free listing

pile sorting

rating scales

• define the domain from informants persepctive
• cultural domain- set of items or things that are all of the same type or category

• set of related items (words, concepts, sentences)
  • family planning methods
  • ways malaria can be transmitted
  • kinds of illness
  • symptoms associated with an illness
• how people assign domains tells you something about how they perceive these items, and how they interpret the meaning for their own lives

• get clear understanding of domains from the emic perspective
• ensures culturally relevant items and language
• allows the researcher to ID most salient items with little effort
• results can e used to develop follow up instruments
• its easy and can e a good ice breaker

• informants asked to list "items:
• open ended questions allow for as many items as possible
• what are all the...?
  • ways you can get malaria?
  • family planning methods you can think of
• followed by a probe: Are there any other...?
• important to pre-test the opening question
  • test wording on 4-5 people

• ID most salient items
  • items that are etter known or more familiar
  • two indices of saliency
    • position on the list

• ensures that domain and concepts are contextually relevant
• get good sense of important issues rapidly
• easy for respondents
• can prevent researchers from investigating irrelevant domains
• prevents use of irrelevant items

• can be difficult to find the right terms or questions to generate lists
• can be difficult to generate lists of items that are productive
• no recognized statistical methods for checking reliability of free lists

• usually after free list (list already generated)
• procedure where informants put together items that are similar
• items can be from free listing, observations, or existing records

• explore possile links between domain items
• to explore links between domains and another concept, item, or topic of interest
• to better understand domain boundaries
• can uncover reasons why domains are constructed the way they are- what is behind perceptions and beliefs

• place words or pictures on index cards
• shuffle/randomize cards before giving to respondents
• ask respondents to sort items according to similarity
  • constrained- researcher defines number of piles
  • unconstrained- respondent sorts into as many piles as makes sense to them
• do respondents determine the criteria?
  • yes- then elicit the criteria used after the sorting is done
  • no- make sure there is common understanding of criteria among respondents

• whats in each pile
• how many piles were made
• similarity of criteria used by respondents
• computerized analysis
  • multidimensional scaling (map with distances between items)
  • cluster analysis (which items go together)
  • factor analysis (underlying variale)

• allows similarity comparisons of a large amount of items
• easy to administer
• often a good way to open up discussion between the researcher and respondent

• need 20 or more respondents to get reliable reults
• only lookin at relationships among items- not appropriate for comparing respondents
• need pictures or actual stimuli with for non literate informants

• consist of one or more statements that respondents "rate" with scaled responses
• can rate one item on several concepts
• can rate different items on same concept
• end points are referred to as "anchors"
• rating on all items can be combine to make an index

• useful to get ordered/continuous data
• provide descriptive information
  • 20% of respondents think that malaria is a serious disease
• can be used to create an index and examine association between a domain (ex: severity of childhood diseases) and other variables (ex: health seeking behavior)
  • bed net use associated with high perceived malaria severity

• ID what you want to include
  • important items from free list
  • most similar items from a pile sort
• write the rating scale based on the concept you want the items to be ordered on (ex: severity, advantages, usefulness, agreement)
• determine whether you want neutral category- choose odd numer or ratings
• determing your scale anchor
  • numberic (scale 1-10)
  • linguistic (ex: strongly agree to strongly disagree)

• self administered
• reliability can be assessed
• familiar format for certain groups

• difficult to administer orally
• response preferences can make it harder to get reliable data and compare data across respondents
• scale construction (anchor points, size of scale, phrasing) can affect results