Sampling is the process of systematically selecting representative elements of a population. When these selected elements are examined closely, it is assumed that the analysis will reveal useful information about the population as a whole.
The systems analyst has to make a decision on two key issues. First, there are many reports, forms, output documents, memos, and Web sites that have been generated by people in the organization. Which of these should the systems analyst pay attention to, and which should the systems analyst ignore?
Second, a great many employees can be affected by the proposed information system. Which people should the systems analyst interview, seek information from via questionnaires, or observe in the process of carrying out their decision-making roles?
The Need for Sampling
There are many reasons a systems analyst would want to select either a representative sample of data to examine or representative people to interview, question, or observe. They include:
- Containing costs.
- Speeding up the data gathering.
- Improving effectiveness.
- Reducing bias.
Examining every scrap of paper, talking with everyone, and reading every Web page from the organization would be far too costly for the systems analyst. Copying reports, asking employees for valuable time, and duplicating unnecessary surveys would result in much needless expense. Sampling helps accelerate the process by gathering selected data rather than all data for the entire population. In addition, the systems analyst is spared the burden of analyzing data from the entire population.
Effectiveness in data gathering is an important consideration as well. Sampling can help improve effectiveness if information that is more accurate can be obtained. Such sampling is accomplished, for example, by talking to fewer employees but asking them questions that are more detailed. In addition, if fewer people are interviewed, the systems analyst can afford the time to follow up on missing or incomplete data, thus improving the effectiveness of data gathering.
Finally, data gathering bias can be reduced by sampling. When the systems analyst interviews an executive of the corporation, for example, the executive is involved with the project, because this person has already given a certain amount of time to the project and would like it to succeed. When the systems analyst asks for an opinion about a permanent feature of the installed information system, the executive interviewed may provide a biased evaluation, because there is little possibility of changing it.