from the sample data. Dependent Variables Independent Variables Null Hypothesis Statistical Significance
Biostatistics for the Clinician Term: Biostatistics for the Clinician Qualitative Variables Qualitative variables are those variables which are either nominal variables or ordinal variables. Interval Variables Level of Measurement Nominal Variables Ordinal Variables Quantitative Variables Ratio Variables Types of Variables Variable
Biostatistics for the Clinician Term: Biostatistics for the Clinician Quantitative Variables Quantitative variables are those variables which are either ordinal variables or interval variables. Interval Variables Level of Measurement Nominal Variables Ordinal Variables Qualitative Variables Ratio Variables Types of Variables Variable
Biostatistics for the Clinician Term: Biostatistics for the Clinician Random Sample A random sample is a sample chosen from a population in a fashion that ensures every object, event, item or individual has an equal chance of being drawn. The selection of any one entity can in no way influence or affect the selection of any other. Most statistical inferential procedures assume that researchers use random samples. The validity of such procedures rests upon the assumption that samples are genuinely representative random samples from populations. Population Sampling Distribution
Biostatistics for the Clinician Term: Biostatistics for the Clinician Range The range of a distribution is one of the simplest measures of variability of a distribution. It is simply the difference between the maximum and minimum values in the distribution. Interquartile Range Standard Error Standard Deviation
Biostatistics for the Clinician Term: Biostatistics for the Clinician Ratio Variables Ratio variables, the fourth and highest level of measurement, have all the properties of interval variables, but in addition have the property that the ratio variable has a meaningful absolute zero. The 'zero' is not arbitrary, as in the case of interval scales (e.g., fahrenheit and celsius temperature scales), but represents the complete absence of any amount of the variable. Interval Variables Level of Measurement Nominal Variables Ordinal Variables Qualitative Variable Types of Variables Variable
Biostatistics for the Clinician Term: Biostatistics for the Clinician Regression Analysis Regression analysis is an inferential statistical method that develops equations from empirical random samples to make predictions about the values of a dependent variable based on the values of one or more independent variables with known probabilities of accuracy. If there is more than one independent variable the method is referred to as "multiple regression". Residual Variable
Biostatistics for the Clinician Term: Biostatistics for the Clinician Relationships among Variables Scientific research investigates relationships among variables. Such relationships take the form that as one variable increases another tends to increase, or as one variable increases another tends to decrease. Theses relationships may be causal, meaning that the changes in one variable depend on the changes in another; or they may be correlational, meaning that the variables tend to change at the same time, but there is not necessarily a causal relationship between the two variables. The relationship between the foot size and vocabulary in a population of grade school students provides an example of a non-causal correlational relationship. Causal Relationship Variable
Biostatistics for the Clinician Term: Biostatistics for the Clinician Research Hypothesis The research hypothesis is also called the alternative hypothesis. It is the opposite of the null hypothesis. The research hypothesis states there is a relationship between the independent and dependent variables. When the null hypothesis is rejected, based on research data, it implies acceptance of the research hypothesis. Alpha Inferential Statistics Null Hypothesis Population Power.
Biostatistics for the Clinician Term: Biostatistics for the Clinician Residual A residual is the difference between the predicted value (often from a regression equation) and the actual or observed value. Examination of residuals in regression analysis will identify atypical cases. Regression Analysis
Biostatistics for the Clinician Term: Biostatistics for the Clinician Sample Size Sample size is the number of subjects (people, plants, etc.) in a group selected from a population. Because sampling error tends to be smaller for larger samples, larger samples have more statistical power. Power Population. Sampling Distribution Standard Error
Biostatistics for the Clinician Term: Biostatistics for the Clinician Sampling Distribution Whenever random samples of a given size are taken repeatedly from a population of scores and a statistic (e.g., the mean) is computed for each sample, the distribution of this computed statistic may be constructed. The resulting distribution is called a sampling distribution (e.g., the sampling distribution of the mean). Population. Random Sample
Biostatistics for the Clinician Term: Biostatistics for the Clinician Statistical Significance The statistical significance, or level of significance, is the probability of rejecting a true null hypothesis. That is, it is the probability that an investigator will conclude that a relationship exists between the independent and dependent variables when no such relationship exists, a false positive. It is represented by the lowercase Greek letter alpha. Alpha Null Hypothesis
Biostatistics for the Clinician Term: Biostatistics for the Clinician Skewness A distribution that is not symmetric is skewed. That is, a skewed distribution is a lopsided distribution. In such distributions the mean, median and mode are typically all different. The more lopsided the distribution is the more skewed it is. Box Plot Distribution Exploratory Data Analysis
Biostatistics for the Clinician Term: Biostatistics for the Clinician Standard Deviation The standard deviation is a measure of variability. It provides a measure that can be thought of as being the "average distance" that values are from the Mean. It is expressed in the same units as the original values. Mathematically it is computed as the square root of the average squared distance from the mean of the values. Interquartile Range Population Range Standard Error
Biostatistics for the Clinician Term: Biostatistics for the Clinician Standard Error The standard error of any statistic is the standard deviation of its sampling distribution. You can find an estimate of the standard error of the mean by dividing the standard deviation by the square root of the sample size. Mean Sampling Distribution Standard Deviation
Biostatistics for the Clinician Term: Biostatistics for the Clinician Summary Statistics Summary statistics or descriptive statistics concern that branch of statistics that has as its primary focus summarizing data from groups. Such statistics enable investigators to describe with precision a collection of quantitative information in a manner that is concise, convenient, easily interpreted, and easy to communicate. For example, a clinician might choose to use the mean to summarize a large number of patients'. EXAMPLES: mean, median, mode, standard deviation, z-score. Inferential Statistics Mean Population. Standard Deviation
Biostatistics for the Clinician Term: Biostatistics for the Clinician t-Test t-Tests are used to detect significant differences in means of quantitative variables. The t-tests generally require the data to be normally distributed and from populations having equal variability unless samples sizes are approximately equal. Tests have the above characteristics are frequently referred to as parametric tests. There are three t-tests, the t-test for independent samples, for paired-samples and for the mean of a single sample. Chi-square Tests Inferential Statistics Population.