A data matrix is a two-dimensional table with rows containing entities and columns containing attributes. In social sciences, entities are usually respondents, and attributes are research variables. Entities do not have to be only respondents, they can also be products of human activity. For example, if the quality of a product is being checked, the rows of the data matrix would be the products, and the columns would be the product properties that are used to check the quality.

In all quantitative research, after defining the research problem, determining the sample and other necessary elements of the research design, the data collection phase follows. In order for the collected data in the research to be available for statistical processing, they must first be entered into the data matrix.

A data matrix consists of columns, rows and matrix cells located at the intersection of a row and a column. The columns contain the variables used in the study, and the rows contain the respondents. The cells of the matrix contain codes, numbered responses of the respondents to a specific question from the questionnaire, measurement results, and observations. The table provides an example of a data matrix used in one study.

No. Sex Age Education Question1 Question2 ... Questionn
1. 1 1 6 3 104    
2. 1 2 7 3 108    
3. 1 1 8 4 108    
...              
4. 2 2 5 4 107    
5. 2 1 6 1 93    
6. 1 1 10 2 103    
...              
n. 1 3 6 2 109    

It is common for the initial columns in a matrix to be independent and/or descriptive research variables, and then dependent variables. From the columns of this matrix, it can be assumed that a questionnaire was used in the research and that it is a survey research.

Variables in the data matrix are usually designated by a name or abbreviation that is associated with the variable itself. If the variable is too long (e.g. “Do you like going to school?“) it is usually named by the ordinal number from the questionnaire. Modern statistical data analysis software (SPSS, R, Python, Stata) allows variable name lengths of 64 to 256 characters, or even more.

Respondents are identified by their serial number in the matrix. It is recommended that the questionnaire of a particular respondent or the instruments completed by one respondent be marked with the same number, so that in the event of an error in data entry, the source of the error can be easily identified.

Matrices are created as two-dimensional. If there are multiple dimensions in the study, this is solved by adding new rows or columns. For example, in longitudinal studies, in addition to respondents and variables, there is also time. Variables of the second or nth measurement are entered in new columns.