The basics of survey processing > Tables and cross-tabulations
Tables
Survey processing is the statistical exploration of the data; it involves observing the variables via simple tabulations (or hole-counts) and cross tabulations. Hole-counts are simply the variables as they appear in the questionnaire, and more of a checking tool. This is why DAISIE produces them systematically each time a variable is created. Cross tabulations show the same type of information, but on several sub-populations simultaneously thereby enabling a comparative analysis.
Cross tabulations
Cross tabulations are basically tables of data obtained by crossing one variable with another. One variable is put in columns and the other in rows. Generally speaking, one of the dimensions is used for the variables to observe and the other for socio-demographic criteria like sex, age, social bracket, etc.
The initial output is often all the questionnaire crossed with several socio-demographic criteria. It’s up to the user to decide on the form of their tables. If they want one table per variable, they will no doubt prefer to put the socio-demographic criteria in rows. Otherwise, they will put the socio-demographic criteria in columns and obtain a long table with all the variables appearing one after the other. The fact that the variables have a “kind” means that the calculations to display can be defined in advance, and different kinds of variables put end to end. For example, a continuous variable, the calculation for which might be a mean score, can be put after a logical variable, the calculation for which might be percentages.
DAISIE memorises queries which is not only a way of checking requests, but also of modifying or exporting them from one study to another.
Intelligent tables
Intelligent tables are tables that are clear and precise, thereby enabling full and immediate comprehension. They should meet these requirements in terms of both content and form.
In terms of content: by choosing the appropriate calculations and showing the calculation bases and marginals; the reader needs to know exactly what the calculations are based on.
In terms of form: via the texts associated with the variables which should be unambiguous and the style given to the table with key figures made obvious straight away through the use of statistical tests or conditional styles.
The power of queries in DAISIE
The DAISIE software program is unequalled when it comes to creating and producing cross tabulations. Cross tabulations are produced by automatically stored and re-usable queries. The same query can produce a group of tables with common characteristics. Dynamic queries cross all the variables in the study with one or several sort criteria.
All statistical calculations are available in the cross tabulations of DAISIE. Their layout in the cells of the table is dictated by formats. Statistical tests like for example the Chi-Square test or Student T-test, can make reading the tables easier.
Examples of intelligent tables
- Clear labels for the variables and a conditional style that highlights the mean scores from 7 to 10 in shading from yellow to green.
- Statistical tests that use upper / lower case letters and bold type to indicate the level of significance.
