Chapter 9 Measurement And Scaling : Non-Comparative Scaling Techniques

Front 1

NoncomparativeScaling Technique

Back 1

employ whatever rating standard seemsappropriate to them. They do not compare the object being rated eitherto another object or to some specific standard. àEvaluate only one object at a time (for thisreason also called monadic scales)

Front 2

Noncomparative Scaling Technique consist of ..

Back 2

1. Continous rating scale --> graphic rating scale

2.Itemized rating scales --> Likert scale , Semantic differntial , Stapel scale

Front 3

WHAT IS a continous rating scale:

Back 3

1) respondents rate the objects by placing a markat the appropriate position in a line that runs from one extreme criterionvariable to the other. à use inmarketing research limited, but with theincrease popularity of computer-assisted personal interviewing and othertechniques use becoming more frequent"

Front 4

advantages and disadvantages of a continous rating scale

Back 4

Advantages: Easy to construct Easily implemented on the internet, automatically scored by computer.

Disadvantage: Scoring is cumbersome and unreliable Provided little new info

Front 5

Itemized rating scale , what is it?

Back 5

1) the respondents are provided with a scale thathas a number or brief description associated with each category. - Categories are ordered in terms of scaleposition; respondents are required to select the specified category that bestdescribes the object being rated. -

Widely used in marketing research & formthe basic components of more complex scales (such as multi-item scales)

Front 6

What is a Likert scale

Back 6

a) (summated scale) a widely used rating scale thatrequires the respondents to indicate a degree of agreement or disagreement witheach of a series of statements about stimulus objects·

Each scale itemhas 5 response categories ranging from ‘stronglydisagree’ to ‘strongly agree’.· Each statementis assigned a numerical score, ranging either from -2 to 2 or 1 to 5.

Front 7

Example of a Likert-scale

Back 7

Sears sellshigh-quality merchandise 1 2 3 4 5 correspod to Strongly disagree disagree Neither agree nor disagree Agree Strongly agree

Front 8

Advantages and disadvantages of Likert scales

Back 8

Advantages:· Easy to construct & administer· Respondents readily understand (suitable for:mail, telephone or personal interviews)

Disadvantage:· Takes longer to complete than any other itemizedrating scales (because respondents have to reach each statement)

Front 9

What is a semantic differntial scale

Back 9

is a 7-oint rating scale with endpoints withbipolar labels that have semantic meaning·

Individualitems on a semantic differential scale may be scored a (-3 to 3) or a (1 to 7)scale.

· Resulting data are commonly analyzed through profileanalysis

Front 10

What is profile analysis when it comes to semantic differntial scale?

Back 10

· Profile analysis: means or median valueson each rating scale are calculated and compared by plotting or statisticalanalysis). ·

Its versatility makes the semanticdifferential a popular rating scale in marketing (widely used in comparingbrand, product and company images)

Front 11

Example of one

Back 11

Example:Sears is?Powerful -:-:-:-:-:-:-:Weak mark the X, how accurately one or the other adjective describes whatthe store means to you

Front 12

What is a staple scale?

Back 12

a) : is a unipolar rating scale with 10categories numbered from -5 to +5

without a neutral point (zero).·

The scale isusually presented vertically. · Respondents areasked to indicate how accurately or inaccurately each item describes the objectby selecting an appropriate numerical response·

Data obtainedusing a stapel scale can be analyzed in the same way as semantic differentialdata!

Front 13

Advantages and disadvantages?

Back 13

Stapel Scale Advantages·

Not require apretest of the adjectivesDisadvantage·

Confusing· Difficult to apply

Front 14

What decision could you make regarding regarding itemized rating scales?

Back 14

No. of scale categories (between 5 and 9) Balanced vs. unbalanced scale (balanced)

Odd or even no. of categories (odd)

Forced vs. non-forced (non-forced)

Nature and degree of verbal description

Physical form of scale (number of options à select the best one)

Front 15

What do they mean with number of scale categories

Back 15

-Two conflicting considerationconcerning number of scale categories.

a) The greater the number of scalecategories, the finer the discrimination among stimulus objects is possible.

b) Most respondents cannot handlemore than a few categories.

Traditional guidelines: appropriatenumber of categories should be between 5 and 9. Number of categories dependson: - Intersted respondents-Nature of the objects-Mode of data collection - how data is analzyed

Front 16

Multi-iteem scale

Back 16

-multi-item scale should be evaluated foraccuracy and applicability àassessment of reliability, validity, generalizabilityagment-

Front 17

Measurement and Measurement error

Back 17

Measurement is a number that reflects somecharacteristic of an object. it is not the true value of the characteristic ofinterest but rather an observation of it.A measurement error results in the measurement observedscore being different from the true score characteristic being measured.i-item ׻-d

Front 18

True score model

Back 18

True Score model: A mathematical model that providesa framework for understanding the accuracy of measurement

Xo=Xt+Xs+Xr

Xo= the observed score or measurementXt+ the true score of the characteristicXs=systematic errorXr=random error

Front 19

Random vs systematic error

Back 19

Systematic error: affects measurements in a constantway. It represents stable factors (such as mechanical factors)

Random error: is not constant. It represents transient factors that affect the observed score in different wayseach time the measurement is made (transient personal or situational factors):

Front 20

Scale evaluation dimensions

Back 20

Front 21

What is reliability?

Back 21

refers tothe extent to which a scale produces consistent results if repeatedmeasurements are made (systematic errors do not have an impact, because theyaffect in a constant way, random error produces inconsistencies!!!) à IF Xr =0,perfect reliable measure !!!!!

Front 22

What are the approaches to asses reliability?

Back 22

Test- retest = respondents are administered identical sets of scale items at twodifferent times under as nearly equivalent conditions as possible

Alternative forms = two equivalentforms of the scale are constructed and then the same respondents aremeasured at two different times

Internal-Consistency = an approach for assessing theinternal consistency of the set of items when several items are summated inorder to form a total score for the scale

spilt-halfreliability = items are divided into two halves and the resulting halfscores are correlated

Coefficient alpha (Cronbach’salpha) = it is the average of all possible split-half coefficientsresulting from different splitting of the scale items

Front 23

What is validity?

Back 23

the extentto which differences in observed scale scores reflect true differences among objects on the characteristic beingmeasured, rather than systematic or random errors

Front 24

What is content validity?

Back 24

systematicevaluation of the content of a scale for the meaning task at hand

Front 25

What is criterion validity?

Back 25

examineswhether the measurement scale performs as expected in relation to other variablesselected as meaningful criteria

Front 26

2aWhat is concurrent criterion validity

Back 26

data and criterion variables are collected atthe same time

Front 27

2bWhat is predicitive criterion validity?

Back 27

data on a scale is collected one time, data oncriterion variable is collected at a future time

Front 28

What is construct validity:

Back 28

addresses the question of what construct orcharacteristic the scale is measuring. An attempt is made to answer theoreticalquestions of why a scale works and what deductions can be made concerning thetheory underlying the scaleeperformרd

Front 29

Consutrct validity can be convergent, discriminant, nomological

Back 29

Convergent – extent to which the scale correlates positively with othermeasures of the same construct

Discriminant – extent to which the scale does not correlate with otherconstructs

Nomological – extent to which the scalecorrelates in theoretically predicted ways with measures of different butrelated constructs

Front 30

Relationshipbetween reliability and validity

Back 30

à perfectvalidity implies perfect reliability, but not vice versa!!!-reliability is a necessary, but NOT SUFFICIENT, condition for validity!