Data Processing
Data Entry Process
Data processing goes through the following stages:
· Data reading: To minimize errors, it is important to give clear instructions to the data entry agents. Verification may be done through each survey sheet or, in the case of very large databases, through statistical samples.
· The description of the variables by preliminary statistical analysis (mean, standard variation, minimum, maximum, mode, median). Many errors and deficiencies in data collection can be identified during data analysis. The best data validation goes effectively through their analysis, and that is the approach we are intending to adopt. The entry mask systems do not detect all data entry errors, and it is common to find anomalies when interpreting the results.
· Development of cross-tabulation: Among all cross-tabulations of variables implemented during processing, we will only keep those setting a significant causal link among two variables, allowing a focus for reflection for the next phases.
· The analysis of interrelationships between variables is a means of identifying new avenues of work, highlighting lines for consideration and/or action.
· The delivery of findings is to summarize the data information on a reduced number of dimensions reflecting at best the proximities between observations and/or variables. One of the great
difficulties of statistical analysis involving a large number of variables is to deliver a sufficiently clear and synthetic summary on all variables.
Documentation in PDF