-
the process of assigning numbers or labels to objects, persons, states, or events in accordance with specific rules to represent quantities or qualities of attributes.
a. We measure attributes, not the person or event
Concept of measurement
-
Specific types of concepts which exist at higher levels of abstraction, usually theoretical
a. They are generally not observable but are inferred through indirect means
b. E.g., brand loyalty, Trust, Satisfaction, Self-esteem
Concept of construct
-
States the central idea or concept; Establishes boundaries for the construct
Theoretical definition
-
Defines which observable characteristics will be measured; Defines the process for assigning a value to the concept
Operational definition
-
Attitudes toward complex objects have many facets
o Unrealistic to capture full picture with one overall attitude-scale question
o Responses are combined into some form of average score.
Multi-item measures
-
measurement in which numbers are assigned to objects or classes of objects solely for the purpose of identification
nominal scale
-
measurement in which numbers are assigned to data on the basis of some order (for ex., more than, greater than) of the objects
Ordinal scale
-
measurement in which the assigned numbers legitimacy allows the comparison of the size of the differences among and between members
inverval scale
-
measurement that has a natural, or absolute, zero and therefore allows the comparison of absolute magnitudes of the numbers
ratio scale
-
Error in measurement due to temporary aspects of the person or measurement situation and which affects the measurement in irregular ways
o Statistical fluctuation that occurs because of chance variation in the elements selected for the sample
o Mood, state of health, fatigue, situation in which measure is taken, ambiguity of question wording can all bias measurement at random
Random error
-
error in measurement that is also known as constant error since it affects the measurement in a constant way.
o Caused by some imperfect aspect of the research design or a mistake in research execution
o Personality, response styles, wording of questions, method of administration can all bias measurement in systematic ways
Systematic error
-
· Non-respondents
o Refuse to cooperate
o Not available
· Self-selection bias
o Over-represents extreme positions
o Under-represents indifference
· Varies by type of interview
Nonresponse error
-
Bias that occurs when respondents tend to answer questions with a certain slant that consciously or unconsciously misrepresents the truth
Response error
-
Response bias due to some people tending to agree with all questions or to concur with a particular position
Acquiescence bias
-
Response bias due to response styles that vary from person to person; some people tend to use extremes when responding to questions
Extremity bias
-
Response bias that occurs because the interviewer’s presence influences answers
Interviewer bias
-
Response bias due to being influenced by the organization conducting the study
Auspices bias
-
Bias caused by respondents’ desire, either conscious or unconscious, to gain prestige or appear in a different social role
Social desirability bias
-
The degree to which a measure is consistent across time, evaluators, and the items forming the scale
o Similarity of results by independent comparable measures of the same construct
Reliability
-
The degree to which a measure measures what it is supposed to measure
o Is the measure “good” (a true reflection of the underlying variable or construct it is attempting to measure)
Validity
-
How well does a measure predict an outcome?
Predictive Validity
-
Do the measures look as though they would measure the construct of interest?
Content Validity (Face validity)
-
How well does the measure actually assess what it is supposed to assess? (The most difficult form of validity to establish)
Construct Validity
-
the ability of a measure to correlate or converge with other supposed measures of the same variable or construct
Convergent Validity
-
the ability of a measure not to correlate with measures from constructs with which it is supposed to differ
Discriminant Validity
-
Steps to follow in developing valid measures
- Step 1 – specify domain of the construct
- Step 2 – Generate sample of items
- Step 3 – Collect data
- Step 4 – Purify measure
- Step 5 – assess validity
-
Three attitude components
- Affective
- - Feelings or emotions toward an object
- Cognitive
- - Knowledge and beliefs
- Behavioral
- - Predisposition to action
- - Intentions
- - Behavioral expectations
-
a method of assessing attitudes that rests on the presumption that a subject’s performance of a specific assigned task (for example, memorizing a number of facts) will depend on the person’s attitude.
Performance of objective tasks
-
a method of assessing attitudes in which the researcher monitors the subject’s response, by electrical or mechanical means, to the controlled introduction of some stimuli.
Physiological reaction
-
a method of assessing attitudes in which individuals are asked directly for their beliefs about or feelings toward an object or class of objects.
Self-report
-
ask about a single concept
Non-comparative Rating Scales
-
ask respondents to rate a concept by comparing it to a benchmark
The resulting data may have only ordinal or rank order properties
Comparative Rating Scales
-
consists of a series of statements that express either a favorable or unfavorable attitude toward a concept
Likert scale
-
•A multiple-item scale consisting of a series of bipolar adjectives or phrases
•Responses are provided on a 7-category scale with no intermediate numerical or verbal labels
Semantic-differential scale
-
Variation of a semantic differential scale
substitute a single adjective for bipolar adjectives of semantic-differential scales
Stapel scale
-
•Ask respondents to rate a concept by comparing it to a benchmark.
•Ask respondents to judge each attribute with direct reference to the other attributes being evaluated.
•The resulting data may have only ordinal or rank order properties.
comparitive scales
-
•A respondent is presented with two objects and asked to select one according to some criterion.
•Ordinal data obtained
•The most widely used comparative scaling technique.With n objects, [n(n - 1) /2] paired comparisons are required
paired comparative scales
-
Respondents are presented with several objects simultaneously and asked to order or rank them according to some criterion.
Rank Order Scales
-
Respondents allocate a constant sum of units, such as 100 points, to attributes of a product to reflect their importance.
Constant Sum Scales
-
Investigation of all individual elements that make up a population
-costs more
-takes longer
-sometimes impossible
-sometimes sampling may be more accurate
Census
-
The selection of a fraction of the total amount of units of interest to decision making, for the ultimate purpose of being able to draw general conclusions about the entire body of units
Sampling
-
the total group of people from whom information is needed
Population
-
a subset of the population of interest
Sample
-
A characteristic or measure of a population
- If it were possible to take measures from all members of a population without error, a true value of a parameter could be determined
Parameter
-
A characteristic or measure of a sample
-Statistics are calculated from sample data and used to estimate population parameters
Statistic
-
The difference between results obtained from a sample and results that would have been obtained had information been gathered from or about every member of the population
Sampling Error
-
A sample in which each target population element has a known, nonzero chance of being included in the sample
Probability Sample
-
A sample that relies on personal judgment in the element selection process
Nonprobability Sample
-
-One can statistically assess level of sampling error and make inferences about the population (and not just the specific sample)
-Thus, results are generalizable from the sample to the population
Probability Sample
-
-Neither sampling error nor the margin of sampling error can be estimated or calculated
-Inferences cannot be made about the population
-Inferences are limited to the sample
-Thus, results are not generalizable from the sample to the population
Nonprobability Sample
-
-Population elements are sampled simply because they are in the right place at the right time
-Sometimes referred to as “accidental” sampling
-Easy to conduct, but no way to know if sample is representative of the population (i.e., cannot statistically assess sampling error)
-Examples include online or television “question of the day” polls
Convenience Sample
-
-Population elements are handpicked because they are expected to serve the research purpose
-The researcher may believe that they are representative of the larger population or that they can offer the information needed
-Best Use: early stages of research, seeking ideas or insights, realization by the researcher of its limitations
-Most Hazardous: used in descriptive or causal studies and its limitations are overlooked
-Examples include hiring panelists who are knowledgeable about the issue at hand rather than selecting them at random
Judgment Sample
-
-Initial respondents selected by probability methods
-Subsequent respondents are selected based on the referrals by initial respondents
snowball sampling
-
-Sample chosen so that the proportion of sample elements with certain characteristics is about the same as the proportion of the elements with the characteristics in the target population
-Stated more simply, certain important characteristics of the population are represented proportionately in the sample
Quota Sample
-
-Each unit included in the population has a known and equal chance of being included in the sample
-Every element is selected independently of every other element
-Typically drawn by a computer or from a physical list using a random number table
Simple Random Sample
-
Sample in which every kth element in the population is selected for the sample pool after a random start
Systematic Sample
-
-The number of population elements that must be drawn from the population and included in the initial sample pool in order to end up with the desired sample size
-Applies to any type of sample, not just systematic samples
Total Sampling Elements (TSE)
-
-Sample in which (1) the population is divided into mutually exclusive and exhaustive subsets and (2) a simple random sample of elements is chosen independently from each group/subset
-Most appropriate when subsets (or strata) are homogeneous within but heterogeneous between with respect to key variables
Stratified Sample
-
Number of observations in the total sample is allocated among the strata in proportion to the relative number of elements in each stratum in the population
Proportionate Stratified Sample
-
Involves balancing the two criteria of strata size and variability
Disproportionate Stratified Sample
-
-Like stratified sampling, (1) the population is divided into mutually exclusive and exhaustive subsets
-Unlike stratified sampling, (2) a simple random sample of subsets (i.e., clusters) is chosen
-Most appropriate when subsets (or strata) are heterogeneous within but homogeneous between with respect to key variables
Cluster Sample
-
¢Degree of accuracy
¢Resources
¢Time
¢Advanced knowledge of population
¢National versus local
¢Need for statistical analysis
Basis for Choosing a Sample Design
-
¢Degree
of acceptable error in an estimate of a population parameter
Precision
-
¢The range into which the true
population parameter will fall, assuming a given level of confidence
-
¢Degree
to which the researcher can feel assured that an estimate approximates the true
value
Confidence
-
Ê1. Specify the level of precision (H=2)
Ê2. specify the level of confidence
(CL=95%)
Ê3. determine the z value associated
with the confidence level
Ê90% confidence: z =1.65
Ê95% confidence: z =1.96
Ê99% confidence: z =2.58
Ê4. Determine the standard deviation of
the population (σ=29)
ÊUse standard deviation from a previous study on the target population.
ÊConduct a pilot study of a few members of the target
population and calculate .
ÊEstimate the range the value you are estimating can take on (minimum and
maximum value) and divide the range by 6.
Ê5. Determine the sample size using the
formula
Determining sample size when estimating means
-
Ê1. Specify the level of precision (H = 2%)
Ê2.Specify the level of confidence (CL=95%).
Ê3. Determine the Z value associated
with the confidence interval (z =
1.96)
Ê4. Estimate the population proportion (π = 25%)
ÊPilot studies
ÊPrevious studies
ÊJudgment
ÊThe most conservative position is to
predict that π = 0.50; this will lead to a larger
sample size
Ê5. Determine the sample size using the
formula.
- Determining
- Sample Size When Estimating Proportions
-
ÊBeyond the statistical approach to
determine sample size discussed to this point, sample size can also be
determined by
ÊThe available research budget
ÊAnticipated analyses and the number of
cases necessary to perform those analyses
ÊHistorical evidence of sample sizes in
similar studies
- Other
- Approaches to Determining Sample Size
-
ÊOther Approaches to Determining Sample SizeSampling error is decreased by increasing sample
size
ÊSampling Error
-
ÊNonsampling Error
- ÊError that arises in research that is
- not due to sampling
-
ÊNonsampling error that arises because of a failure
to include some units, or entire sections, of the defined target population in
the sampling frame
Noncoverage Error
-
ÊNonsampling error that represents a failure to
obtain information from some elements of the population that were selected and
designated for the sample
Nonresponse Error
-
ÊNonsampling error that arises because some
designated respondents refuse to participate in the study
ÊRefusals
-
ÊNonsampling error that arises because some
designated respondents are not at home hen the interviewer calls
ÊNot-at-Homes
-
ÊNonsampling errors that arise in the editing,
coding, or analysis phases of research
Office Error
-
ÊRespondent Interest in Topic
Ê“Foot-in-the-door” Technique
ÊInterviewer Characteristics and
Training
ÊGuarantee of Confidentiality or
Anonymity
ÊPrenotification
ÊPersonalization
ÊSponsor Disguise
ÊResponse Incentives
ÊSurvey Length
ÊFollow-Up Surveys
-
ranks a list of brand alternatives on numerous attributes
attribute belief ranking
|
|