Data collection technique

Characteristics

Strengths

Weaknesses

Unstructured interviews
(conversation style)

Questions emerge from the immediate context and are asked in the natural course of things; there is no predetermination of question topics or wording.

Increases relevance of questions; questions can be matched to individuals and circumstances. Good for exploration of new topics.

Different information has to be collected from different people with different questions. Less systematic and comprehensive. Data organisation and analysis can be difficult.

Semi-structured interviews
(with interview guide)

Topics to be covered are specified in advance, but the interviewer decides on the sequence and wording of the questions during the interview.

More systematic and comprehensive, but interviews still fairly conversational and situational.

Important topics may be inadvertently omitted. Interviewer flexibility in sequence and wording of questions can result in different responses from different perspectives, thus reducing comparability.

Standardised open-ended interview

Exact wording and sequence of questions determined in advance. Worded so that responses are completely open-ended.

Increased comparability of responses; data complete for each respondent. Facilitates data organisation and analysis. Best once a fairly good understanding of the topic has been obtained.

Little flexibility in relating interview to particular individuals and circumstances; standardised wording of questions may constrain and limit naturalness and relevance of questions and answers.

Free listing

Asks informants to name all items in an informant-defined domain.

Useful first step in all research involving definition of new domains. Helps ensure that concepts are culturally relevant.

Possible difficulties involved with finding appropriate generic terms to start the listing process. Sometimes list too sparse.

Pile sorts

Participants sort cards or other elements according to their differences or similarities.

Helps to discover relations among items within a domain. Can be conducted using large numbers of items and are easy to administer.

Unless items are pictures or actual stimuli, pile sorts cannot be done with non-literates. Also not appropriate for all domains since not all have underlying taxonomic structure.

Triad

Informants are asked to choose, for a set of 3 items, the one item that is the most different.

As with pile sorts, they help to discover relations among items within a domain. Easy to administer, can be used with non-literates.

Not practical for comparisons among large numbers of items.

Rating scales

Respondents are asked to rate items one-by-one according to a pre-designed scale that can be graphical or numerical.

Scales can be created for any number of concepts or attributes. They are easy to administer.

Requires literate informants or visual stimuli. Sensitive to response bias.

Rank order methods

Informants are asked to rank 2 or more items in terms of a specific characteristic.

Very productive for the short time it takes to conduct. Good for studying individual differences. Paired comparisons are especially easy to administer and can be used with non-literates.

If there are many items it can become tedious for non-literates.

Focus Groups

A semi-structured interviewing technique, which relies on discussion among participants.

Can produce a lot of information quickly. Good for identifying and exploring beliefs, attitudes and behaviours and for identifying relevant questions for individual interviews.

Provides no information abut frequency or distribution of beliefs or behaviours. Are difficult to conduct; require skilled moderators. Participants may influence each others’ responses so care must be taken in analysing results.

Group interviewing techniques
(non-focus groups)

A more formal, group interview technique, which tends to use pre-existing natural groups. Sometimes group members are asked to work together to complete a task, such as building a map of the community. The purpose is to get community actively involved in identifying and exploring topics of interest.

Good for rapport building, identifying and exploring community problems and priorities, raising awareness of local problems, involving people in identification of potential solutions. Fun and interesting for participants.

Less systematic than focus groups, so inter-group comparisons are ore difficult. Requires preparation, but perhaps less skilled group moderators than for focus groups. Raising awareness can lead to raised expectations – researchers should be careful not to create unfulfilled expectations.

Social network analysis

A set of methods for discovering and describing social structures and patterns of social interaction. Based on individual interviewing techniques and participant observation.

Can identify important social groups; provide information about value, meaning and boundaries of network relationships; describes network-related attributes of individuals in a population; describes the contacts and ties that relate individuals or group to one another.

Some methods may require intensive long-term relationship building with informants. Other methods are lengthy and boring to administer. Data analysis is fairly sophisticated.

Participant observation

Not really a data collection technique, but rather an approach whereby the researcher becomes an active member of the culture under study. Involves mainly unstructured observation and unstructured interviewing.

Facilitates all other data collection activities by building rapport and reducing reactivity. Helps to formulate relevant questions. Good for understanding processes, events and relationships in their social context.

Can be time consuming, requires that the researcher speak the local language very well and be a skilled observer and note taker.

Unstructured observation

Observer is outsider, non-participant. What is observed is only generally defined. Broadly focused, aim is to observe behaviour in holistic context.

Good for discovering unknown aspects of a problem. Allows for discovery of “surprises”. Good for understanding behaviours in their physical and social context.

Do not provide precise, repeatable measures of behaviour, and thus cannot be used to monitor changes in behaviour.

Structured observation

Observer is outsider, non-participant. Observes and records observations in a pre-determined manner.

Provides precise numerical data which are amenable to statistical analysis and can be repeated to monitor change over time.

The problem to be studied needs to be well-defined. Training of observers is intensive and time-consuming. Pre-determined structure of observations limits discovery of other potentially relevant behaviours.

Ethnographic decision-tree modelling

Uses interviews and participant observation to elicit people’s own decision criteria. These are then combined in the form of a decision tree or flowchart. The model is then tested with separate sample of decision-makers

Ethnographic decision-tree modelling shares the strengths and weaknesses of the qualitative data collection methods it draws on. It is a systematic means for understanding how people make decisions; good for gaining an understanding of observed behaviours and for predicting behaviour.

Modelling process can be difficult of researcher has no prior experience. Requires knowledge of computers for testing models.