An Introduction to Scientific Research Methods in Geography Chapter 6 Explicit Reports

1. An Introduction to Scientific Research Methods in Geography Chapter 6 Explicit Reports

2. What is an explicit report? • Measures of beliefs people have about all sorts of things, including themselves or other people, places or events, activities or objects • Can request many different types of beliefs: behaviors, knowledge, opinions, attitudes, expectations, intentions, experiences, and demographic characteristics

3. What is an explicit report? • The defining characteristic is their explicitness • People know they are providing information • Cannot be judged as right or wrong • Can be characterized as common or unusual or related to variables such as demographics

4. Types of Explicit Reports • Surveys • Interviews • Sociometric ratings • Activity diaries/logs • Contingent valuation • Focus groups • Protocol analyses • Tests

5. Format of Explicit Reports • A specific survey or interview is called an instrument • The specific questions or statements that make it up are called items • Closed-ended items provide a finite number of specific response options • Open-ended items do not provide specific response options for respondents

6. Format of Explicit Reports • Standardized items are presented in a predetermined and consistent format to all respondents • Nonstandardized items are not consistent across respondents • Follow up questions are nonstandardized • Branching format • Free-format

7. Closed-ended items • Should provide exhaustive alternatives • Should usually provide mutually exclusive alternatives • Providing an “other” alternative can make it somewhat more open-ended • “Don't know” and “no opinion” avoid forced responses

8. Rating Scales • Respondents provide a number or mark a line to indicate the amount or extent of something, including the degree of belief they have in something • Semantic differential • Likert scales • Paired comparison ratings

9. Rating Scale Design Issues • How many scale options to provide? • Generally, 5 to 9 for adults, usually an odd # • Measurement level – treat as ordinal • Only non-parametric ordinal stats can be used • Measurement level – treat as interval • Used by many researchers • Believe people treat equally spaced scale intervals as approximately equally spaced • Better for visually/numerically spaced values

10. Administration of Explicit Reports • Can be self or researcher-administered • Individually or in groups • In person, through the mail, over the phone, or on the internet • Interviews could be done with a computer program or recorded with audio or video

11. Administration of Explicit Reports • 1. Cost – Can vary greatly, in money, time and effort, by administration • Interviewing more costly than surveys • 2. Number and nature of items – surveys can contain more items and more complex items than interviews • Complex interviews can be too difficult • Surveys provide more anonymity and can increase honesty

12. Administration of Explicit Reports • 3. Response rate – in-person requests for participation are most effective but are costly • People tend to ignore mail and e-mail surveys and become frustrated with phone calls • 4. Potential for follow-up

13. Administration of Explicit Reports • 5. Nature of respondents – different types of respondents differ in their ability to handle different administration methods and questions • Example: revealing some info to a non-present researcher can be strange and not work • 6. Possible interview artifacts – the appearance, demographic characteristics, or personal style of the researchers can affect responses

14. Using the Internet to Collect Explicit Reports • Major benefit is the efficiency and low cost • Can reach many, many people for very cheap • Are stored in digital form and analyzed automatically • Branching-format follow-up questions possible • However, there is concern about the presence of sampling bias

15. Other Difficulties with Internet Research • Repeated participation • Does not occur commonly and can be avoided • Respondents may take studies less seriously and provide frivolous responses • Research indicates that internet results are similar to traditional studies • Privacy issues – obtaining informed consent

16. Design and Generation • Use existing literature to generate items • Conduct one or more open-ended, unstructured interviews to generate ideas • Construction of items is important in how respondents understand and respond to them • Should avoid confusing, biased, and ambiguous wording • Items should be unidimensional

17. Design and Generation • Should be visually easy to understand and use • Should be long enough to obtain the information you need, but not longer than that • Order effects – potential influence on responses to particular items that arise due to the order the item was presented • A context effect may arise because of the context created by the order of items • The optimal solution is counterbalancing

18. GPM...PMT • Generate, pretest, modify...pretest, modify, test • 1st generate items to tap into your constructs of interest, then pretest the first draft • May pretest with friends or colleagues, but should also pretest with a small sample of the type of respondents you will use for real data • Also ask them how they interpret items, instructions, etc. • Then modify the instrument, pretest it, and modify again as many times as is necessary

19. The Census • A count of the number of people in a country & an assessment of their characteristics, such as family structure, economic activities, etc. • Secondary data source • U.S. constitution mandates a census – 10 years • American Community Survey – 1% of the population • U.S. census is neither de facto or de jure, but counts people at their “usual residence,” legal or not • What kind of problems might this create for geographers?

20. Counting issues • Accepted that the U.S. census is an undercount • Some are overcounted and others under, but the net result is an undercount • Illegal immigrants, those without permanent residences, and those of lower socioeconomic status in inner cities may be undercounted • Students and those with multiple residences may be counted twice

21. Census Units • Census data are aggregated to spatial units designed to ensure the privacy of individual respondents • The entire U.S., regions, divisions, states, counties, county subdivisions, census places, census tracts, block groups, and blocks are the basic units • Change somewhat from census to census, especially at smaller units

22. Where do I find it?

24. Limitations of Explicit Reports • Some basic limitations include: • Respondents may give socially desirable responses - called response sets • Respondents may attempt to damage the efforts of the researcher • Interviewers can distort results • Memory can be a limitation • Keep the time frame short and recent

25. Limitations of Explicit Reports • Language is another limitation • Not all meaning and experience can be verbally expressed • Language expresses meaning in a profoundly contextual way • Altering the wording of items can produce different responses

26. Limitations of Explicit Reports • Explicit reports require respondents to access their mental states and externalize them but this is complicated • “What” questions are easier than “why” questions to answer • We don't do what we do because of what we believe • Cultural norms and expectations, habits, action opportunities and constraints, personal gain or loss, past experience, and belief that one's actions are public or anonymous factor in too

27. Review Questions • How do closed-ended and open-ended items differ and what are some strengths/weaknesses of each? • What issues need to be considered when deciding how to administer explicit reports? • What are some desirable characteristics of explicit report items? • What are strengths/weaknesses of the U.S. census as a source of data for scientific research?