Measurement in Research¶
“Measurement is undertaken to facilitate adequacy, uniformity, comparisons, consistency, accuracy and precision in describing and assessing concepts”
Measurement can either be quantitative (numerical) or qualitative (symbolic), though this will focus on quantitative measurement
Theory, Conceptualisation and Operationalisation¶
Theory is an explanation of a natural or social behaviour, event or phenomenon.
Concepts are the building blocks of theory, they are ideas expressed in symbols, formulae or words. They act as the bridge between theory and research
Some concepts can’t be measured directly, so we need to conceptualise the idea into something more tangible. How you conceptualise will impact on how you measure.
To operationalise is to identify the indicators of a concept that can be directly measured. Concepts can often be multidimensional and will have multiple indicators
Concept: academic performance
Conceptualisation: academic performance can be nominally defined as the level of achievement in an educational institution
Indicators: Test results, assignment results, final grades
This is an important practice, and we need to thoroughly conceptualise our concepts to determine useful measures.
The practices of conceptualisation and operationalisation ensure that we understand the significance of the work that we’re doing. It demonstrates how the work fits in to the bigger picture of the theory.
Variables are empirical constructs that take on more than one value or intensity. A variable can only relate to one concepts (or it may be confounding) and must ultimately be measurable.
All variable have attributes, which are the values or categories of the variables
Variable: eye colour
Attributes: blue, green, brown, hazel, grey
Variables can be:
- Continuous or discrete
- Continuous variables have an infinite number of values and numbers that flow along a continuum e.g. PM 2.5 (ppm)
- Discrete variables are separate, countable and often refer to attributes. They may act as categorisation e.g. Visibility rating
- Dependent (DV) or independent (IV)
- The dependent depends on the independent
- Extra variables that you are not directly interested in however are a part of your study
- Extraneous variables that may interfere the DV and IV relationship
- They can impact on the validity of the study
A hypothesis is a tentative explanation of the research problem, a possible outcome of the research or an educated guess about the outcome.
More simply put, it’s a statement that proposes a causal link between two variables. it offers a clear framework for how we can collect, analyse and interpret the data.
Criteria for a Hypothesis¶
- Must have at least two variables
- Must express a causal relationship between two variables
- Can be expressed as a prediction or an expected outcome
- Must be logically linked to a research question and to theory
- Must be empirically testable and falsifiable
- Must focus on only one issue
- Must be clear, specific and precise
In science we have a few more things to consider. A hypothesis must use deductive reasoning (If… then… logic) and must serve as a tentative answer to a well framed question.
Failure to falsify the hypothesis does not prove the hypothesis.
Validity is a measure of precision accuracy and relevance, though it also refers to the connection between the indicator/concept connection
It answers the question; do the instruments/indicators measure what they’re supposed to?
There are three types of validity:
- Internal validity - to what extent can we draw conclusions about cause and effect
- External validity - how generalisable are the findings
- Construct validity - the extent to which the IV and DV represent the theoretical constructs that were intented
Is a measure of objectivity, stability, consistency and precision that measures the quality of the indicators and instruments. It refers to the ability of the experiment to come to the same conclusions every time.
It answers the question: does the instrument/indicator produce consistent results
Levels of Measurement¶
There is a hierarchy of the value of the orders of measurement that go as such:
- Ratio - The values have a meaningful zero, are continuous in scale and can be given as a ratio
- e.g. temperature (K), weight, distance
- Interval - these are the differences between two values, the deltas. There is no absolute zero though is still quantitative
- e.g temperature (C), Gibbs free energy
- Ordinal - These are discrete values, though there is an importance associated with them, one value is bigger/better than another
- e.g. Letter grades (F, P, C, D, HD), opinion scales
- Nominal - essentially categorical data without any sense of order. All categories have equal importance.
- e.g. gender, race