ERIC EJ789965: Determining Individual Preschoolers' Preferences in a Group Arrangement

2008·0.26 MB·English

by  ERIC

JOURNALOFAPPLIEDBEHAVIORANALYSIS 2008, 41, 25–37 NUMBER1 (SPRING2008)
DETERMINING INDIVIDUAL PRESCHOOLERS’ PREFERENCES IN A
GROUP ARRANGEMENT
STACY A. LAYER, GREGORY P. HANLEY, NICOLE A. HEAL, AND JEFFREY H. TIGER
UNIVERSITYOFKANSAS
This study sought to determine the accuracy of an assessment format in which selection
outcomes were delayed and probabilistic; these are unavoidable features of an assessment
designed to determine preferences of multiple children simultaneously. During the single
arrangement, preference hierarchies were established by having a child repeatedly select from
amongseveral foodsandbysequentiallyrestricting preferred items fromthearray. Afterbeing
taughttheassociationsbetweencoloredcardsandthesamefooditems,groupassessmentswere
conducted with 3 children simultaneously, in which each child chose a card and all children
receivedthefoodcorrelatedwitharandomlyselectedcardfromthosethathadjustbeenselected.
The group assessment appeared to be accurate and perhaps more efficient for determining
preferences;thus,wepositthatthisarrangementissuitablefordeterminingcontextpreferences
ofmultiple children simultaneously.
DESCRIPTORS: concurrent-chainsarrangements,delay,preferenceassessments,preschool
children,probabilistic outcomes
_______________________________________________________________________________
Preferenceassessmentstypicallyidentifypoten- ments (e.g., Lohrmann-O’Rourke, Browder, &
tially reinforcing items, activities, or contexts Brown,2000;Risley,1996)byallowingindivid-
throughdirectobservationofaselectionbetween uals who are typically unable to advocate for
two or more alternatives. These assessments are themselves to express their preferences for how
usually conducted with individuals for whom theirenvironmentistobearranged.
verbal report of preferences may be unreliable or Hanley and colleagues have used various
impossible to obtain (e.g., young children and permutationsofaconcurrent-chainsarrangement
persons with severe developmental disabilities). (Catania & Sagvolden, 1980) to determine
Many studies have described procedures for individuals’ preferences among contexts such as
assessing preferences among discrete items (e.g., protractedactivities(Hanley,Iwata,&Lindberg,
foods, toys) to be used as reinforcers during 1999), function-based interventions (Hanley,
interventionorinstruction(e.g.,DeLeon&Iwata, Piazza, Fisher, Contrucci, & Maglieri, 1997;
1996;Fisheretal.,1992;Pace,Ivancic,Edwards, Hanley, Piazza, Fisher, & Maglieri, 2005), and
Iwata,&Page,1985;Roane,Vollmer,Ringdahl, effective instructional practices for preschoolers
&Marcus,1998).However,relativelyfewstudies (Heal & Hanley, 2007; Tiger, Hanley, &
have described methods for determining prefer- Hernandez,2006).Theseassessmentsallinvolved
ences for important contexts or experiences, the modified concurrent-chains arrangements in
results of which could produce lifestyle enhance- which children chose among discriminative
stimulithatwerecorrelatedwithvariouscontexts
Thisresearchisbasedonathesisbythefirstauthorin
in the initial link of a chain; children then
partial fulfillment of the requirements for the master’s
degree. Gregory Hanley is now at Western New England experiencedthecorrelatedcontextintheterminal
College, Nicole Heal is now at the May Center for linkofthechain.Thisprocesswasthenrepeated
Education and Neurorehabilitation, and Jeffrey Tiger is
for each child until a preference for a particular
nowatLouisiana StateUniversity.
RequestsforreprintsshouldbeaddressedtoGregoryP. contextwasdeterminedthroughvisualinspection
Hanley, Psychology Department, Western New England of selection patterns in the initial links.
College, 1215 Wilbraham Road, Springfield, Massachu-
Heal and Hanley (2007) introduced two
setts01119(e-mail: [email protected]).
doi:10.1901/jaba.2008.41-25 procedural changes. First, the initially preferred
25
26 STACY A. LAYER et al.
context was removed and the assessment problembehaviorordifferentfree-playarrange-
continued until only one context remained to ments, it would be relatively easy to provide
establishapreferencehierarchyamongcontexts. multiple children an opportunity to express a
Second, assessments also were replicated by personal preference in the initial links of the
reintroducing all contexts and reapplying the assessment. Children could, for example, all
same restriction criteria. These changes provid- touch one of multiple shapes when entering the
ed better information regarding children’s classroom from outdoor play, with each shape
preferences for contexts, but they also extended corresponding to a different context. Because it
the time required to conduct the assessment. isnotlikelythatallchildrenwillselectthesame
The utility of determining children’s prefer- contextoneachday,themoredifficultaspectof
ences for behavior management strategies, the procedures would be allowing multiple
group instructional techniques, or designs of children to experience their selected context.
classroom activities (e.g., free play, circle, or For instance, it would not be feasible for
mealtime) is predicated on determining the classroom teachers to provide each child with
individual preferences of all members of a a different behavior management strategy each
classroom. Result from such assessments could day, and it would be impossible to allow
inform individualized or alternating preferred multiple free-play arrangements to cooccur
classroom practices; conditions consistent with simultaneously due to the shared nature of
the preferences of the majority also could be these contexts.
based on assessment results. However, applica- Although Hanley et al. (1997) showed that
tion of the context preference assessments, as differential reinforcement of selections (i.e.,
theyhavebeendescribedmostrecently(Heal& providing access to the contexts following
Hanley, 2007; Heal, Hanley, & Layer, unpub- selections) was critical for obtaining accurate
lished manuscript; Tiger, Hanley, & Heal, descriptions of preferences, they did not dem-
2006), would require a prohibitive amount of onstrate that this reinforcement needs to be
time for a scientific investigation and certainly continuous.Itispossiblethatintermittentaccess
would preclude routine application as a practi- to selected contexts may yield accurate descrip-
cal means to determine children’s preferences tionsofpreference,andthisintermittencywould
onaclasswidebasis.Forexample,120to1603- certainly make these assessments more efficient.
min trials were required to determine a context Forinstance,assessmentsthatscheduledaccessto
preference for each child in Heal and Hanley’s selected contexts on every other selection would
study. If these procedures were extended to require half the time to conduct. If intermittent
determine the context preferences of a class- access to selected contexts yielded accurate
room of 20 children and a similar amount of assessmentsofpreference,thenitispossiblethat
selection variability is assumed, more than multiple children’s preferences for various con-
140 hr would be required to complete the texts could be determined simultaneously. As a
classwide assessment. meanstothisend,atypeoflotterysystemcould
Heal and Hanley (2007) noted that experi- be adopted in which all children are provided
encing the contexts in the terminal links is the with an opportunity to select their preferred
time-consumingaspectoftheconcurrent-chains context,butonlyonecontext,randomlyselected
procedures; by contrast, observing selections in from the chosen contexts, is experienced by the
the initial links—the actual measure of prefer- group. All children would have a chance of
ence—requires little time. For instance, if one experiencing their selected context, but they
wanted to determine preschool children’s would experience their selected context only
preferences for different teacher responses to intermittently.
PREFERENCE ASSESSMENTS IN A GROUP ARRANGEMENT 27
With this group-oriented concurrent-chains and typically developing. Participation was a
arrangement, the selected outcomes are neces- function of the mutual availability of child and
sarily delayed because all children make selec- experimenter in addition to parental consent.
tions before any context is experienced. The Sessions took place in a small room in close
outcomes also are probabilistic in that each proximity to the child’s regular classroom. The
child’s selected context will be experienced on room included a child-sized table and chairs.
only a proportion of trials. This probability Each single assessment was completed during
would vary as a function of the number of one session, and each group assessment was
contexts from which to select and the extent to completedwithineitheroneortwosessions,with
which a particular child’s selection was consis- each session separated by less than 24 hr.
tent with the majority of selections. Thus,
although a group-oriented assessment might Dependent Variables and Data Collection
and Analysis
allow a determination of preferences for
individual children in an efficient manner, it Theprimarydependentmeasurewasselection,
is unknown whether the delayed and probabi- defined as pointing to one item in an array of
listic outcomes that are inherent in these items. Although these children could vocalize
procedures would compromise the accuracy of theirpreferences,amotorresponsewaschosenas
the assessment results. the target behavior because motor responses are
Thus,theultimatereasonforconductingthis usually used in concurrent-chains assessments
study was to determine if group-oriented duethecomplicatednatureoftheexperiencesfor
procedures could be applied to determine whichpreferencesarebeingassessed,andbecause
children’s preferences for various contexts it allowed a child’s selections to remain covert
accurately. However, to conduct a comparative during group arrangements when multiple
analysis of a group concurrent-chains arrange- children were present simultaneously. Depend-
mentandatypicalarrangementfordetermining ingontheassessmenttype,theitemswereeither
preferences for contexts (referred to hereafter as actualfooditemsorsmallcoloredcardsthatwere
the single arrangement), it was necessary to correlated with the same food items. Rankings
simplify our analysis. We did this by determin- for each food were calculated by dividing the
ing preference hierarchies with individual totalnumberofselectionsofaparticularfoodby
children for edible items that could be thetotalnumberoftrialsduringwhichthatsame
consumed within seconds, as opposed to food was available, multiplying by 100%, and
contexts that require minutes to experience, then ordering the foods from the highest to
and comparing the hierarchies obtained with lowest percentage of selections (i.e., a 1 was
the single and group procedures. Our primary assigned to the food with the highest selection
question was whether the delayed and probabi- percentageanda4wasassignedtothefoodwith
listic nature of the group arrangement imposed the lowest selection percentage). Selection per-
unwanted variability on children’s preferences centages were equal occasionally, but these ties
as identified using single arrangements. were broken by assigning the food item chosen
on the last trial the higher rank; thus, all ranks
wereabsolutenumbers.Thelastfoodchosenwas
METHOD
assigned the higher rank because later choices
Participants and Setting followed the longer experience with the assess-
Participants included 14 children who were ment items. Food rankings across single and
enrolled in an inclusivepreschool classroom. All group arrangements were examined for each
participantswerebetween3 years2 monthsand child to determine if the group procedures
5 years 3 months old (M 5 4 years 3 months) imposed unwanted variability on preferences.
28 STACY A. LAYER et al.
Interobserver agreement was measured dur- ‘‘Point to the one you like the best’’) and then
ing 100% of sessions for every assessment and deliveredtheselecteditemtothechild;heorshe
child.Agreementwascalculatedbydividingthe hadupto30 stoconsumetheitem.Anyportion
number of trials with selection agreements of the item left on the plate after 30 s was
(observers recorded the same chosen food or disposed of so that the presence of the food did
colored card) by the number of trials and not affect selections on the next trial. If a child
multiplying by 100%. Mean agreement was didnotmakeaselection,theexperimenterwould
97% (range, 15% to 100%) for all children in have waited 5 s and then delivered the prompt
thesinglearrangementand98%(range,83%to again;however,thisdidnotoccurforanyofthe
100%) for all children in the group arrange- children. If the child pointed to two items
ment.Forthesinglesessioninwhichagreement simultaneously or made an unclear selection by
was under 80%, an examination of the data pointing between the two plates, the experi-
revealed that the secondary observer scored two menter said, ‘‘I am not sure which one you
selections as occurring during the same trial in wanted.Let’strythatagain.Pointtotheoneyou
thebeginningofthesession,andthisaffectedall likethebest.’’Afteraselection,theexperimenter
trials for the remainder of the session. replaced the food and then rotated the items so
thatthelastitemonthechild’srightbecamethe
Procedure firstitemontheleft,andtheplateswerearranged
Single arrangement. Our single arrangement equidistant from each other again. A new trial
procedures were similar to those described by thenbeganinwhichtheexperimenterinstructed
DeLeon and Iwata (1996) and Windsor, Piche, the child to pick his or her most preferred item.
and Locke (1994) in that we conducted a This was repeated until a preference was
multiple-stimulus preference assessment. The indicated according to one of the following
assessments were conducted with a single child, restriction criteria:(a) The childchose thesame
using the same four items for each child (jelly food four times consecutively, (b) a particular
beans, chocolate peanut butter balls, cheese food was selected four more times overall, (c)
crackers, and raisins). These particular foods two foods were selected equally compared to
werecutintosmallpiecesapproximatelythesize each other and four more times overall, or (d)
of the raisin and were chosen because they each 20totaltrialshadbeenconductedwithnoother
had a distinct flavor (e.g., sweet, nutty, salty, restriction criteria met. If a child met criteria
fruity).Acupofwaterwasavailablecontinuously (a), (b), or (c), that particular food was
to the childrenin all sessions. A singlechild was restricted and the child was presented with the
present during this assessment. Priorto the start remaining foods until the same criteria were
of the evaluation, a one-time sampling period met to obtain a rank of preference among
was arranged in which the experimenter labeled the food items. If a child met criterion (d), the
eachitemasheorsheplaceditonaplateinfront assessment ended. During each assessment, the
ofthechild,andthensuggestedthatthechildtry experimenter delivered a brief statement of
the item (e.g., ‘‘This is a raisin; try it’’). At the praise or social conversation unrelated to the
beginningoftheinitialassessmentforeachchild child’s choices (e.g., ‘‘You are sitting so nicely,’’
(the Single 1 assessment), the experimenter or ‘‘It is so sunny, I can’t wait to go outside
presented an array of four items in a half circle later’’) following every other trial, and also
onthetableinfrontofthechildsothatallitems briefly responded to social interactions initiated
were equidistant from each other and from the by the child at any time.
child. The experimenter instructed the child to Exposure. It was necessary to pair the food
point to his or her most preferred item (i.e., items with arbitrary and distinct stimuli (i.e.,
PREFERENCE ASSESSMENTS IN A GROUP ARRANGEMENT 29
2.5 cm by 2.5 cm colored cards) for two balls’’) following at least 24 hr since a previous
reasons. First, during the group arrangement, exposure session, the exposure period was
each child’s selection was placed in a box with considered complete and the children moved
theotherchildren’sselections,onewasdrawnat intoagrouparrangement.Themeannumberof
random, and all children received the item exposure sessions required for a child to learn
correlated with the card. It was more practical the correlations was 2 (range, 1 to 5).
to use cards instead of foods for this portion of Group arrangement. Prior to making selec-
the procedures because placing food items in tions,3childrenwereseatedatatableandwere
the box presents obvious problems with sanita- provided with instructions to keep their
tion and food waste. In addition, the group selections a secret to avoid the possibility of
arrangement was intended to simulate the one child’s selection influencing another’s. The
concurrent-chains procedures used to evaluate experimenter attempted to group children so
preferences for contexts; therefore, in future that as many children as possible within the
applications of these methods, the cards will grouphaddifferingfoodsrankedfirstaccording
represent experiential contexts that cannot to the single arrangement, and one group
possibly be placed in box. participated at a time. All children made their
Therefore,priortoevaluatingpreferencesina selections in private; to accomplish this, the
group arrangement, we conducted exposure experimenter held up a folding poster-board
sessions to teach the children the correlations barrier (1 m by 1 m) between the child
betweenthedifferentfoodsandcoloredcardsto choosing and the other children. An array of
be used as initial-link stimuli in the group thefourcardsthathadbeenpairedwitheachof
arrangement. Four colors were chosen random- the foods in the exposure sessions was arranged
ly to be correlated with each of the four foods infrontofeachchild.Eachchildwasinstructed
presented in the single arrangement. Three to point to the one that he or she liked best.
children simultaneously participated in this After each selection, the colored card corre-
procedure. During each trial, all children were sponding to each child’s choice was placed
instructed to point to one of four available inside a voting box, and this was repeated until
colored cards (each child had his or her own set each child had a card in the box. The
of four cards). After they pointed to the experimenter then set aside the barrier, showed
specified card, the experimenter provided brief theboxtoallofthechildren,anddrewonecard
praise and presented every child with the food out of the box. All children received the item
that was correlated with that color. The child correlated with the chosen card. This method
had access to the food for up to 30 s, and any allowedeachchildachanceof havinghisorher
portion of the item not left on the plate after selected card chosen; however, this also resulted
30 s was removed. This sequence of events was in food deliveries that were both delayed and
repeated for 20 trials, with each color and food probabilistic relative to the single arrangement.
association presented on five trials each in a The probability of a child receiving the item
randomandcounterbalancedorder.Priortothe that was selected was measured and is reported
start of each subsequent session, the experi- below.Bycontrast,delaysbetweenselectingand
menter asked each child what happened when receiving the foods were not measured but were
he or she pointed to each of the four cards. estimatedtorangebetween1to2 minpertrial.
When the child was able to describe the Eachcardwasreplacedaftereveryselectionso
contingencies correctly (e.g., the experimenter thatallchildrencontinuedtoselectfromthefull
asked, ‘‘What do you get when you touch array of four cards until a preference emerged.
pink?’’and the child responded, ‘‘peanut butter The different cards and their associated edible
30 STACY A. LAYER et al.
consequences were restricted using the same group arrangement was conducted next, fol-
criteria described in the single arrangement. lowedbyareversaltoasinglearrangement,and
Onceapreferencehierarchywasestablishedfora finally a replication of the group arrangement.
particularchild,heorshewasremovedfromthe All 14 children experienced all conditions in a
pool of children that could be grouped together similar manner.
for a session. Toensure groups of three for each
session, children who completed the assessment
RESULTS
weresometimesincludedingroupsatthecloseof
the overall evaluation, but their data were not The primary research question regarding the
used in the analyses. imposition of variability via the group arrange-
In sum, the single and group arrangements ment was addressed by evaluating changes in
were the same in that every child chose among rank of each of the four foods across repeated
the same four foods, and the same criteria for measures of the same arrangements, and also
determining preferences and assigning ranks to repeated measures of the two different arrange-
food items were used. The few important ments. Trial-by-trial data for 2 children are
differences between the two assessment formats shown in Figure 1. These are examples of two
were as follows. First, each child chose from an common types of performance patterns. Al’s
array of foods in the single arrangement, data provide an example of consistent prefer-
whereas each child chose a card that was ence rankings across both single and group
correlated with a food in the group arrange- assessment formats. He selected a peanut butter
ment. Second, each child was guaranteed to ballonthefirsttrial,followedbyajellybeanon
receive every item that he or she chose in the the second trial. Al again selected peanut butter
single arrangement, whereas the number of balls on the third and fourth trials, and then he
timeseach childreceived hisor her chosen item selected a cheese cracker. He next selected
was probabilistic in the group arrangement. peanut butter balls two more times. Peanut
Third, there was a longer delay between butter balls were then restricted because they
selections and delivery of food in the group hadbeen selectedfour moretimesoverall.After
arrangement than in the single arrangement. peanut butter balls were removed from the
Finally, each child was alone in the single array,Alswitchedselectionsbetweenjellybeans
arrangement,whereas3childrenwerepresentin andcheesecrackersuntileachwereselectedfour
the group arrangement. more times than the raisin (which was never
selected), and the end-of-session criterion had
Experimental Design also been met. Al continued to show similar
Concurrent-chains designs allowed prefer- patterns of responding throughout the rest of
ences to be determined within each assessment. the assessments, until the second group assess-
In addition, an AABAB design was used to ment when he chose each of the foods four
evaluatedifferencesinselectionsasafunctionof times consecutively. Throughout every assess-
the group arrangement. Initially, two single ment, peanut butter balls were always the most
arrangements (Condition A) were conducted preferred food, jelly beans second, cheese
consecutively to determine if preferences were crackers third, and raisins last. Thus, it did
stable across repeated measures of the same not appear that the group arrangement altered
assessment type before we attempted to deter- the primary outcome of the preference assess-
mine if the group arrangement (Condition B) ment in any appreciable way; the same
imposed unwanted variability on preferences. preference hierarchy resulted from all assess-
Next, the exposure condition was implemented ments. There was not a single participant for
to teach the card and food associations. The whom preferences were stable only across the
PREFERENCE ASSESSMENTS IN A GROUP ARRANGEMENT 31
Figure1. CumulativeselectionsacrosstrialsandrelativerankingsoffooditemsaredisplayedforAlinthetoppanel.
Thesedataillustrateanexampleofachildwhosepreferencehierarchydidnotchangeduringtheentireassessment.The
bottompanelshows cumulative selectionsacross trialsandindifference scores for Kay.Indifference scores describe the
amountofselectionvariabilitywithineachassessment.Thesedataillustrateanexampleofachildwithhighlyindifferent
responding during the single assessments andhighly discriminated responding during the group assessments.
single arrangements; preferences were either Considering that there was no difference
stableacrossbothassessmenttypesorpreferences between the percentage of children for whom
were variable regardless of assessment type. rankings were generally consistent for these two
We also examined the data using a criterion comparisons (e.g., Single 1 to Single 2 and
for determining an agreement. According to Single 2 to Group 1 on all rankings or just the
this criterion, a ranking agreement occurred top items), it seems unlikely that unwanted
when all foods did not move more than one variability in food rankings was imposed by the
rank across the two assessments. When com- group arrangement.
paring the first and second single arrangement To further describe the extent of agreement
using this criterion, rank agreements occurred within and across assessment types, a Spear-
for 69% of children. A comparison of the first man’s correlation analysis was conducted on
group assessment with the preceding single food rankings (Table 1). First, the rankings for
assessment revealed that rank agreements also all children for the initial assessment were
occurred for 69% of children. Another way to comparedtotherankingsyieldedbythesecond
examine rank consistency includes determining singleassessment(Single1/Single2inTable 1);
if the first ranked food remained the same this yielded a moderate, positive correlation of
across assessments. For single assessments, 57% .60.Correlationsofsimilarstrengthwerefound
of the children had their first ranked food between all like assessment types (r 5 .60 for
remainthesameacrossrepeatedmeasuresofthe Single1/Single2,r5.68forSingle2/Single3,
single arrangement. A comparison of the first r 5 .57 for Single 1/Single 3, and r 5 .72 for
group assessment with the preceding single Group 1/Group 2). These correlations were
assessment revealed that the first ranked food close to or higher than a generally acceptable
remained the same for 57% of children. test-retestreliabilitycoefficientof.60(Cicchetti
32 STACY A. LAYER et al.
Table 1
Correlations andDays ExpiredBetween Assessments
Samearrangementtype Differentarrangementtype
Single1/ Single2/ Single1/ Group1/ Single2/ Group1/ Single3/
Single2 Single3 Single3 Group2 Group1 Single3 Group2
Correlationsbetween .60 .68 .57 .72 .65 .74 .71
assessments
Meannumber(range)of 19(1–34) 30(7–60) 44(8–68) 28(7–70) 19(6–47) 10(1–49) 17(3–69)
daysbetweenassessments
Note.All correlations betweenassessments arestatistically significant atp ,.05.
& Sparrow, 1981). These data showed that that the probabilities that each child received
both single and group assessment types share a the same item selected varied considerably
similar amount of consistency across applica- across children. For example, the probability
tions. The correlations generated between the that Eva received the item that she selected in
singleandgroupassessmentswerealsoallabove the second group arrangement was .83, whereas
.60(r5.65forSingle2/Group1,r5.74for the probability that Kat received the item that
Group 1/Single 3, r 5 .71 for Single 3/Group she selected in the second group arrangement
2). All of the above correlations are statistically was .08 (Table 2). The mean probability of
significant (p , .05). By comparing the sets of each child receiving the item that he or she
correlations, it is apparent that the correlations chose in the group arrangement was .53 in the
within assessment types were of the same current study. Thus, children’s selections re-
magnitude as those correlations between assess- sulted in access to their chosen item on about
ment types. Thus, these data are also consistent half the trials. To determine if the probabilistic
with the notion that the group assessment did outcomes influenced the extent of agreement
not impose unwanted variability on the out- between the single and group assessments, a
comes of the preference assessments. Pearson correlation analysis between the mean
Because the amount of time between assess- probability that a child received selected items
ments varied unsystematically, we determined and the correlations between the single and
whether the amount of time between assess-
ments affected the correlations between assess- Table 2
ments (see Table 1 for the mean number of ProbabilityThataChildReceivedtheSameItemSelected
days between assessments). A Pearson correla-
Assessment
tion analysis involving time between assess-
ments and the rank correlations between the Group1 Group2
assessmentsshowedthatthesevariableswerenot Ed .44 .43
correlated (r 5 .04), suggesting that variations Kay .39 .67
Ben .43
in the rank correlations were probably not due Roy .44 .75
to the differences in time between assessments. Lea .56 .50
Mia .57 .69
Oneofthefeaturesofthegrouparrangement
Roz .44 .42
was that it was probabilistic that each child Kat .43 .08
Abe .43 .53
would receive the item that he or she preferred
Mel .58 .53
the most; traditional preference assessments Eva .62 .83
Joy .46 .58
guarantee that the child receives every item that
Jim .65 .53
he or she selects (i.e., the probability of Al .77 .50
M .52 .54
receiving a selected item is 1). Table 2 shows
PREFERENCE ASSESSMENTS IN A GROUP ARRANGEMENT 33
group arrangements was conducted. Our con-
cern was that children with lower probabilities
of receiving the items they selected (e.g., Kat)
would be more likely to have inconsistent
assessment outcomes. A weak, positive, and
statistically insignificant correlation found be-
tweenthesevariables(r5.38,p..05)suggests
that this was not the case. In addition, a scatter
plot showed that this weak relation was highly
dependent on two outliers. Thus, the probabi- Figure 2. The mean indifference scores and standard
listic selection outcomes were likely not respon- deviationsofthosescoresforall14children.Indifference
sible for any disruptions in preference observed scores refer to the amount of selection variability within
an assessment.
between single and group assessments.
Although preference hierarchies were not
systematically different between single and criteriametplusone)toobtainascoreof3.This
group arrangements, the patterns in the data is representative of the lowest possible i score.
were different for each assessment type with Kay’sdata(Figure 1)alsoillustrateacommon
respect to within-assessment variability. As we pattern of within-session variability observed
analyzed the data with respect to our primary within the single and group arrangements.
question, it became apparent that children’s During the first and second single assessments,
selections were more highly discriminated (i.e., Kayallocatedafairlyequalnumberofselections
there was less within-session selection variabil- between peanut butter balls, jelly beans, and
ity) during the group arrangements. cheese crackers. In the first group assessment,
A metricto describeand quantify theamount selectionsweremorehighlydiscriminated.When
of selection variability within and across assess- a third single assessment was conducted, she
mentswasderivedfromthedata.Werefertothis again frequently switched selections among all
metric as an indifference score (i score) because the foods. Finally, in the second group assess-
indiscriminate responding would yield relatively ment, Kay engaged in the most discriminated
high scores and discriminated responding would responding by selecting peanut butter balls
yield relatively lower scores. These scores were four times consecutively, followed byjelly beans
calculated by dividing the total number of trials fourtimesconsecutivelyandcheesecrackersfour
in each assessment by the number of restriction timesconsecutively.Kay’sdatarepresentastrong
criteriametplusone(onerepresentedtheend-of- example of the differencesin variability between
session criterion). The i scores ranged from 3 the single and group assessments. Her respond-
(highly discriminated) to 20 (highly indifferent). ing was indifferent during all single assessments
For an example of a high i score, see Kay’s first with i scores of 20 (the maximum score), and
single assessment (Figure 1). Twenty trials were highly discriminative responding was observed
conducted without Kay meeting a single restric- during the group assessments with scores of 3.3
tion criterion. Therefore, the i score was and 3.0 (the minimum score), respectively,
calculated by dividing the 20 total trials by 1, during each group assessment. These data show
for a score of 20, the maximum score that could that the group assessment resulted in more
be achieved. By contrast, Kay’s second group discriminated selections than the single assess-
assessment wasconductedfor atotalof12 trials, ments for Kay.
andthreerestrictioncriteriaweremet.Theiscore Figure 2 shows the mean i scores across
was calculated by dividing 12 by 4 (restriction children for each of the five assessments. The
34 STACY A. LAYER et al.
Figure 3. The mean indifference score byassessment type for each individual child.
data reveal that mean i scores were higher in tions and disrupt the rank of preferences. The
each of the single arrangements than in the evidence against negative effects of the group
group arrangements. In addition, the standard arrangement were as follows: (a) There was no
deviation of scores was higher in each of the single example of stability in ranks being
single arrangements than in the group arrange- disrupted systematically by the group assess-
ments. This effect was replicated in a reversal ment procedures, (b) the consistency in rank-
design for the grouped data. A different ings was similar for like and unlike assessment
representationoftheiscoresisseeninFigure 3, types, and (c) correlation coefficients were
which shows the mean i score by assessment similar for like and unlike assessment types. In
typeforeachindividualchild.Forallchildren,i fact, for 11 of 14 participants, more highly
scores in the single arrangements were either discriminated responding, and thus more
higher than or equal to i scores in the group efficient determinations of preference, was
arrangements. In addition, the range of group- associated with the group assessment. Thus,
arrangement i scores was minimal (3 to 7.5) notonlydidthegrouparrangementnotimpose
relative to the range of single-arrangement i unwanted variability on selections during the
scores (3 to 20). To determine if a relation identification of preferences, but the assessment
existed between the probability that a child actually minimized selection variability for 12
received the same item he or she selected and i of14childrenandbyamagnitudeofover50%
scores, a Pearson correlational analysis was for 10 of these 12 children.
conducted on these variables. A weak, negative However, there were some limitations of this
correlation of 2.28 resulted, indicating that study that are worth noting. First, because all
selections were slightly more likely to reflect assessments were conducted in the same order
indifference when the child was less likely to for all children, it is possible that the more
receive his or her selected item. highly discriminated performances during the
groupassessmentsmayhavebeendependenton
the fact that the single assessments routinely
DISCUSSION
preceded them. Conducting the conditions in a
This study sought to determine if group- BBABA design for some participants would
oriented preference assessments, in which have allowed the influence of an order effect to
outcomes were both delayed and probabilistic, be detected, although not necessarily controlled
would impose unwanted variability on selec- (Hains & Baer, 1989). In addition, children