A data dictionary is part of your research methodology. Your project dictionary or code book lists each variable's name with description, format -- numeric, alphanumeric or alphabetic -- date obtained, group and individual identifiers, location of the data in the database and notes. Descriptive data from surveys, coded grounded theory, pre-test and post-test or other experimental design are checked for accuracy and completeness. The data is entered directly into a database and a statistical analysis package, IBM's Statistical Package for the Social Sciences, SPSS, or North Carolina State University's Statistical Analysis System, SAS.
Experimental and quasi-experimental research compares two groups, a control group and an experimental group or pre- and post-test results of the same sample. The simplest inferential test is whether there is a difference between two groups or between pre- and post-test results. Another simple inferential test is comparison of the average of results. For example, students average math scores before and after completing a computerized program of study indicate whether students' proficiency improved. A t-test compares the difference between two groups. Analysis of variance, regression analysis, scales and cluster analysis are other inferential statistical analysis methods.
Descriptive statistics describe each variable and the distribution, central tendencies and dispersion of each variable. Distribution is the range of each variable and the frequency of its occurrence. Graphs and charts display the distribution of data. Central tendencies are mean, median and mode. Dispersion is the distance of a point from the central tendency. Variance is calculated as difference from the point to the mean, median or mode. Standard deviation is calculated by subtracting the average of the numbers from the number itself. Variance is the value minus the mean squared, divided by the number of variables minus one. The square root of the variance is the standard deviation.
The conclusion of a quantitative research study implies relationships within the data. Descriptive statistics determine the accuracy of the conclusion. If the data is internally and externally valid and the constructs of the study are valid, the conclusion is probably accurate. Possible conclusions are whether a hypothesis is true or false. Causal relationships within the data are calculated to a degree of accuracy and reliability. If the improvement because of computerized study is pronounced, it is easier to determine the benefit of the study, but the improvement might be because of another variable, such as different test questions. Mild improvement occurs by chance.