Elements of Research

Mass Communication Research
Elements of Research

UNDERSTANDING CONCEPTS AND CONSTRUCTS

Concept: A term that expresses an abstract idea formed by generalizing from particulars and summarizing related observations. Important for two reasons:

simplifies research process by combining characteristics or objects into general categories
facilitates communication among people who have a shared understanding of the terms.

Construct: Three distinct characteristics:

an abstract notion that is broken down into dimensions represented by lower-level concepts or a combination of concepts
because it is abstract it cannot be directly observed
it is designed for a particular research purpose so that its exact meaning only relates to the context in which it is found.

UNDERSTANDING VARIABLES

Variable
Phenomena and events that can be measured or manipulated in research

Operational definition
A definition that specifies patterns of behavior and procedures in order to measure or experience a concept.

Causal
Independent variable
Systematically varied by the researcher

Dependant variable
Is observed and their values are presumed to depend on the effects of the independent variables.

Antecedent variable
A variable used for predictions or assumed to be causal, sometimes called the predictor.

Criterion variable
A variable that is predicted or assumed to be affected.

Precision of Measurement

Discrete variable
A variable that can by divided into a finite number of indivisable parts (e.g. - political affiliation, population or gender).

Continuous variable
Takes on any value (including fractions) and can be meaningfully broken into smaller subsections.

Variables measured at nominal level are always discrete; at the ordinal level they are generally discrete though there may be an underlying continuous measurement dimension. Variables measured at interval or ratio levels can either be discrete or continous.

UNDERSTANDING MEASUREMENT

Measurement — Assigning numbers to observations according to rules.

Qualitative
A description or analysis of a phenomena that does not depend on the measurement of variables.

Quantitative
A description or analysis of a phenomena that involves specific measurement of variables.

Levels of measurement

You will come to appreciate the importance of understanding level of measurement as the semester progresses. Only certain levels of measurement can be used to get viable results from some descriptive statistics and some statistical tests.

Ratio
Has the property of an interval scale and has a true zero point. A good example is GPA. You can have a zero GPA and equal intervals. Annual income is good too. You can earn zero dollars or $100,000. If you make $100,000 that's twice as much as $50,000, so you have equal intervals.

Nominal
Arbitrary numerals or other symbols are used to classify persons, objects or characteristics. Gender is a good example of nominal data. You have two possible values: male and female. The researcher assigns the numbers. Year in school is good too. Freshman, sophmore, junior or senior.

Ordinal
Items are ranked on a continuum. A good example of ordinal data is the college football polls. The No. 1 team can have 1,212 points, the No. 2 team 1,200 points, and the No. 3 team 900 points. You have rank order but no equal intervals.

Interval
A scale in which the intervals between adjacent points are equal. Temperature is a good example of interval data. IQ is good too. You can't have a zero IQ so to speak. The zero is not meaningful.

SPSS allows us to get the mean for all the data we gather. But how valuable would the mean be for a question regarding gender with the responses being 1=male and 2-female? We know the mean would fall between 1 and 2. But could gender be 1.5? No.

So, knowing what tests to use for which type of data is critical. First, though, you have to be able to recognize the level of measurement of the data. Here is a table that should help.

Levels of Measurement

Level/ Property	Nominal	Ordinal	Interval	Ratio
Can Have Categories	Yes	Yes	Yes	Yes
Has Intervals	No	Yes	Yes	Yes
Has Equal Intervals	No	No	Yes	Yes
Has True Zero Point	No	No	No	Yes

Reliability and Validity

Components of Reliability

Stability - consistency of a result or measure at different points in time
Internal Consistency - consistency of performance among items composing a scale. Cronbach's alpha uses the analysis of variance approach to assess the internal consistency of a measure.
Equivalency - relative correlation between two parallel forms of a test, a.k.a. cross-test reliability. If two or more observers judge the same phenomenon (e.g. content analysis), intercoder reliability assesses the degree to which a result can be achieved or reproduced by other observers.

Assessing Validity of Measures - A valid measuring device measures what it's supposed to measure.

Judgment-based	Criterion-based	Theory-based
Face validity	Predictive validity Concurrent validity	Construct validity

If you don't understand something in this Web note, please e-mail Dr. Sitton.

©M. Mark Miller & Ronald W. Sitton 2009
Revised 092811 — http://www.uamont.edu/FacultyWeb/sitton/crz/mrea/elements.html