Data Collection Techniques

Powell and Connaway:

Sources of error: Error with questionnaires occur frequently, but they are not limited to just questionnaires.

-Researcher bias: When the researcher develops the questionnaire so they receive the desired results.

-Sponsorship bias: When a researchers caters to the sponsors and skews the results.

-Imperfection of design: Weaknesses in the design of the questionnaire can result in innaccurate results.

-Respondent interpretations: Interpretation of "facts" may be skewed.

-Time lapse: Answers to questions tend to vary over time.

-Circumstances: Mood and careless answers can skew results.

-Response bias: Number of respondents may be too low and skew results.

-Reactive insight: Questionnaires can bring up sensitive topics which cause the survey taker to create a bias based on their reaction.

Final editing - There are a couple of suggestions that will make your final survey desirable and easy to take. Making the survey as short as possible will encourage accurate and well thought out answers. Always ask questions that the researcher doesn't already know. Unnecessary and redundant questions should be avoided.

Wildemuth:

Data Collection - Should be a relatively straightforward activity in the case of server-side transaction log studies. Researchers should monitor data logging and ensure that data are being captured as expected. (Wildemuth pg.170)

Transaction Log Analysis - The captured data represent a record of events as they actually occurred, without re-framing and recall errors prevalent in many other data collection methods. The quality of your data will not be dependent on the study participant's memory of the interaction or on his or her ability to describe the interaction. (Wildemuth pg.167)

Think-aloud Protocol - A research method used to understand the subjects' cognitive processes based on their verbal reports of their thoughts during experiments. You request subjects to speak aloud, reporting what they are thinking while they are performing tasks during an experiment. (Wildemuth pg. 178).

Types of think-aloud protocol:

-Concurrent protocols - they allow the subject to first complete tasks without saying anything

-Retrospective protocols - allows the subjects to complete the tasks in a more natural way