-
8/13/2019 34094347 Types of Scales
1/23
Types of ScalesA scale refers to any series of items that are
arranged progressively according tovalue or magnitude,into which an
item can be placed according to its quantification. In other words,
ascale is a continuousspectrum or series of categories.It is
traditional to classify scales of measurement on the basis of the
mathematicalcomparisons that areallowable with these scales. Four
types of scales are nominal, ordinal, interval, andratio.Nominal
ScaleA nominal scale is the one in which the numbers or letters
assigned to objects serveas labels foridentification or
classification. This measurement scale is the simplest type.
Withnominal data, we arecollecting information on a variable that
naturally or by design can be grouped intotwo or morecategories
that are mutually e clusive, and collectively e haustive.!ominal
scales are the least powerful of the four scales. They suggest no
order ordistance relationshipand have no arithmetic origin.
!evertheless, if no other scale can be used, one canalmost always
oneset of properties into a set of equivalent classes.Ordinal
Scale"rdinal scales include the characteristics of the nominal
scale plus an indicator oforder. If a is greaterthan b and b is
greater than c, then a is greater than c. The use of ordinal
scaleimplies a statement of #greater than$ or #less than$ without
stating how much greater or less. "therdescriptors can be%#superior
to,$ #happier than,$ #poorer than,$ or #above.$Interval
ScaleInterval scales have the power of nominal and ordinal scales
plus one additionalstrength% theyincorporate the concept of
equality of interval &the distance between ' and ( equalsthe
distance between( and )*. For e ample, the elapsed time between )
and + A. . equals the timebetween - and A. ."ne cannot say,
however, + A. . is twice as late as ) A. . because #/ero time$ is
anarbitrary origin.In the consumer price inde , if the base year is
'01), the price level during '01)will be set arbitrarily as'22.
Although this is an equal interval measurement scale, the /ero
point isarbitrary.Ratio Scale3atio scales incorporate all thee
powers of the previous scales plus the provision forabsolute /ero
ororigin. 3atio data represent the actual amounts of variable.
easures of physicaldimensions such as
-
8/13/2019 34094347 Types of Scales
2/23
weight, height, distance, and area are the e amples. The
absolute /ero represents apoint on the scalewhere there is an
absence of the given attribute. If we hear that a person has
/eroamount of money, weunderstand the /ero value of the amount.
CRITERIA FOR GOOD MEASUREMENT!ow that we have seen how to
operationally define variables, it is important toma4e sure that
theinstrument that we develop to measure a particular concept is
indeed accuratelymeasuring the variable,and in fact, we are
actually measuring the concept that we set out to measure. This
ensures that inoperationally defining perceptual and attitudinal
variables, we have not overloo4edsome importantdimensions and
elements or included some irrelevant ones. The scales
developedcould often beimperfect and errors are prone to occur in
the measurement of attitudinal variables.
The use of betterinstruments will ensure more accuracy in
results, which in turn, will enhance thescientific quality of
theresearch. 5ence, in some way, we need to assess the #goodness$
of the measuredeveloped.What should be the characteristics of a
good measurement6 An intuitive answer to
this question is thatthe tool should be an accurate indicator of
what we are interested in measuring. Inaddition, it should beeasy
and efficient to use. There are three major criteria for evaluating
ameasurement tool% validity,reliability, and
sensitivity.alidity
'. 7ontent validity(. 7onstruct validity
-
8/13/2019 34094347 Types of Scales
3/23
). 8redictive validity-. 7oncurrent validity9. Face :alidity+.
7onvergent validity. ;iscriminant validity
:alidity is the ability of an instrument &for e ample
measuring an attitude* tomeasure what it issupposed to measure.
That is, when we as4 a set of questions &i.e. develop
ameasuring instrument* withthe hope that we are tapping the
concept, how can we be reasonably certain thatwe are
indeedmeasuring the concept we set out to do and not something
else6 There is no quic4answer.3esearchers have attempted to assess
validity in different ways, including as4ingquestions such as
#Isthere consensus among my colleagues that my attitude scale
measures what it issupposed to measure6$and #;oes my measure
correlate with others< measures of the =same< concept6$and
#;oes the behaviore pected from my measure predict the actual
observed behavior6$ 3esearcherse pect the answers toprovide some
evidence of a measure
-
8/13/2019 34094347 Types of Scales
4/23
"$ @hould men and women get equal pay for equal wor46%$@hould
men and women share household tas4s6
These two questions do not provide coverage to all the
dimensions delineatedearlier. It definitely fallsshort of adequate
content validity for measuring feminism.A panel of persons to judge
how well the instrument meets the standard can attestto the content
validityof the instrument. A panel independently assesses the test
items for a performancetest. It judges eachitem to be essential,
useful but not essential, or not necessary in assessingperformance
of a relevantbehavior.Face validity is considered as a basic and
very minimum inde of content validity.Face validityindicates that
the items that are intended to measure a concept, do on the face of
itloo4 li4e theymeasure the concept. For e ample a few people would
accept a measure of collegestudent math abilityusing a question
that as4ed students% ( ( B 6 This is not a valid measure
ofcollege>level math abilityon the face of it. !evertheless, it
is a subjective agreement among professionalsthat a scale
logicallyappears to reflect accurately what it is supposed to
measure. When it appearsevident to e perts that themeasure provides
adequate coverage of the concept, a measure has face validity.!%#
Criterion&Related alidity7riterion validity uses some standard
or criterion to indicate a construct accurately.
The validity of anindicator is verified by comparing it with
another measure of the same construct inwhich research
hasconfidence. There are two subtypes of this 4ind of
validity.Concurrent validity: To have concurrent validity, an
indicator must be associated with apree istingindicator that is
judged to be valid. For e ample we create a new test to
measureintelligence. For it tobe concurrently valid, it should be
highly associated with e isting IC tests &assumingthe same
definitionof intelligence is used*. It means that most people who
score high on the oldmeasure should also scorehigh on the new one,
and vice versa. The two measures may not be perfectlyassociated,
but if theymeasure the same or a similar construct, it is logical
for them to yield similarresults.
Predictive validity:7riterion validity whereby an indicator
predicts future events that are logicallyrelated to a construct
iscalled a predictive validity. It cannot be used for all measures.
The measure and theaction predictedmust be distinct from but
indicate the same construct. 8redictive measurementvalidity should
not be
-
8/13/2019 34094347 Types of Scales
5/23
confused with prediction in hypothesis testing, where one
variable predicts adifferent variable in future.?oo4 at the
scholastic assessment tests being given to candidates
see4ingadmission in differentsubjects. These are supposed to
measure the scholastic aptitude of the candidates Dthe ability
toperform in institution as well as in the subject. If this test
has high predictivevalidity, then candidateswho get high test score
will subsequently do well in their subjects. If students withhigh
scores performthe same as students with average or low score, then
the test has low predictivevalidity.!'# Constr(ct alidity7onstruct
validity is for measures with multiple indicators. It addresses
thequestion% If the measure isvalid, do the various indicators
operate in consistent manner6 It requires a definitionwith
clearlyspecified conceptual boundaries. In order to evaluate
construct validity, we considerboth theory andthe measuring
instrument being used. This is assessed through convergent
validityand discriminantvalidity.Conver)ent alidity* This 4ind of
validity applies when multiple indicators converge orare
associatedwith one another. 7onvergent validity means that multiple
measures of the sameconstruct hangtogether or operate in similar
ways. For e ample, we construct #education$ byas4ing people how
mucheducation they have completed, loo4ing at their institutional
records, and as4ingpeople to complete atest of school level
4nowledge. If the measures do not converge &i.e. people
whoclaim to have collegedegree but have no record of attending
college, or those with college degreeperform no better than
highschool dropouts on the test*, then our test has wea4 convergent
validity and weshould not combine allthree indicators into one
measure.Discriminant alidity* Also called divergent validity,
discriminant validity is theopposite of convergent validity. It
means that the indicators of one construct hang together
orconverge, but alsodiverge or are negatively associated with
opposing constructs. It says that if twoconstructs A and E arevery
different, then measures of A and E should not be associated. For e
ample, wehave '2 items thatmeasure political conservatism. 8eople
answer all '2 in similar ways. Eut we havealso put 9 questionsin
the same questionnaire that measure political liberalism. "ur
measure ofconservatism has
-
8/13/2019 34094347 Types of Scales
6/23
discriminant validity if the '2 conservatism items hang together
and are negativelyassociated with 9liberalism ones.Relia+ility
The reliability of a measure indicates the e tent to which it is
without bias &errorfree* and hence ensuresconsistent
measurement across time and across the various items in
theinstrument. In other words, thereliability of a measure is an
indication of the stability and consistency with which
theinstrumentmeasures the concept and helps to assess the
=goodness$ of measure.Sta+ility of Meas(res
The ability of the measure to remain the same over time D
despite uncontrollabletesting conditions orthe state of the
respondents themselves D is indicative of its stability and
lowvulnerability to changes inthe situation. This attests to its
#goodness$ because the concept is stablymeasured, no matter when it
isdone. Two tests of stability are test>retest reliability and
parallel>form reliability.!"# Test&retest Relia+ility*
Test>retest method of determining reliability
involvesadministering thesame scale to the same respondents at two
separate times to test for stability. Ifthe measure is stableover
time, the test, administered under the same conditions each time,
shouldobtain similar results. Fore ample, suppose a researcher
measures job satisfaction and finds that +- percentof the
population issatisfied with their jobs. If the study is repeated a
few wee4s later under similarconditions, and theresearcher again
finds that +- percent of the population is satisfied with their
jobs,it appears that themeasure has repeatability. The high
stability correlation or consistency between thetwo measures attime
' and at time ( indicates high degree of reliability. This was at
the aggregatelevel the samee ercise can be applied at the
individual level. When the measuring instrumentproduces
unpredictableresults from one testing to the ne t, the results are
said to be unreliable because oferror in measurement.
There are two problems with measures of test>retest
reliability that are common toall longitudinalstudies. Firstly, the
first measure may sensiti/e the respondents to theirparticipation
in a researchproject and subsequently influence the results of the
second measure. Further if thetime between themeasures is long,
there may be attitude change or other maturation of the
subjects.
Thus it is possiblefor a reliable measure to indicate low or
moderate correlation between the first andthe second
-
8/13/2019 34094347 Types of Scales
7/23
administration, but this low correlation may be due an attitude
change over timerather than to lac4 of reliability.!%#
,arallel&Form Relia+ility* When responses on two comparable
sets of measurestapping the sameconstruct are highly correlated, we
have parallel>form reliability. It is also
calledequivalent>formreliability. Eoth forms have similar items
and same response format, the onlychanges being thewording and the
order or sequence of the questions. What we try to establish hereis
the error variabilityresulting from wording and ordering of the
questions. If two such comparable formsare highlycorrelated, we may
be fairly certain that the measures are reasonably reliable,
withminimal errorvariance caused by wording, ordering, or other
factors.Internal Consistency of Meas(resInternal consistency of
measures is indicative of the homogeneity of the items inthe
measure that tapthe construct. In other words, the items should
=hang together as a set,< and becapable of
independentlymeasuring the same concept so that the respondents
attach the same overallmeaning to each of theitems. This can be
seen by e amining if the items and the subsets of items in
themeasuring instrumentare highly correlated. 7onsistency can be e
amined through the inter>itemconsistency reliability
andsplit>half reliability.!"# Inter&item Consistency
relia+ility* This is a test of consistency of respondentshalf
method theresearcher may ta4e theresults obtained from one half of
the scale items !e.g. odd>numbered items* andchec4 them against
theresults from the other half of the items &e.g. even numbered
items*. The highcorrelation tells us there issimilarity &or
homogeneity* among its items.It is important to note that
reliability is a necessary but not sufficient condition ofthe test
of goodness of a measure. For e ample, one could reliably measure a
concept establishing highstability and
-
8/13/2019 34094347 Types of Scales
8/23
consistency, but it may not be the concept that one had set out
to measure. :alidityensures the abilityof a scale to measure the
intended concept.Sensitivity
The sensitivity of a scale is an important measurement concept,
particularly whenchanges in attitudes orother hypothetical
constructs are under investigation. @ensitivity refers to
aninstrument
-
8/13/2019 34094347 Types of Scales
9/23
,RO.A.I/IT0 AND NON&,RO.A.I/IT0 SAM,/ING There are several
alternative ways of ta4ing a sample. The major alternativesampling
plans may begrouped into probability techniques and
non>probability techniques. In pro+a+ilitysamplin) everyelement
in the population has a known nonzero probability of selection. The
simplerandom is the best4nown probability sample, in which each
member of the population has an equalprobability of beingselected.
8robability sampling designs are used when the representativeness
of the
sample is of importance in the interest of wider
generalisability. When time or other factors,rather
thangeneralisability, become critical, non>probability sampling
is generally used.In non>probability sampling the probability of
any particular element of thepopulation being chosen isun4nown. The
selection of units in non>probability sampling is quite
arbitrary, asresearchers relyheavily on personal judgment. It
should be noted that there are no appropriatestatistical techniques
formeasuring random sampling error from a non>probability
sample. Thus projectingthe data beyond the
sample is statistically inappropriate. !evertheless, there are
occasions when non>probability samplesare best suited for the
researcherprobability sampling designs, the elements in the
population do not have anyprobabilitiesattached to their being
chosen as sample subjects. This means that the findingsfrom the
study of thesample cannot be confidently generali/ed to the
population. 5owever theresearchers may at times beless concerned
about generalisability than obtaining some preliminary
informationin a quic4 and
ine pensive way. @ometimes non>probability could be thee only
way to collect thedata.Convenience Samplin)7onvenience sampling
&also called haphazard or accidental sampling * refers to
samplingby obtainingunits or people who are most conveniently
available. For e ample, it may beconvenient andeconomical to sample
employees in companies in a nearby area, sample from apool of
friends and
-
8/13/2019 34094347 Types of Scales
10/23
neighbors. The person>on>the street interview conducted by
T: programs is anothere ample. T:interviewers go on the street with
camera and microphone to tal4 to few peoplewho are convenient
tointerview. The people wal4ing past a T: studio in thee middle of
the day do notrepresent everyone&homema4ers, people in the
rural areas*. ?i4ewise, T: interviewers select peoplewho loo4
#normal$ tothem and avoid people who are unattractive, poor, very
old, or inarticulate.Another e ample of hapha/ard sample is that of
a newspaper that as4s the readersto clip a questionnairefrom the
paper and mail it in. !ot everyone reads thee newspaper, has an
interestin the topic, or willta4e the time to cut out the
questionnaire, and mail it. @ome will , and the numberwho do so may
seemlarge, but the sample cannot be used to generali/e accurately
to the population.7onvenience samples are least reliable but
normally the cheapest and easiest toconduct.7onvenience sampling is
most often used during the e ploratory phase of aresearch project
and isperhaps the best way of getting some basic information
quic4ly and efficiently."ften such sample ista4en to test ideas or
even to gain ideas about a subject of interest.,(rposive
Samplin);epending upon the type of topic, the researcher lays down
the criteria for thesubjects to be included inthe sample. Whoever
meets that criteria could be selected in the sample. Theresearcher
might selectsuch cases or might provide the criteria to somebody
else and leave it to hisGher
judgment for the actualselection of the subjects. That is why
such a sample is also called as 1(d)mental ore2pert opinionsample$
For e ample a researcher is interested in studying students who are
enrolledin a course onresearch methods, are highly regular, are
frequent participants in the classdiscussions, and often comewith
new ideas. The criteria has been laid down, the researcher may do
this jobhimselfGherself, or mayas4 the teacher of this class to
select the students by using the said criteria. In thelatter
situation we areleaving it to the judgment of the teacher to select
the subjects. @imilarly we cangive some criteria tothe fieldwor4ers
and leave it to their judgment to select the subjects accordingly.
Ina study of wor4ingwomen the researcher may lay down the criteria
li4e% the lady is married, has twochildren, one of herchild is
school going age, and is living in nuclear family.3(ota Samplin)A
sampling procedure that ensures that certain characteristics of a
populationsample will be represented
-
8/13/2019 34094347 Types of Scales
11/23
to the e act e tent that the researcher desires. In this case
the researcher firstidentifies relevantcategories of people
&e.g. male and female or under age )2, ages )2 to +2, over+2,
etc* then decideshow many to get in each category. Thus the number
of people in various categoriesof sample is fi ed.For e ample the
researcher decides to select 9 males and 9 females under age )2,'2
males and '2females aged )2 to +2, and 9 males and 9 females over
age +2 for a -2 personsample. This is quotasampling."nce the quota
has been fi ed then the researcher may use convenience
sampling.
The conveniencesampling may introduce +ias . For e ample, the
field wor4er might select theindividual according tohisGher li4ing,
who can easily be contacted, willing to be interviewed, and belong
tomiddle class.Cuota sampling can be considered as a form of
proportionate stratified sampling, inwhich apredetermined
proportion of people are sampled from different groups, but on
aconvenience basis.@peed of data collection, lower costs, and
convenience are the major advantages ofquota samplingcompared to
probability sampling. Cuota sampling becomes necessary when asubset
of a population isunderrepresented, and may not get any
representation if equal opportunity isprovided to each.Although
there are many problems with quota sampling, careful supervision of
thedata collection mayprovide a representative sample of the
various subgroups within the population.Sno4+all Samplin)@nowball
sampling &also called network, chain referral, or reputational
sampling * is a methodforidentifying and sampling &or
selecting* cases in the networ4. It is based on ananalogy to a
snowball,which begins small but becomes larger as it is rolled on
wet snow and pic4s upadditional snow. Itbegins with one or a few
people or cases and spreads out on the basis of lin4s tothee
initial cases.
This design has been found quite useful where respondents are
difficult to identifyand are best locatedthrough referral networ4s.
In the initial stage of snowball sampling, individuals
arediscovered and mayor may not be selected through probability
methods. This group is then used tolocate others whopossess similar
characteristics and who, in turn, identify others. The
#snowball$gather subjects as itrolls along.For e ample, a
researcher e amines friendship networ4s among teenagers in
acommunity. 5e or she
-
8/13/2019 34094347 Types of Scales
12/23
begins with three teenagers who do not 4now each other. Hach
teen names fourclose friends. Theresearcher then goes to the four
friends and as4s each to name four close friends,then goes to those
fourand does the same thing again, and so forth. Eefore long, a
large number of peopleare involved. Hachperson in the sample is
directly or indirectly tied to the original teenagers, andseveral
people may havenamed the same person. The researcher eventually
stops, either because no newnames are given,indicating a closed
networ4, or because the networ4 is so large that it is at theelimit
of what he or shecan study.Se5(ential Samplin)@equential sampling
is similar to purposive sampling with one difference. Inpurposive
sampling, theresearcher tries to find as many relevant cases as
possible, until time, financialresources, or his or herenergy is e
hausted. The principle is to get every possible case. In
sequentialsampling, a researchercontinues to gather cases until the
amount of new information or diversity is filled.
The principle is togather cases until a saturation point is
reached. In economic terms, information isgathered, or
theincremental benefit for additional cases, levels off or drops
significantly. It requiresthat the researchercontinuously evaluates
all the collected cases. For e ample, a researcher locatesand plans
in>depthinterviews with +2 widows over 2 years old who have been
living without a spousefor '2 or moreyears. ;epending on the
researcher
-
8/13/2019 34094347 Types of Scales
13/23
8robability samples that rely on random processes require more
wor4 thannonrandom ones. Aresearcher must identify specific
sampling elements &e.g. persons* to include in thesample. Fore
ample, if conducting a telephone survey, the researcher needs to
try to reach thespecific sampledperson, by calling bac4 several
times, to get an accurate sample.3andom samples are most li4ely to
yield a sample that truly represents thepopulation. In
addition,random sampling lets a researcher statistically calculate
the relationship betweenthe sample and thepopulation D that is the
si/e of sampling error. A non>statistical definition of
thesampling error is thedeviation between sample result and a
population parameter due to randomprocess.Simple Random Sample
The simple random sample is both the easiest random sample to
understand andthe one on which othertypes are modeled. In simple
random sampling, a research develops an accuratesampling frame,
selectselements from sampling frame according to mathematically
random procedure,then locates the e actelement that was selected
for inclusion in the sample.After numbering all elements in a
sampling frame, the researcher uses a list ofrandom numbers
todecide which elements to select. 5e or she needs as many random
numbers asthere are elements to besampled% for e ample, for a
sample of '22, '22 random numbers are needed. Theresearcher can
getrandom numbers from a random number table, a table of numbers
chosen in amathematically randomway. 3andom>number tables are
available in most statistics and research methodsboo4s. The
numbersare generated by a pure random process so that any number
has an equalprobability of appearing in anyposition. 7omputer
programs can also produce lists of random number.A random starting
point should be selected at the outset.3andom sampling does not
guarantee that every random sample perfectlyrepresents the
population.Instead, it means that most random samples will be close
to the population most ofthe time, and thatone can calculate the
probability of a particular sample being inaccurate. Aresearcher
estimates thechance that a particular sample is off or
unrepresentative by using information fromthe sample toestimate the
sampling distribution. The sampling distribution is the 4ey idea
thatlets a researchercalculate sampling error and confidence
interval.Systematic Random Sample
-
8/13/2019 34094347 Types of Scales
14/23
-
8/13/2019 34094347 Types of Scales
15/23
-
8/13/2019 34094347 Types of Scales
16/23
In a disproportionate, sample si/e for each stratum is not
allocated in proportion tothe population si/e,but is dictated by
analytical considerations.Cl(ster Samplin)
The purpose of cluster sampling is to sample economically while
retaining thecharacteristics of aprobability sample. Kroups or
chun4s of elements that, ideally, would haveheterogeneity among
themembers within each group are chosen for study in cluster
sampling. This is incontrast to choosingsome elements from the
population as in simple random sampling, or stratifyingand then
choosingmembers from the strata, or choosing every nth case in the
population insystematic sampling. Whenseveral groups with
intra>group heterogeneity and inter>group homogeneity
arefound, then a randomsampling of the clusters or groups can
ideally be done and information gatheredfrom each of themembers in
the randomly chosen clusters.7luster samples offer more
heterogeneity within groups and more homogeneityamong
andhomogeneity within each group and heterogeneity across
groups.7luster sampling addresses two problems% researchers lac4 a
good sampling framefor a dispersedpopulation and the cost to reach
a sampled element is very high. A cluster is unitthat contains
finalsampling elements but can be treated temporarily as a sampling
element itself.3esearcher first samplesclusters, each of which
contains elements, then draws a second a second samplefrom within
the clustersselected in the first stage of sampling. In other
words, the researcher randomlysamples clusters, andthen randomly
samples elements from within the selected clusters. 5e or she
cancreate a goodsampling frame of clusters, even if it is
impossible to create one for samplingelements. "nce theresearcher
gets a sample of clusters, creating a sampling frame for elements
withineach cluster becomesmore manageable. A second advantage for
geographically dispersed populations isthat elements withineach
cluster are physically closer to each other. This may produce a
savings inlocating or reaching eachelement.A researcher draws
several samples in stages in cluster sampling. In a
three>stagesample, stage ' israndom sampling of big clusters
stage ( is random sampling of small clusterswithin each selected
bigcluster and the last stage is sampling of elements from within
the sampled withinthe sampled small
-
8/13/2019 34094347 Types of Scales
17/23
clusters. First, one randomly samples the city bloc4s, then
households within bloc4s,then individualswithin households. This
can also be an e ample of m(ltista)e area samplin)$
The unit costs of cluster sampling are much lower than those of
other probabilitysampling designs.5owever, cluster sampling e poses
itself to greater biases at each stage ofsampling.Do(+le
Samplin)
This plan is adopted when further information is needed from a
subset of the groupfrom which someinformation has already been
collected for the same study. A sampling designwhere initially a
sampleis used in a study to collect some preliminary information of
interest, and later asub>sample of thisprimary sample is used to
e amine the matter in more detail, is called doublesampling.
alidity in E2perimentsH periments are judged by two measures.
The first, internal validity indicateswhether the
independentvariable was the sole cause of the change in the
dependent variable. It implies theresearcher
-
8/13/2019 34094347 Types of Scales
18/23
that the treatment was the true causal factor producing change
in the dependentvariable. The secondmeasure, e ternal validity,
indicates the e tent to which the results of thee periment are
applicable inthe real world.Internal validity is high in the
laboratory e periment, reason being the control overall the
confoundingfactors. H ternal validity &generalisability* is not
sure because of the effect ofvariety of factors. Fielde periments
have more e ternal validity but less internal validity because it
iscloser to the realsituations.Factors Affectin) Internal alidityIn
choosing or evaluating e perimental research design, researchers
mustdetermine whether they haveinternal and e ternal validity.
There are eight major types of e traneous variablesthat may
jeopardi/einternal validity% 5istory effect, maturation effect,
testing effect, instrumentationeffect, selection biaseffect,
selection bias effect, statistical regression, mortality, and
mechanical loss."$ -istory Effect* A specific event in the e ternal
environment occurring between thefirst and secondmeasurement that
is beyond the control of the e perimenter and that affects
thevalidity of ane periment. Advertisement of a particular product
&mineral water* and its sale isaffected by an event inthe
society &contamination of drin4ing water*. The researcher does
not have controlon such happeningswhich have an impact on the L and
relationship.%$ Mat(ration Effect* C ause and effect relationship
can also be contaminated by theeffects of thepassage of time D
another uncontrollable variable. @uch contamination is
calledmaturation effect. Thematuration effects are a function of
the processes D biological and psychological Doperating within
thesubjects as a result of the passage of time. H amples of
maturation processes couldinclude growing older, getting tired,
feeling hungry, and getting bored . In other words there could
bematuration effecton the dependent variable purely because of the
passage of time. For e ample, letus say that an 3 M ;director
intends that an increase in the efficiency of wor4ers would result
withinthree months< time if advanced technology is introduced in
the wor4 setting. If at the end of three monthsincreasedefficiency
is indeed found, I will be difficult to claim that the advanced
technology&and it alone*increased the efficiency of wor4ers,
because with the passage of time, employeeswould also gained
-
8/13/2019 34094347 Types of Scales
19/23
e perience, resulting in better performance and therefore
improved efficiency. Thus, the internalvalidity also gets reduced
owing to the effects of maturation in as much as it isdifficult to
pinpoint howmuch of the increase is attributable to the
introduction of the enhanced technologyalone.'$ Testin) Effects*
Frequently, to test the effects of treatment, subjects are given
whatis called a
pretest &say7a short questionnaire eliciting their feelings
and attitudes*. That is, ameasure of thedependent variable is ta4en
&pretest*, then the treatment given, and after that asecond
test, called
posttest , administered. The difference between the posttest and
the pretest scores isthen attributed tothe treatment. 5owever, the
very fact that the subjects were e posed to the pretestmight
influence theirresponses on the posttest, which will adversely
impact on internal validity. It is alsocalled sensiti/ationthrough
previous testing.8$ Instr(mentation Effects* Instrumentation
effects are yet another source of threat tointernalvalidity. These
might arise because of a change in the measuring instrumentbetween
pretest andposttest, and not because of the instrument
-
8/13/2019 34094347 Types of Scales
20/23
e treme values or a tendency for random error to move group
results towards theaverage. If e tremesare ta4en then they tend to
regress toward the mean. Those who are on either endof the e treme
wouldnot truly reflect the cause and effect relationship."ne
situation arises when subjects are unusual with regard to dependent
variable.Eecause they beginas unusual or e treme, subjects are
li4ely to respond further in the same direction.For e ample,
aresearcher wants to see whether violent films ma4e people act
violently. Theresearcher chooses agroup of violent criminals from a
high security prison, gives them a pretest, showsviolent films,
andthen administers a posttest. To the researcherprisoners who did
see the film are slightly moreviolent thanbefore. Eecause the
violent criminals began at an e treme, it is unli4ely that
atreatment could ma4ethem more violent by random chance alone, they
appear less e treme whenmeasured a second time.If participants
chosen for e perimental group have e treme scores on thedependent
variable to beginwith then the laws of probability say that those
with very low scores on a variablehave a greaterprobability to
improve and scoring closer to mean on the posttest after
treatment.
This phenomenon of low scorers tending to score closer to the
mean is 4nown as #regressing toward themean.$?i4ewise, those with
high scores have a greater tendency to regress toward themean D
will score loweron the posttest than on pretest. Thus the e tremes
will not #truly$ reflect the causalrelationship D athreat to
internal validity.;$ Mortality* ortality, or attrition, arises when
some subjects do not continuethroughout thee periment. Although the
word mortality means death, it does not necessarily meanthat
subjects havedied. If a subset of subjects leaves partway through
an e periment, a researchercannot whether theresults would have
been different had the subjects stayed. Hven with departure offew
subjects, thegroups do not remain balanced.7onsider for e ample of
a training e periment that investigates the effects of
closesupervision of salespersons &high pressure* versus low
supervision &low supervision*. The highpressure condition
maymisleadingly appear to be superior if those subjects who
completed the e perimentdid very well. If,
-
8/13/2019 34094347 Types of Scales
21/23
however, the high>pressure condition caused more subjects to
drop>out than theother condition, thisapparent superiority may
be due to a self>selection bias &those who could not bearthe
pressure had left Dmortality* D perhaps only very determined andGor
talented salespersons made itthrough the end of thee periment. the
researcher puts the new drug in the yellow capsule, puts anold drug
in the pin4one, and ta4e the green capsule a placebo D a false
treatment that appears to be real&e.g., a sugar capsulewithout
any physical effects*. Assistants who give the capsules and record
theeffects do not 4nowwhich color contains the new drug. "nly
another person who does not deal withsubjects directly 4nowswhich
colored capsule contains the drug and e amines the results.E2ternal
alidity
-
8/13/2019 34094347 Types of Scales
22/23
Hven if the researcher eliminates all concerns for internal
validity, e ternal validityremains a potentialproblem. H ternal
validity is the ability to generali/e e perimental findings to
reallife situations.Without e ternal validity, the findings are of
little use for both basic and appliedresearch i.e. we shallnot be
able to develop any theories that could be applicable to similar
othersituations.Reactivity* A T6reat to E2ternal alidity@ubjects
may react differently in an e periment than they would in real
lifebecause they 4now they arein a study. The Hawthorn Effect, a
specific 4ind of reactivity to the e perimentalsituation is a goode
ample in this respect. The e periment was conducted in the 5awthorn
Hlectric7ompany where theperformance of the participants was
supposed to change due to the change in theenvironmentalconditions
i.e. improvement on the environmental conditions will have a
positiveeffect on theeperformance. The researchers modified many
aspects of the wor4ing conditions andmeasuredproductivity.
8roductivity rose after each modification. 8roductivity rose even
ifthere was no realmodification but it was announced that there is
a modification. The behavior changewas simply areaction to the
announcement of modification and some other factors li4e
theparticipants were beingwatched and had a feeling of being =very
important persons.
