Measurement

Measurement• The process of assigning numbers or labels to units of analysis in order

to represent conceptual properties.• The measurement process begins with formulation of research problem

or hypothesis. • Every research problem contains terms – concepts or variables.

Cont.…• The ultimate goal of measurement is to specify

clearly observable referents of the terms contained in one’s hypothesis.• Thus, the entire measurement process consists of

moving from the abstract (concepts) to the concrete (measures of the concept.

Conceptualization• Initial step is to clarify the concepts

embedded in one’s hypothesis with words and examples.• Process of formulating and clarifying

concepts is called conceptualization. It is linked to theory testing and construction. • The conceptualization is an ongoing process

that may occur prior to any particular empirical investigation and is usually continues through research as theories and their constituent concepts are refined and elaborated.

Cont.…• The conceptualization of complex concepts such as social capital

often requires careful distinctions among similar ideas and breaking down the concept into various components or dimensions.

• Measurement assumes the possibility of assigning different values or categories to unit of analysis; hence, we measure concepts tat vary, which we refer to as variables.

• However, many social science concepts like social capital, or wealth etc are not directly measurable. These are rather manifestation of several directly measurable variables. After conceptualization , the next step is to identify such manifestations of one’s concepts.

Operationalization• Once the meaning of the concept is clarified and

the concept is constructed as variable, the process of operationalization of the measurement process begins. • First step of the operationalization is to set up an

operational definition as a counterpart of conceptual definition that corresponds reasonably good to the concept in question. • An operational definition describes the research

operation that will specify the value and category of the variable on each case.

Cont.…

• While creating operational definitions, a researcher may consider many different empirical representations or indicators.

• An indicator consists of a single observable measure, such as a single questionnaire item in the survey.

• No two indicators measure a given concept or variable in the same way, and no one indicator is likely to correspond perfectly to its underlying concept.

Cont.…• Indicators provide imperfect representations of the

concepts for two reasons:- Indicators often contain errors of classification- Indicators rarely capture all the meaning of a

concept. • Because of the imperfect correspondence between

indicators and concepts, researchers often choose to rely on more than one indicator when operationalizing a concept.

Cont.…Operationalizing the concept of religiosity• My religious beliefs are what really lie behind my

whole approach to life. (1) this is definitely not so (2) Probably not so (3) Probably so (4) definitely so.

Verbal reports• Self reports provide simple and generally accurate

measures of background variables such as age, gender, marital status and education.• Composite measures: in self report attitude

measurement, responses to several questions frequently are combines to create and index or scale.

Cont…• There are numerous composite measures of

prejudice, combining from two to twenty questions or more-• Do you think there should be laws against marriages

between blacks and whites? Yes/no• White people have the right to keep blacks out of their

neighborhoods if they want to, and black should respect that right. Do you agree strongly, agree slightly, disagree slightly, or disagree strongly with this statement

Cont…• Observation: observation provides direct and

generally unequivocal evidence of overt behavior, but it also is used to measure subjective experience such as feelings and attitudes.• Archival records: which refers to existing recorded

information, provide another invaluable source of measurement.

Level of measurement• Nominal• Ordinal• Interval• Ratio

Nominal measurement• System in which cases are classified into two or

more categories on some variable, such as gender, race, religious preference etc.

• In nominal measurement numbers (more accurately numerals) are assigned to the categories simply as labels or codes for the researcher’s convenience in collecting and analyzing the data.

• Categories of the variable should possess two characteristics: they must be exhaustive and mutually exclusive.

Ordinal Measurement• In ordinal measurement, number indicate only the rank order of cases on some

variable.• Ordinal measurement allows the researchers to make an accurate judgment

about one thing compared to another, even when they can not make an accurate absolute judgment. Ex.:

Individual’s ranking of certain leisure activities in terms of the pleasure derived from them. The three activities are ranked as:

1.Playing tennis2.Watching television3. Reading sociology , where one, two and three represent individuals’ ranking of the leisure activities

• In the realm of social measurement one can probably say with some certitude whether security or chance for advancement is the more important job characteristic, without being able to say how important either characteristic is. • The ability of human observers to make such

comparative judgments permits a wide range of reasonably accurate social measurements at the ordinal level. For example, measures of socioeconomic status, intelligence etc.

Cont.…

Interval measurement• Interval measurement has the quality of the nominal

and ordinal levels, plus the requirement that equal distances or intervals between numbers represents equal distances in the variable being measured.• For Ex. Fahrenheit temperature scale: the difference

between 20ºF and 30ºF is the same as the difference between 90ºF and 100ºF- 10ºF. We can infer not only that 100ºF is hotter than 90ºF but also how much it is.

• Though interval measurement system does not have absolute zero that is it is not a fixed but an arbitrary point.

• Therefore, we can not say that 100ºF is twice as hot as 50ºF as it will require a fixed zero point which interval measurement system does not have.

Cont.…

Ratio Measurement• Includes the feature of other levels plus an

absolute zero point. • The presence of an absolute zero makes it possible

to multiply or divide scale numbers meaningfully and thereby form ratios.• Ex. Rs. 10000 and Rs. 20000, one can divide the

other

Information provided by the four levels of measurement

Information provided

Nominal Ordinal Interval Ratio

Classification X X X X

Rank Order X X X

Equal Intervals X X

Non arbitrary Zero

X

Reliability and validity• Reliability is concerned with questions of stability

and consistency.• An ex. of a highly reliable measuring instrument is

a steel tape measure. A cloth tape measure would be somewhat less reliable.• SRS data is reliable.• Unreliable things cannot be valid.

Cont…• Measurement validity refers to the extent of

matching, congruence, or goodness of fit between an operational definition and the concept it is purported to measure.• Ex. Amniocentesis, it is a valid measure of

biological sex which can determine with virtually perfect accuracy.

Sources of errorObserved value= true value + systematic error+

random errorThe first source of variation is true differences in the

concept the operation is intended to measure.Ex. IQ test ought to reflect only true differences in

intelligence and nothing else.

Cont…• Systematic measurement error results from the

factors that systematically influence either the process of measurement or the concept being measured. • When the respondent’s sensitivity or

responsiveness to a measure is affected by the process of observation or measurement, we refer to this a s a reactive measurement effect.

Cont…• Random measurement error is unrelated to

true differences in the concept being measured. It is the result of temporary, chance factors.• Ex. A tired and bored respondent may give

erroneous responses by not attending carefully to the questions asked.• Similarly an ambiguously worded question will

produce random errors by eliciting responses that vary according to respondent’s interpretation of the question’s meaning.• Such a error is random because its presence,

extent and direction are unpredictable.

Reliability assessment• Reliability indicates consistency, or the extent to

which a measure does not contain random error.• Test retest reliability: the procedure involves

testing the same persons or units on two separate occasions.

Cont…

Low reliability High reliability High reliabilityLow validity Low validity High validityHigh random error Low random error Low random errorLow systematic error High systematic error Low systematic error

Split half and internal consistency reliability• The second set of procedures for assessing

reliability estimates the agreement or equivalence among the constituent parts or items of a multi-item measure.• Ex. Age of women and age at marriage.

Validity assessment• Reliability assessment is relatively simple, validity

assessment by contrast is more problematic. • Systematic errors which affect validity but not

reliability, are more difficult to detect than random errors.• The issue of measurement validity generally

cannot be divorced from larger theoretical concerns.

Subjective validation• Face validity: refers simply to a personal judgment

that an operational definition appears, on the face of it, to measure the concept it is intended to measure.• Content validity: concerns the extent to which a

measure adequately represents all facets of a concept.

Thank you