measurement.ppt
- 2. Importance of measurement • research conclusions are only as good as the data on which they are based • observations must be quantifiable in order to subject them to statistical analysis • the dependent variable(s) must be measured in any quantitative study. • the more precise, sensitive the method of measurement, the better.
- 3. Direct measures • physiological measures • heart rate, blood pressure, galvanic skin response, eye movement, magnetic resonance imaging, etc. • behavioral measures • in a naturalistic setting. • example: videotaping leave-taking behavior (how people say goodbye) at an airport. • in a laboratory setting • example: videotaping married couples’ interactions in a simulated environment
- 4. Self reports or “paper pencil” measures • oral interviews • either in person or by phone • surveys and questionnaires • self-administered, or other administered • on-line surveys • standardized scales and instruments • examples: ethnocentrism scale, dyadic adjustment scale, self monitoring scale
- 5. Indirect measures • relying on observers’ estimates or perceptions • indirect questioning • example: asking executives at advertising firms if they think their competitors use subliminal messages • example: asking subordinates, rather than managers, what managerial style they perceive their supervisors employ. • unobtrusive measures • measures of accretion, erosion, etc. • example: “garbology” research—studying discarded trash for clues about lifestyles, eating habits, consumer purchases, etc.
- 6. Miscellaneous measures • archived data • example: court records of spouse abuse • example: number of emails sent to/from students to instructors • retrospective data • example: family history of stuttering • example: employee absenteeism or turn- over rates in an organization
- 7. Levels of data • Nominal • Ordinal • Interval (Scale in SPSS) • Ratio (Scale in SPSS) nominal ordinal interval ratio
- 8. Nominal data • a more “crude” form of data: limited possibilities for statistical analysis • categories, classifications, or groupings • “pigeon-holing” or labeling • merely measures the presence or absence of something • gender: male or female • immigration status; documented, undocumented • zip codes, 90210, 92634, 91784 • nominal categories aren’t hierarchical, one category isn’t “better” or “higher” than another • assignment of numbers to the categories has no mathematical meaning • nominal categories should be mutually exclusive and exhaustive
- 9. Nominal data-continued • nominal data is usually represented “descriptively” • graphic representations include tables, bar graphs, pie charts. • there are limited statistical tests that can be performed on nominal data • if nominal data can be converted to averages, advanced statistical analysis is possible
- 10. Ordinal data • more sensitive than nominal data, but still lacking in precision • exists in a rank order, hierarchy, or sequence • highest to lowest, best to worst, first to last • allows for comparisons along some dimension • example: Mona is prettier than Fifi, Rex is taller than Niles • examples: • 1st, 2nd, 3rd places finishes in a horse race • top 10 movie box office successes of 2006 • bestselling books (#1, #2, #3 bestseller, etc.) 2nd 3rd 1st
- 11. More about ordinal data • no assumption of “equidistance” of numbers • increments or gradations aren’t necessarily uniform • researchers do sometimes treat ordinal data as if it were interval data • there are limited statistical tests available with ordinal data • •Top 10 Retirement Spots, according to USN&WR Sept. 20, 2007 • Boseman, Montana • Concord, New Hampshire • Fayetteville Arkansas • Hillsboro, Oregon • Lawrence, Kansas • Peachtree City, Georgia • Prescott, Arizona • San Francisco, California • Smyrna, Tennessee • Venice, Florida
- 12. Interval data (scale data) • represents a more sensitive type of data or sophisticated form of measurement • assumption of “equidistance” applies to data or numbers gathered • gradations, increments, or units of measure are uniform, constant • examples: • Scale data: Likert scales, Semantic Differential scales • Stanford Binet I.Q. test • most standardized scales or diagnostic instruments yield numerical scores
- 13. More about interval data • scores can be compared to one another, but in relative, rather than absolute terms. • example: If Fred is rated a “6” on attractiveness, and Barney a “3,” it doesn’t mean Fred is twice as attractive as Barny • no true zero point (a complete absence of the phenomenon being measured) • example: A person can’t have zero intelligence or zero self esteem • scale data is usually aggregated or converted to averages • amenable to advanced statistical analysis
- 14. Ratio data • the most sensitive, powerful type of data • ratio measures contain the most precise information about each observation that is made • examples: • time as a unit of measure • distance as a unit of measure (setting an odometer to zero before beginning a trip) • weight and height as units of measure
- 15. More about ratio data • more prevalent in the natural sciences, less common in social science research • includes a true zero point (complete absence of the phenomenon being measured) • allows for absolute comparisons • If Fred can lift 200 lbs and Barney can lift 100 lbs, Fred can lift twice as much as Barney, e.g., a 2:1 ratio
- 16. Examples of levels of data • nominal: number of males versus females who are HCOM majors • ordinal: “small,” “medium,” and “large” size drinks at a movie theater. • interval: scores on a “self-esteem” scale of Hispanic and Anglo managers • ratio: runners’ individual times in the L.A. marathon (e.g., 2:15, 2: 21, 2:33, etc.)
- 17. Application to experimental design • As far as the dependent variable is concerned: • always employ the highest level of measurement available, e.g, interval or ratio, if possible • rely on nominal or ordinal measurement only if other forms of data are unavailable, impractical, etc. • try to find established, valid, reliable measures, rather than inventing your own “home-made” measures.