Handout #7: - Measurement
Data are pieces of information that any particular situation gives to the observer. They are limited by measurement, and may be substantial (variables) or insubstantial (concepts).
Insubstantial measurement exists only as concepts (opinions, ideas, feelings and other intangible entities). e.g. status while substantial measurement is measurement of observable objects or variables and has an obvious basis in the real world. e.g. area, height.
A concept is an abstract word that refers to the property of subjects e.g. weight and status. Concepts are not measurable so they are usually defined by precise reliable observations. The process of formulating and clarifying concepts into measurable variables is called conceptualisation. The concept “status” may be conceptualised to observable ownership of property. Operationalisation is the process of developing an expression of what a concept really means through the use of operational definitions. e.g. Upper class means owns home and Lower class means rents home.
A variable that is influenced by another variable is called a dependent variable while a variable that influences another is called an independent variable. Weight (dependent variable) varies with time (independent variable). A variable may be continuous (infinite number of possible values) or discrete (finite number of possible values).
All measurement falls into one of four categories or scales. These are:
Nominal – This is the lowest scale of measurement. It divides data into discrete categories that limits the data to these categories. It does not give any indication of value or rank. e.g. black, white, blue or male, female.
Ordinal – This not only divides data into categories but also indicates order or rank. e.g. first, second, third and unskilled, semiskilled, skilled. The distance between first and second or between unskilled and semiskilled has no meaning.
Interval scale – The interval between categories has meaning on this scale. The IQ scale and the Celsius scale are examples of interval scales. A change in temperature from 10oC to 30oC is the same as that from 40oC to 60oC. However, because an interval scale does not have a true zero, it is not correct to say that 40oC is twice as hot as 20oC.
Ratio scale – The ratio scale has a true zero and it is possible to say that a value on the ratio scale is twice as large as another. Measurement of weight using a scale is an example of a ratio scale. It is correct to say that a weight of 50 kg is twice as large as 25 kg.
Leedy and Ormrod (2001) citing Senders (1958), summarise the differences as follows;
If you can say that
· One object is different from another you have a nominal scale;
· One object is bigger or better or more of anything than another, you have an ordinal scale;
· One object is so many units (degrees, inches) more than another, you have an interval scale;
· One object is so many times as big or bright or tall or heavy as another, you have a ratio scale.
VALIDITY AND RELIABILITY OF RESEARCH
Research should be repeatable and should produce interpretable results from which meaningful and accurate conclusions can be drawn. In other words research should be reliable and valid.
Validity – This concerns the interpretability and accuracy of research findings.
If a research project is designed to determine the effects of drug use on mental ability, it is valid if meaningful and defensible conclusions can be drawn from the data collected. In other words the study should address the problem.
Research validity is dependent on internal and external validity.
Internal validity – This concerns the ability of the method to produce interpretable results. Interpretability may be affected by:
i) Factors not being properly isolated. Faults in the method of the drug use study may result in other factors such as socio-economic background or educational background masking the effects of the factor under study making it impossible to interpret the results. In a cause-and-effect study, all factors except the one under study must be kept constant.
ii) Subjects behaving differently because they are in a study.
External validity – This concerns the degree to which the findings can be generalised. If the subjects in the drug use study are adult males 18-25 years old and attending UTECH, the research findings are valid for that group only. It would be inappropriate to use the findings to predict the behaviour of all males in Jamaica. If Jamaicans are the intended population then the subjects should represent Jamaicans. The sample must be representative of the population or condition for the research to have external validity.
Laboratory experiments have external validity concerns since the results may not apply to the “real world”. This is why laboratory results are field-tested.
Reliability – is a measure of the repeatability of research. A research project is reliable if each time it is conducted similar findings are obtained.
Internal reliability is concerned with repeatability due to the instrument used or agreement between observers. Questions about the ability of a measurement instrument to generate consistent results or about two observers getting the same readings are questions about internal reliability.
External reliability is concerned with repeatability by other researchers in the same or other situations. To ensure external reliability the procedures and conditions must be clearly and completely stated. Vague terminology should be avoided.