Microsoft Word research method fm doc

2. Cross-tabulations 
Further analysis of the data usually requires the combination of information on two or more 
variables in order to describe the problem or to arrive at possible explanations for it. 
For this purpose it is necessary to design CROSS-TABULATIONS. 
Depending on the objectives and the type of study, two major kinds of cross-tabulations may 
be required: 
• 
Descriptive cross-tabulations that aim at describing the problem under study.
• 
Analytic cross-tabulations in which groups are compared in order to determine 
differences, or which focus on exploring relationships between variables.
When the plan for data analysis is being developed, the data, of course, are not yet available. 
However, in order to visualize how the data can be organized and summarized it is useful at 
this stage to construct so-called DUMMY cross-tabulations. 
A DUMMY TABLE contains all elements of a real table, except that the cells are still empty. 
In a research proposal dummy tables should be prepared to describe the study population in 
order to show the crucial relationships between variables. 
Some practical hints when constructing tables: 
• 
If a dependent and an independent variable are cross-tabulated, the headings of the 
dependent variable are usually placed horizontally, and the headings of the 
independent variable vertically.
• 
All tables should have a clear title and clear headings for all rows and columns.
• 
All tables should have a separate row and a separate column for totals to enable you 
to check if your totals are the same for all variables and to make further analysis 
easier.

---

Research methodology 
77
• 
All tables related to a certain objective should be numbered and kept together so the 
work can be easily organised and the writing of the final report will be simplified.
To further analyse and interpret the data, certain calculations or statistical procedures must 
usually be completed. Especially in large cross-sectional surveys and in comparative studies, 
statistical procedures are necessary if the data are to be adequately interpreted. Statistical 
tests should, for example, indicate whether differences are true differences or due to chance. 
When conducting such studies it is advisable to consult a person with statistical knowledge 
from the start in order that: 
• 
correct sampling methods are used and an appropriate sample size is selected;
• 
decisions on coding are made that will facilitate data processing and analysis; and
• 
a clear understanding is reached concerning plans for data processing, analysis and 
interpretation, including agreement concerning which variables need to be cross-
tabulated.