Data Collection Techniques Readings

Powell Chapter 5

A summated scale uses a set of statements or items for rating responses, but does not structure or distribute the statements along any continuum.  Likert scales are a well-known and frequently-used type of summated scale.

A semantic differential scale provides pairs of antonyms and synonyms along a rating scale that appears to follow a continuum.  These are familiar to many people, and appear frequently in survey research.

Wildemuth Chapters 18-28

Ch. 18

Transaction log anaylisis is a research method in which records of transactions or occurrences stored in a computer device are examined for patterns of usage and feedback data.  The computer system in question stores precise data about how users utilize its systems, how users move through established processes, and what resources users actually use.  This data is used to help make decisions regarding how resources might be allocated or how some processes might be improved.

Data cleansing and preparation involves discarding corrupt and/or extraneous data and processing of raw data files to build secondary data sets.

Ch. 19

Think-aloud protocols (also called verbal reports or verbal protocols) are research methods in which subjects follow given directions but must speak what they are thinking aloud as they perform the tasks necessary to complete the directions given.  They allow researchers to observe reactions and feelings and the sequence of subjects' thought processes.

Concurrent protocols ask subjects to speak while they are performing tasks, while retrospective protocols allow subjects to detail their thoughts verbally after the fact.

Ch. 20

Direct observation is the observation of subjects in situ rather than by eliciting details through "indirect" methods such as interviews or questionnaires.

Participant observation involves the observer playing some role as a participant in the setting that is being observed.  Non-participant observation occurs when the observer takes no part in the setting.

Ch. 21

Participant observation can take place when the observer engages in passive participation, participant-as-observer, or limited observation; these terms indicate that the observer is merely present in the observation setting, but takes a very limited or passive role in the setting itself.

On the other hand, complete participation, full membership, and active participation involve the researcher taking a much more active role in the observation setting.

Note: the level of participation chosen by the observer should be appropriate to the setting itself and to the information needs or research needs of the researcher.

Ch. 22

Research diaries are self-reports made by participants in a research study that occur repeatedly over time.  Subject to a wide range of structure types, the diary is a special form of data collection distinct from questionnaires, interviews, and observation in that the subject or participant self-reports.

Diaries are often combined with other research methods, and one method often employed by researchers is the diary-interview method.  This method combines the use of interviews with the use of diaries, and can, if conservatively used, clarify diary entries.  Liberally used, this method can allow much greater-depth exploration of diary entries after they are recorded.

Ch. 23

A structured interview is one with predefined questions that are asked of all participants or subjects in the same order.  An unstructured interview is more flexible and, while it may rely on a guide with suggested questions, allows for the interviewer to modify the question set and question order to gather extra information when such an opportunity presents itself.  In unstructured interviews, questions asked may often depend on context or on the answers given by participants to earlier questions.

There are three types of questions (especially crucial to unstructured interviews): 1) descriptive questions, which elicit descriptions from subjects about their activities, thoughts, etc.; 2) structural questions, which focus on how subjects organize their thoughts and knowledge; and 3) contrast questions, which ask subjects to make comparisons and discuss the meanings of the situations in which they are placed.

Ch. 24

A semistructured interview is one that allows for some movement of question order and some flexibility with questions asked of subjects, but also has essential questions which must be asked because they are the foci of the research at hand.  Other questions might be added (which can be classified as "extra", "throw-away", or "probing").  Semistructured interviews vary somewhat, but fall toward the center of a continuum between completely structured interviews and completely unstructured ones.

Problematic questions should be avoided, and include questions that elicit inappropriate emotional responses (affectively worded questions), questions that incorporate two issues in a single question (double-barreled questions), and questions that are too complex or ambiguous in nature (complex questions).

Ch. 25

A focus group is a group that is put together by research with the specific task of discussing and commenting on a research topic or subject based on their personal experience.

When creating a focus group, the control characteristics are the characteristics of the people that the researcher would like to include in the focus group.  These most often involve figuring out the characteristics of who might be most useful or insightful during participation in a given study.

Ch. 26

A survey instrument is the item or set of items that is used to generate responses from participants in a research survey.  Examples would include interview questions, questionnaires, and statements that might elicit reactions that could be recorded and later analyzed.

Pretesting is the process of survey instrument review by experts or members of the target group.  Pilot testing is the administration of a survey to a sample of the target group in a realistic way.  Both of these processes assist in evaluating the survey instrument itself.

Ch. 27

Cognitive variables are those invisible or otherwise [directly] unobservable variables that revolve around a subject's cognitive style or method of thinking while participating in a study.

Affective variables are similarly unobservable, but involve the feelings, attitudes, etc. that occur at different stages of the information-seeking process.

Both of these types of variables are expressed by researchers in the form of constructs, or defined and operationalized sets of measurement characteristics that can be expressed concisely.

Ch. 28

A scale is a logical structure that expresses or embodies the relationships between numbers and the presence of whatever construct might be in use.  Put another way, once a construct is developed, results of research involving the construct produce numbers which are measured and compared by a scale.

A measurement's reliability is in its consistency.  That is, scores or measurements should not vary unreasonably as a result of factors irrelevant to the research at hand.  Consistency without such variation is expressed as reliability.

References

Connaway, L. & Powell, R. (2010). Basic research methods for librarians. 5th ed. Santa
Barbara, CA: Libraries Unlimited.  

Wildemuth, B. (2009). Applications of social research methods to questions in information and library science. Westport, CT: ABC-CLIO, Incorporated.