Chapter 8 Principles of Measurement Flashcards Preview

Research in Nursing > Chapter 8 Principles of Measurement > Flashcards

Flashcards in Chapter 8 Principles of Measurement Deck (53)
Loading flashcards...

Construct validity

1. Extent to which an instrument or test measures an *intended hypothetical concept or construct*.
2. The most valuable - yet the most difficult (takes years) - way to assess an instrument's validity.
3. Example: The Safe Sex Behavior Questionnaire was determined to have construct validity based on significant correlations with measures of risk-taking and self-expression.


Content validity

1. Extent to which an instrument or test measures an *intended content area*. The *most common type* of validity.
2. Determined by a *panel of experts*. Example: Experts in nutrition were chosen to assess the items in the Barriers to Healthy Eating Scale.
3. Usually this type of validity is used in the development of a questionnaire, interview schedule, interview guide, or instrument.


Continuous variable

1. Variable that takes on an *infinite number* of different values presented on a *continuum*.
2. Examples: Age in years; education in years.


Criterion-related validity

1. Extent to which an instrument or test measures a particular concept compared with a *criterion*.
2. Measured by a *validity coefficient*; a higher coefficient indicates high criterion-related validity.
3. Example: A new instrument is compared with an older, more reputable instrument.
4. Two types: *Concurrent and predictive.*


Cronbach's alpha (coefficient alpha)

1. Widely used index of the extent to which a measuring instrument is *internally stable*.
2. A measurement of the extent to which all the items in an instrument measure the same *concept*.
3. An acceptable level of reliability for any measurement instrument is an alpha coefficient of *0.70*.
4. Can be used with instruments composed of items that can be scored with three or more possible values, such as a Likert-type scale.


Dichotomous variable

A *nominal variable* that consists of *two* categories.



1. A device, piece of equipment, or paper-and-pencil test that *measures* a concept or variable of interest.
2. Examples from nursing research: Questionnaires, surveys, and rating scales.


Interval level of measurement

1. Level of measurement characterized by a *constant unit of measurement or equal distances between points on a scale*.
2. Possesses all characteristics of a nominal and ordinal scale in addition to having an equal interval size based on an actual unit of measurement, but *lacks* a true zero point.
3. May be referred to as "interval data" or "interval variables."



1. The assignment of numerical values to concepts, according to *well-defined rules*.
2. The numerical values reflect *properties* of the concepts under study.


Nominal level of measurement

1. The *lowest level of measurement*, which consists of assigning numbers as labels for categories. These numbers have *no numerical interpretation* (i.e., they are not stating that one category has "more" and one category has "less").
2. Data simply show the *frequency of subjects or objects* in each category.
3. May be referred to as "nominal scale," "nominal data," "nominal measurement," or "categorical data."


Operational definition

1. A definition that *assigns meaning* to a variable and the terms or procedures by which the variable is to be *measured*.
2. Especially important for quantitative research. Usually found in the *methods* section.
3. Concepts such as "spiritual well-being" must be translated to measurable definitions that are valid reflections of the concepts.


Ordinal level of measurement

1. Level of measurement that yields *rank-ordered data* (i.e., highest to lowest, most to least).
2. Specifies the *order* of items being measured, *without* specifying how far apart they are.
3. May be referred to as "ordinal scale," "ordinal data," "ordinal variables," or "ordinal measurement."


Psychometric evaluation

Evaluating properties of reliability and validity in relation to instruments being used to measure a particular concept or construct.


Predictive validity

1. Ability to predict future events, behaviors, or outcomes.
2. Example: A university admissions committee uses applicants' GRE scores to decide whether to admit them to a graduate program, based on the idea that a high GRE score is predictive of success in their program.


Ratio level of measurement

1. The *highest level of measurement*, characterized by equal distances between scores having an *absolute zero point*.
2. The zero point indicates *absolutely none of the property*.



1. Value that refers to the *consistency* with which an instrument or test measures a particular *concept*. Different ways of assessing reliability include test-retest, internal consistency, and interrater.
2. An instrument with high reliability should yield essentially the same scores each time the test is administered.
3. Asks, "Is it consistently generating the same measurements?".


Test-retest reliability

1. An approach to reliability examining the extent to which scores are *consistent* over time (their *stability*).
2. A researcher measures a group of individuals twice with the same measuring instrument or test, with the two testings separated by a particular period of time.



1. Value that refers to the accuracy with which an instrument or test *measures what it is supposed to measure*. Different types of validity include content, criterion, and construct.
2. Asks, "Are the measurements meaningful?".


Relationship between reliability and validity

1. *Both* are essential to the *meaning and accuracy* of the results produced by an *instrument*.
2. They are the two *most important issues* to consider when examining the worth of any *instrument* used to measure variables in a study.
3. Reliability refers to whether an instrument generates consistent measurements; validity refers to whether an instrument measures what it is supposed to measure.
4. An instrument's data *must* be reliable if they are to be valid; an instrument *can* be reliable without being valid.
5. They are *context/population-specific* - an instrument designed for premenopausal women is not reliable and valid for homeless youth, unless *tested and proven*.


In order to make sense out of data collected, each variable must be _

*Defined operationally* using a variety of measurement techniques.


The selection of a measurement technique depends upon _

The particular *research question* and the availability of *instruments*.


The four levels of measurement

*In order of increasing sophistication*:
1. Nominal (lowest level - numbers are merely labels for categories).
2. Ordinal (more precise than nominal, but does not lend itself well to statistical analysis).
3. Interval (equal interval sizes and units allow addition, subtraction, and computation of averages, but there is *no* true zero point).
4. Ratio (highest level - has an absolute zero point; data can be manipulated using *any* arithmetic operation).


Examples of nominal data

The numbers are labels arranged in random order, and do not mean that one category is "more" or "less" than another:
1. Gender (1 for males, 2 for females).
2. Religion (1 for Catholic, 2 for Protestant, 3 for Jewish).
3. Blood type (1 for type AB, 2 for type A, 3 for type B, 4 for type O).


If there are only two categories associated with nominal data (such as gender), then the variable of interest is said to be _ in nature.



Many instruments and scales used to measure _ variables yield ordinal data.

Psychosocial. (For this reason, ordinal data are common in nursing research.)


Examples of ordinal data

Only the order of categories is important, as intervals between ranks are *not* equal:
1. Satisfaction (1 for "very satisfied;" 2 for "somewhat satisfied;" 3 for "not satisfied").
2. Education (1 for 6 years or less; 2 for 8 years; 3 for 11-12 years; 4 for 13-16 years).
3. Household income (1 for $0-$4,999; 2 for $5,000-$9,999; 3 for $10,000-$19,999; 4 for $20,000-$29,999; 5 for $30,000-$49,999).


Many instruments and scales used in nursing research report "scores" and are usually referred to as _

*Interval* data or *interval* variables. (Many highly reliable *psychosocial* instruments or scales yield interval data.)


Example of interval data

Numerical values with equal interval sizes and an actual unit of measurement, but no meaningful zero point:
Temperature (Fahrenheit or Celsius: 0° does not indicate an "absence" of temperature; 90°F is not "twice as hot" as 45°F).


Examples of ratio data

Categories are different, ordered, separated by a constant unit of measurement, and have an absolute zero point:
1. Age (e.g., number of years).
2. Time (e.g., number of hours).
3. Weight (e.g., number of kilograms).
4. Length (e.g., number of inches).


From a statistical standpoint, _ and _ measurements are treated as the same.

Interval and ratio. (The mean and standard deviation can be calculated for both of these.)