Data gathering Techniques

Research basics: choosing a data-gathering technique and crafting questionnaires

The following excerpt explains two precursors to a project's success: choosing a data-gathering technique and crafting a questionnaire.

Choose your data-gathering technique based on the project's needs. Such techniques include in-person interviews, questionnaires sent via regular mail, phone interviews or use of the Internet. Keep in mind that different cultures and subcultures respond differently to various techniques, and that you should not limit yourself to one technique if combining two or three methods provides optimal results.

In-person interviews.
Researchers who conduct in-person interviews can benefit by:
* Receiving high contact and response rates
* Conducting long, in-depth interviews
* Supplementing participants' responses with observations of them
* Allowing respondents a chance to interact with stimuli (e.g., pictures, products)

At the same time, in-person interviews have their disadvantages. Such interviews:
* Can take weeks to complete
* Are expensive (P200.00- P500.0-0 plus per interview)
* Fail to contact gated communities
* Can be skewed by possible interviewer bias through expressions, comments, etc.

Mail.
Like in-person interviews, questionnaires sent via regular mail provide participants the chance to submit thoughtful responses, since they have time to look up information and reflect before responding, plus there is a high contact rate. Mailing out questionnaires is effective for surveying homogeneous groups, and there is relatively low cost involved with surveying a wide geographic area. This technique should not be used for general population studies, however. Some people cannot read or write well, so their responses can be unclear or skewed by misinterpretation of the question. Also, mailing questionnaires offers low return rates and can take weeks to complete. Hence, a company should entice potential participants by providing:
* An immediate prize (e.g., a small gift sticker, etc., included with questionnaire)
* A future prize (e.g., collectible gift, stamp, etc., sent after participants return questionnaires)
* A chance to win something (e.g., entering name in raffle drawing)

Additionally, the company should explain the value of the study and send a letter before the study announcing it and a reminder after potential participants receive the questionnaire with a set deadline.

Phone
While mailing questionnaires works when targeting a homogeneous group, a phone interview is the most popular method for researching general populations. This method boasts a good response rate, is effective in reaching most people and is quick and inexpensive. However, phone interviews are typically limited to 8 to 10 minutes, which demands that participants give short, top-of-mind responses. Phone interviews also limit whom you contact, since most people have answering machines or caller ID, or they mistake a researcher for a telemarketer.

Database generators
For a more accurate depiction of continuing behavior, you may want to merge existing information the company has about its customers' lifestyles with information obtained by tracking how often they use checks, credit cards, discount cards, etc. Likewise, employee databases provide a way to assess productivity by tracking their sales volume and expenses, for instance. Also, you can combine employment statistics (e.g., length of employment, job classification) with behavioral statistics (e.g., use of in-house communication, e-mail, Internet) to develop communication.

Internet tracking
Internet tracking that records how often a web site is "hit" can be skewed by search engines that count a hit even if a person does not see the page. Also, a hit does not indicate that a person paid attention or found the information useful. To address this, install a cookie on the computer that keeps track of an audience's behavior while on the web site.

Internet and e-mail
Using the Internet and e-mail to gather information is inexpensive and quick, and offers immediate analysis of perhaps thousands of responses. This technique, useful for surveying within an organization, is not suggested for general population studies, as it is limited to those with Internet access. Setting up such a complex database can prove costly, and privacy concerns arise. Researchers who use this method encounter similar problems as those who use regular mail.

Consider the following:

* How to control who gets a questionnaire
* How to prevent the same person from responding twice
* Whether all computers format the questionnaire correctly

DEVELOPING A QUESTIONNAIRE

When designing a questionnaire, keep in mind:
* Questions and response categories embody your variables.
* Response categories quantify variables for analysis.
* Standardizing how data is collected reduces bias and allows the study to be replicated.
Keep in mind also that a general population study is usually limited to an eight-minute phone call or four-page mail.

Types of questions

Researchers typically ask questions leading to responses that are fixed or open-ended. Fixed responses, used when you know the likely answers, are easy to code and analyze. Consider open-ended questions when you are uncertain of the response. Such questions should be used sparingly because they take time away from other short questions, and answers must be categorized. These questions should be written with "probes" (e.g., "Are there any other issues?") that allow respondents to elaborate.

Rules of thumb. These rules will make questionnaires more efficient and less biased. Sometimes there is a good reason for breaking these rules, but for the most part, you should:
* Keep it short, simple and to the point
* Write clearly in the language of the target population
* Keep a logical order for asking questions
* Save sensitive items for the end
* Avoid leading/loaded questions or combining multiple questions into one

Pre-test. Have people fill out the questionnaire or answer on the phone, and ask them at the end if they found anything difficult to answer. Make all necessary adjustments, and pretest the questionnaire again.