Scales

875412_330130201.jpgScales are a research tool, often used in surveys, which prompt informants to respond in structured ways to a set of related questions. The use of scales can allow researchers to quantify seemingly abstract, subjective data. Scales measure both the direction and intensity of an individual’s thoughts, feelings, attitudes, experiences and more about a variable. Collectively, groupings of individual’s data can be used to reflect wider understandings of beliefs or behaviors that can be presented on their own or used to guide subsequent methodology. Either a single question or multiple questions are used to understand a variable, depending on the complexity. Although an existing scale may be adapted for a particular context, construction of a new scale may be more appropriate for a unique population study or assessing anthropologically relevant variables. For example, scales might be used to order the importance of governmental responses to a community need or estimate the satisfaction with an agencies response to a situation.

Relevant characteristics


  • Allows for measurement of perceptions, cognitions, opinions and other latent constructs that cannot be directly measured.
  • Can be designed to allow for items to be directly compared (comparative scaling) or independently scaled (noncomparative scaling).
  • Within those categories, information is collected primarily in four different ways.[1]
    • Nominal – observations placed into one exclusive category. Ex. “male” or “female”
    • Ordinal- observations placed in a “ranking” order with no concern given to the amount of difference between the rankings. Ex. Preference rankings
    • Interval- observations placed along a scale with the distance between units distributed equally over all levels of the scale with no absolute zero point. Ex. Fahrenheit Temperature
    • Ratio- observations placed along a scale with the distance between units distributed equally over all levels of the scale with an absolute zero point which allows for ratio comparisons. Ex. Time (10 hours is twice as long as 5 hours).


Examples of Scales

  • Likert scale is a format in which a statement or question is presented and a rating scale indicates level of agreement. Likert scales are typically used for multidimensional variables.[2]
  • Guttman scale is a format in which a statement or question has a binary outcome (yes/no). A Guttman scale is typically used to measure unidimensional variables.[3]
  • Semantic differential scale is a format in which the target item is provided and paired adjectives are used along with a rating scale.[4] .

likert.gif



Method Made Easy

Scale Construction

  1. Identify the variable to be measured. This should be found by understanding your research question and choosing several key themes to be explored.
  2. Use focus groups to generate questions based upon the key themes identified. Focus groups should allow the researcher to align the perceived research objectives with the social reality. It gives the researcher opportunity to develop methodology, including the use of scales, to maximize the collection of the informant’s perspective.
  3. Analyze transcripts from focus groups and mine for possible scale questions. This is an opportunity to further discover research themes. These themes should direct the questions used in the scale construction. Also important will be understanding what the scale results will be used for.
  4. Determine type of scale to use. Will it be a comparative or a noncomparative scale? Will the desired data be nominal, ordinal, interval or ratio in type?
  5. Construct scale. In close relationship to the previous step is the determination of both the questions and responses to be used in the scale. The research themes already identified should drive the question construction as will the desired research objectives. Some things to consider include.[5]
    1. How many divisions should be used? (1-4; 1-5; 1-10; etc)
    2. Should you allow for a neutral center value? An odd number division will allow for a neutral response if that type of response is desired.
    3. How should the scale be designed physically? (vertical, horizontal, linear, graphical)
    4. Should responses be forced or will a null response be an option?
    5. Determine the type of statistical analysis that will be used. It is crucial to consider how the data will be analyzed in the construction stage to allow for a streamlined analytic phase.
    6. Pilot scale and modify as necessary. Scales are often constructed in versions that are tested on smaller sample groups and deconstructed by the researcher to consider the overall success of the scale design. It is common to find that questions may need to be re-worded, response options might need to be clarified or portions of the scale need to be completely overhauled.

Scale Administration

Much of the actual administration of the constructed scale depends on the design of the scale. Scales can be given individually or in a group setting. Today it is common to find scales administered on-line via computers. Scales given in that manner need to have a greater amount of consideration given to issues of data storage and security. Scales are versatile and are found in such diverse settings as the medical office, supermarket aisle and research labs. The keys to successful scale administration are clear instructions, clear expectations of the informant and careful scale design.

Advantages

  • Scales are a tool to quantify qualitative data
  • Scales measure degrees of a characteristic rather than presence or absence
  • Short questionnaires with between four and ten items are often sufficient
  • Scales can take advantage of differences in intensity among the indicators of the variables


Limitations

  • Participants may have a particular response style for choosing answers instead of providing accurate answers (ie. always choosing a neutral answer).[6]
  • Specific scoring criteria must be used to maximize interrater reliability
  • Minimize cultural bias in development, analysis and interpretation of scale
  • Self-report data is generated with varying degrees of accuracy


Analysis


  • Internal reliability to ensure each question measures the same variable. The researcher wants to ensure that the scale will produce consistent results. The scale should be tested-retested under similar circumstances. Also the results should be checked using different forms of the scale to further ensure reliability.[7]
  • Use Cronbach's alpha to measure internal reliability. Do all questions in the scale measure the same variable? Cronbach's Alpha using SPSS
  • Face validity to determine if the scale measured the intended variable. This can be done by using “known-groups technique” which requires the researcher to administer the scale to groups expected to give different results. The desired population should give results that differ from the groups with non-relevant characteristics[8]
  • Statistical analysis of results using statistical software (SPSS, ANTHROPAC). These statistical software packages allow for computation of complicated statistical measurements and will also generate the data in presentable format. Tables and graphs can be populated that assist in validating research questions. A short video is included in the “Online Resources” that offers a beginners’ tutorial to SPSS.


pie_likert.png


Method in Context

  • Scale construction may be laborious; however the results obtained may provide a clearer picture of the study population. Proxy questions are used to identify information related to the variable of interest.
  • Focus groups used for question development are a guided interview instrument to generate perspectives and conceptualizations
  • Scales allow measurement of both frequency and subjective experience for life events at both the individual and cultural levels
  • Scales are used in a variety of disciplines; anthropology, psychology, education


Online Resources



























Further Reading


Cowherd, R.E.
2012 The Effects of Food Insecurity on Mental Wellbeing in Monteverde Costa Rica. (Unpublished masters thesis). University of South Florida, Tampa

Ice, Gillian H., and Jaja Yogo
2005 Measuring Stress Among Luo Elders: Development of the Luo Perceived Stress Scales. Field Methods 17(4):394-411.

Lende, Daniel H.
2008 Wanting and Drug Use: A Biocultural Approach to the Analysis of Addiction. Ethos 33(1):100-124.

Luhrmann, T.M., H. Nusbaum and R. Thisted
2010 The Absorption Hypothesis: Learning to Hear God in Evangelical Christianity. American Anthropologist 112(1):66–78.

Snipes, Shedra A., Beti Thompson, Kathleen O’Connor, Ruby Godina and Genoveve Ibarra
2007 Anthropological and Psychological Merge: Design of a Stress Measure for Mexican Farmworkers. Culture, Medicine and Psychiatry 31(3):359-388.

Reeve, Bryce and et al.
2011 Comparing Cognitive Interviewing and Psychometric Methods to Evaluate a Racial/Ethnic Discrimination Scale. Field Methods 23(4):397-419.

Rudmin, F.
2009 Constructs, measurements and models of acculturation and acculturative stress. International Journal of Intercultural Relations 33:106-123.

Schrauf, Robert W. and Ellen Navarro
2005 Using Existing Tests and Scales in the Field. Field Method 17(4):373-393.




Discussion Board

Subject Author Replies Views Last Message
No Comments



References


  1. ^ Bernard, H. Russell
        • 2006 Research Methods in Anthropology: Qualitative and Quantitative Approaches.
        • Lanham,New York, Toronto, Oxford: AltaMira Press.
  2. ^ Bernard, H. Russell
        • 2006 Research Methods in Anthropology: Qualitative and Quantitative Approaches.
        • Lanham,New York, Toronto, Oxford: AltaMira Press.
  3. ^ Bernard, H. Russell
        • 2006 Research Methods in Anthropology: Qualitative and Quantitative Approaches.
        • Lanham,New York, Toronto, Oxford: AltaMira Press.
  4. ^ Bernard, H. Russell
        • 2006 Research Methods in Anthropology: Qualitative and Quantitative Approaches.
        • Lanham,New York, Toronto, Oxford: AltaMira Press.
  5. ^ LeCompte, M.D. and J.J. Schensul
        • 2010 Designing and conducting ethnographic research. Maryland: Atlamira press.
  6. ^ Hodge, D.R. and D.F. Gillespie
        • 2003 Phrase Completions: An Alternative to Likert Scales. Social Work Research
        • 27(1):45-55.
  7. ^ Kennedy, David P.
        • 2005 Scale Adaptation and Ethnography. Field Methods 17(4):412-431.
  8. ^ Brown, Ryan A., Carol M. Worthman, Jane Costello and Alaattin Erkanli
        • 2006 The Life Trajectory Interview for Youth: method development and psychometric
        • properties of an instrument to assess life-course models and achievement.
        • International Journal of Methods in Psychiatric Research 15(4):206-215.