Biostatics and Research Design in Psychiatry Flashcards
What is the difference between descriptive and inferential statistics?
Descriptive statistics include mean, median, mode, frequency distributions, histograms, variance, SD, interquartile range, and other ways of summarizing a view of the data and calculating a single number that conveys something important about the data. Inferential statistics requires judgment and technique because a conclusion will be drawn about a population based on information in a sample that has been drawn from that population. It is based on the rules of probability and takes into account uncertainty, and chance is considered regarding the extend of the effects and the exposure can be attributed. In other words, the likelihood of cause and effect.
Dependent vs. independent variable
Dependent variable = the outcome of interest in the study, e.g. HAM-D score reduction, weight gain or weight loss.
Independent variable = the variable used to predict the value of the dependent variable, such as a treatment.
Interval data: what kind of data is this and what is an example? Why is this important?
Interval data is a type of continuous data. An example would be degrees in F or temperature. This kind of data has a zero arbitrarily assigned to it, so it cannot be used as a ratio. Interval data does not have an absolute zero.
Ratio data: what kind of data is this what is an example.
Ratio data can be used as ratios to compare outcomes, it is a type of continuous data. An example would be blood glucose or serum levels. Ratio data has an absolute zero, such as blood pressure, blood glucose, trigs.
Ordinal data: what are examples, and are the distances between points equal?
Ordinal data is a type of discrete data, meaning there are only so many values or categories. The distance between categories is not equal. For example, stage 1, 2, 3, or 4 cancer is ordinal data. Ordinal = order or rank, categories, series.
Define sensitivity
Probability of a true responder
Define specificity
Probability of a true non-responder
Positive predictive power (PPP)
The probability that a patient selected by the model will respond in the given time frame e.g. 6 weeks.