The Management of Data and Variates in Epidemiological Studies

In epidemiological studies, health scientists should collect data that are as reliable and objective as possible, in order for the statistical analysis to lead to precise and valid conclusions. Mistakes in the way that data concerning the cases in a study are collected, and in the input of the data into the appropriate electronic databases, along with limited knowledge of database management, may all result in a considerable decrease in the reliability of data analysis, leading to invalid results. Before the start of statistical analysis, the correct input of the data into database that is going to be used for the analysis is essential. This involves the correct management and coding of the variates, in order to ascertain and rectify possible errors and omissions. The variates (and by extension the data as a whole) are divided, according to their mathematical properties, into qualitative and quantitative; the former are divided into nominal and ordinal, and the latter into interval scale variates and ratio scale variates. Quantitative variates are also divided into continuous and discrete. Missing values and outliers have a particular meaning in data analysis. Observations that have not been made on some variates are called missing values. An increase of the proportion of missing values in the analysis leads to decrease in credibility of the results of the analysis. Outliers are values of observations that are surprisingly extreme when compared to the values of the other observations. Outliers may be due to either errors in observation recorded during data collection or incorrect input of the observations into the database, or they may be actual values that just differ considerably from those of the other observations.