The role of self-esteem in the misinformation effect

This article was downloaded by: [Jo Saunders]On: 13 January 2012, At: 03:34Publisher: Psychology PressInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office: MortimerHouse, 37-41 Mortimer Street, London W1T 3JH, UK

MemoryPublication details, including instructions for authors and subscription information:http://www.tandfonline.com/loi/pmem20

The role of self-esteem in the misinformationeffectJo Saunders aa Psychology Department, Swansea University, Swansea, UK

Available online: 13 Jan 2012

To cite this article: Jo Saunders (2012): The role of self-esteem in the misinformation effect, Memory,DOI:10.1080/09658211.2011.640690

To link to this article: http://dx.doi.org/10.1080/09658211.2011.640690

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions

This article may be used for research, teaching, and private study purposes. Any substantial orsystematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distributionin any form to anyone is expressly forbidden.

The publisher does not give any warranty express or implied or make any representation that thecontents will be complete or accurate or up to date. The accuracy of any instructions, formulae, anddrug doses should be independently verified with primary sources. The publisher shall not be liable forany loss, actions, claims, proceedings, demand, or costs or damages whatsoever or howsoever causedarising directly or indirectly in connection with or arising out of the use of this material.

The role of self-esteem in the misinformation effect

Jo Saunders

Psychology Department, Swansea University, Swansea, UK

Previous research using the Gudjonnson suggestibility scale has suggested a role for self-esteem insuggestibility, with participants low in self-esteem being more suggestible than participants high in self-esteem. Four experiments are presented examining the role of self-esteem in the misinformation effectand whether enhanced suggestibility effects in participants low in self-esteem reflect genuine memoryimpairment. In Experiments 1 and 4 participants completed a standard recognition test. In Experiment 2participants completed the modified recognition test. In Experiment 3 participants completed a freerecall test. In Experiments 1 and 4 participants low in self-esteem demonstrated greater misinformationeffects than participants high in self-esteem. In Experiment 3 a 3-day retention interval was employedwith the modified test and no differences were found between the two groups on the reporting of the newitem. The findings suggest that participants low in self-esteem are particularly sensitive to demandcharacteristics and post-event suggestion but do not suffer from genuine memory impairment.

Keywords: Self-esteem; Misinformation effect; Eyewitness memory.

The accuracy of eyewitness memory has been atopic of fascination for decades. In an attempt tounderstand the processes governing eyewitnessmemory few mental stones have been left un-turned. A prodigious level of research has sug-gested that eyewitnesses are notoriouslyunreliable, misremembering ‘‘Yield’’ signs as‘‘Stop’’ signs (Loftus, Miller, & Burns, 1978),green cars as blue (Loftus, 1977), screwdriversas wrenches (Loftus, Donders, Hoffman, &Schooler, 1989), and green water pitchers asyellow (Belli, 1988). Peripheral information alsoappears to be most vulnerable to misleading post-event information (Wright & Stroud, 1998), andthe presence of a weapon exacerbates the forget-ting effect (Saunders, 2009; for a review seePickel, 2007).

The suggestibility of memory was first studiedby Loftus and colleagues (1978). They showedparticipants a slide sequence of a car accident. Inthe slides participants saw a car stopped at eithera stop sign or a yield sign. Participants then

received misleading information about the sign,embedded in a question. Participants then com-pleted a forced-choice recognition task wherebythey had to choose the correct sign they had seenin the slides from a choice of the correct item orthe misinformation item. Loftus and colleaguesfound that participants who had been misledchose the original item less often than partici-pants who had not been misled, thereby demon-strating the misinformation effect.

To account for these findings Loftus andcolleagues (1978) proposed that the presentationof post-event misleading information led to theerasure of the original item from memory. Thistrace alteration account suggested that the mem-ory trace of the original item was destroyed andpermanently erased from memory. This contro-versial claim was quickly contested. Bekerian andBowers (1983) suggested that the original itemremains in memory but is temporarily unavailabledue to retrieval failures. They suggested that theretrieval cues present on the final recognition test

Address correspondence to: Jo Saunders, Psychology Department, Swansea University, Singleton Park, Swansea SA2 8PP,

Wales, UK. E-mail: [email protected]

MEMORY, 2012, iFirst, 1�10

# 2012 Psychology Press, an imprint of the Taylor & Francis Group, an Informa businesshttp://www.psypress.com/memory http://dx.doi.org/10.1080/09658211.2011.640690

Dow

nloa

ded

by [

Jo S

aund

ers]

at 0

3:34

13

Janu

ary

2012

were a better match to the misleading item thanto the original item. When they presented theforced-choice recognition test in the same tem-poral order as the original slide sequence themisinformation effect was abolished; when therecognition test was presented in a random order,participants favoured the misinformation item.This suggests that when the questions on therecognition test were presented in a sequentialorder there was a better match between theretrieval cues at test and encoding allowing forthe selective access of the original item. Toaccount for these findings Bekerian and Bowersproposed the coexistence hypothesis which sug-gests that the original and misleading tracesremain in memory but that retrieval failureslead to preferential access to the misleading trace(see also Christiaansen & Ochalek, 1983).

Misled participants may also make sourcemisattribution errors concerning the source ofthe misleading item. The sources of the originaland post-event information are very similar toone another, both referring to an eyewitnessevent. Thus the ideal conditions occur for a sourcemisattribution to occur. Lindsay and Johnson(1989a) suggested that the original and misleadingitems remain in memory but that misled partici-pants make a source-monitoring error concerningthe misleading item; that is, they make an errorconcerning the source of their memories (see alsoLindsay, 1990; Lindsay & Johnson, 1989b).

It may be the case, however, that misinforma-tion has no effect on memory but rather reflectsthe actions of a flawed paradigm (McCloskey &Zaragoza, 1985). It has been suggested that therecognition test used, which places the originalitem in opposition to the misleading item, en-courages demand characteristics. McCloskey andZaragoza suggest that there may be a percentageof participants in the misled group who rememberthe original and misleading items, but know thatthe experimenter designed the post-event ques-tions so must know what was in the slides.Participants might therefore infer that the experi-menter may want them to choose the misleadingitem even though they can remember the originalitem. No such demand characteristics exist in thenon-misled control group. Thus participants in themisled group may choose the misleading item forreasons other than the misleading item affectingthe accessibility of the original item.

Misled participants also have a second sourceto rely on to guide their memory choices on therecognition test: the post-event source. If a

percentage of participants never encoded theoriginal item then they can rely on the post-eventsource to guide their memory. Participants in thenon-misled control condition do not have thissecond source of memory to base their choices on.

The main source of these confounds is therecognition test. Placing the original item incompetition with the misleading item may en-courage demand characteristics and change parti-cipants’ choice at test even if the misleading itemhas no effect on memory. One-way around theseproblems is to remove the misleading item as achoice on the test and replace it with a novel item.This modified recognition test removes the effectof demand characteristics and overcomes theproblem of participants who did not encode theoriginal item but do remember the misleadingitem.

When McCloskey and Zaragoza (1985) usedthe modified recognition test they failed to findany misinformation effects. They suggested thatthe misinformation effect is not a genuine findingand is the product of a flawed paradigm. How-ever, others have found misinformation effectsusing the modified test (Chandler, 1989, 1991;Chandler & Gargano, 1998). Typically thesestudies have used retention intervals of 3 to 7days (Belli, Windschitl, McCarthy & Winfrey,1992; Ceci, Ross & Toglia, 1987; Toglia, Ross,Ceci, & Hembrooke, 1992) while studies thathave failed to find misinformation effects usingthe modified test have used shorter retentionintervals of 30 minutes or less (Belli, 1993; Belliet al., 1992; Bonto & Payne, 1991; Loftus et al.,1989). This suggests that a degree of decay of theoriginal memory trace must occur prior to ex-posure to the misleading item for the misinforma-tion effect to occur.

Aside from the experimental approach, sug-gestibility has also been investigated using indivi-dual differences. This form of suggestibility hasbeen termed ‘‘interrogative suggestibility’’ andGudjonsson and Clarke (1986) described it as‘‘the extent to which, within a closed socialinteraction, people come to accept messagescommunicated during formal questioning, as aresult of which their subsequent behaviouralresponse is affected’’ (p. 84). Interrogative sug-gestibility is examined using the Gudjonssonsuggestibility scale (GSS) whereby participantsare read a narrative that they then recall im-mediately and after a delay. Participants are thenasked 20 questions of which 15 are leadingquestions. After completing all of the questions

2 SAUNDERS

Dow

nloa

ded

by [

Jo S

aund

ers]

at 0

3:34

13

Janu

ary

2012

they receive negative feedback that they got some

of their answers wrong. The 20 questions are then

repeated. Two measures of the GSS are taken:

yield which refers to the tendency to go along

with leading questions, and shift which refers to

the tendency to change answers following nega-

tive feedback.Individual differences have been noted to

influence interrogative suggestibility. Participants

who are anxious have been found to be particu-

larly suggestible (Gudjonsson & Lister, 1984).

Gudjonsson and Lister suggested that inter-

viewers who attempted to lower participants’

self-esteem might make participants particularly

prone to suggestibility. Singh and Gudjonsson

(1984) added that self-esteem is most likely to

interact with unfamiliar interview procedures so

that participants low in self-esteem become more

suggestible when they are in unfamiliar interview

contexts. Gudjonsson and Sigurdsson (2003) sug-

gested that participants low in self-esteem are

‘‘vulnerable to giving into pressure from others

because of their eagerness to please and accom-

panying reluctance to engage in confrontation

with people’’ (p. 122).The role of self-esteem in memory perfor-

mance remains under-investigated. The current

experiments aim to examine whether individuals

low in self-esteem are more likely to demonstrate

more memory errors than individuals high in

self-esteem; more specifically, are they more

likely to report post-event misleading informa-

tion than individuals high in self-esteem. Do

individuals low in self-esteem demonstrate gen-

uine memory impairment or do they simply go

along with post-event suggestion from the ex-

perimenter? One-way to examine this issue is to

compare participants’ performance on the stan-

dard recognition test whereby the misleading

item is presented in competition with the original

item with participants’ performance on the

modified recognition test where the misleading

item is not presented as an option on the test. If

participants with low self-esteem have genuine

memory impairment, then they will select the

original item less than participants with high self-

esteem on the modified recognition test. If

participants with low self-esteem are vulnerable

to demand characteristics that are not present on

the modified recognition test, then these partici-

pants will select the original item at a similar rate

to participant high in self-esteem.

EXPERIMENT 1: MISINFORMATIONEFFECTS USING THE STANDARD

RECOGNITION TEST

Method

Participants and design. A total of 100 under-graduate psychology (50 males, 50 females,M age�23.5) students participated in exchangefor course credit. The study had a 2 (self-esteem:high or low)�2 (misled: misinformation or nomisinformation)�2 (misled item: salient andnon-salient) mixed design with repeated measureson the latter factor.

Materials and procedure. Participants arrived atthe laboratory and were greeted by a femaleexperimenter. Participants first completed theRosenberg (1965) self-esteem questionnaire,which consists of 10 statements about generalfeelings the participants have about themselves(e.g., at times I think that I am no good at all; Ifeel I do not have much to be proud of). TheRosenberg questionnaire is based on a Likertscale and participants have to indicate whetherthey strongly agree, agree, disagree, or stronglydisagree with each statement. Participants werethen divided into two groups based on a mediumsplit of their scores, producing a high self-esteemgroup (M��28) and low self-esteem group(M�B 28). Participants then watched a slide-show of an event depicting a woman walkingthrough the park before being approached by agroup of men who produce a knife and steal herhandbag. Six slides were shown and each slidewas shown for 5 seconds. Participants thenreceived a questionnaire containing 10 questionsabout the event containing questions about sali-ent and non-salient items. Salient and non-salientitems were determined in a pilot study whereby20 participants watched the slide show and thencompleted a free recall test. This produced a listof items that were recalled by the majority ofparticipants and a list of items that were poorlyremembered. The items that were well remem-bered formed the salient items and the items thatwere poorly remembered formed the non-salientitems. From this list two salient items and twonon-salient items were used as a misleading item.The salient misleading questions were ‘‘Whicharm was the girl carrying the Sainsbury’s carrierbag on?’’ when the carrier bag was a Tesco bag,and ‘‘Which hand was the attacker with the Nikehooded top holding the knife?’’ when the top read

MISINFORMATION EFFECT 3

Dow

nloa

ded

by [

Jo S

aund

ers]

at 0

3:34

13

Janu

ary

2012

‘‘Gap’’. The non-salient misleading questionswere ‘‘How many red dog bins were in thebackground?’’ when the correct colour was green,and ‘‘Which arm was the woman carrying thebrown handbag’’ when the correct colour wasblack. Three questions were also asked aboutsalient items and three questions askedabout non-salient items. Order of presentationof salient and non-salient questions was rando-mised. Participants in the non-misled conditionworked on anagrams for 2 minutes. Participantsthen completed a name generation task for 5minutes. Following completion of the distractortask participants were presented with a recogni-tion task. Participants were asked questions aboutsalient and non-salient items. One question re-lated to the salient misleading item and one to thenon-salient misleading item. Participants weregiven two choices: the correct item and themisleading item. On completion, participantswere thanked, debriefed, and dismissed.

Results and discussion

To examine whether there were any differences inthe reporting of the original item on the recogni-tion test between high and low self-esteem groupsa one-way ANOVA was conducted on the salientitems. It revealed an effect of self-esteem, F(3,96)�3.26, pB.05, h2�.09, suggesting a differ-ence in the reporting of the original item betweengroups. A one-way ANOVA was also conductedon the non-salient items and revealed an effect ofself-esteem, F (3, 96)�4.94, pB.01, h2�.14.

To determine whether any misinformationeffects were evident a series of independent t-testswere conducted. They revealed that in the highself-esteem group no differences were detected forsalient items between misled (M�.52) and non-misled participants (M�.60), t(48)��.70, ns.For non-salient items no differences were foundbetween the misled (M�.60) and non-misledparticipants (M�.66), t(48)��.54, ns. In thelow self-esteem group misled participants re-ported the original salient item (M�.32) lessoften than non-misled participants (M�.62),t(48)��2.98, pB.01, and misled participantsreported the non-salient original item (M�.28)less often than non-misled participants (M�.48),t(48)��1.96, pB.05.

In the misled conditions participants in thehigh self-esteem condition reported the salientoriginal item (M�.52) more often than the low

self-esteem participants (M�.32), t(48)�2.06,pB.05, see Table 1. Participants in the low self-esteem condition reported the non-salient origi-nal item (M�.28) less often than participants inthe high self-esteem condition (M�.60),t(48)�3.07, pB.01.

Memory for the non-target salient and non-salient details was also examined. There werethree salient and three non-salient items. Recog-nition of salient items in the high self-esteemcondition was .66 (SD�.27) and recognition ofnon-salient items was .51 (SD�.26). In the lowself-esteem condition recognition of salient itemswas .61 (SD�.30) and recognition of non-salientitems was .47 (SD�.26). A 2 (detail: salient andnon-salient)�2 (self-esteem: high or low) mixedANOVA was conducted and revealed an effect ofdetail, F(1, 98)�16.12, pB.001, h2�.14, but nointeraction between detail and self-esteem, F(1,98)�.03, ns. To determine whether there wereany differences between the two groups in recallof salient and non-salient details independent t-tests were conducted. They revealed no differ-ences between the two groups in the reporting ofsalient, t(98)�.37, ns, and non-salient details,t(98)�.59, ns.

Participants with low self-esteem were foundto be more suggestible to post-event suggestionwith more misinformation effects found for bothcentral and peripheral items. However, perfor-mance on non-target items showed no differencein recognition of these items between groups,suggesting that low self-esteem participants mayhave been responding to demand characteristicsin the standard recognition test. The standardrecognition task is known to suffer from demandcharacteristics whereby participants may selectthe misleading item because they know that theexperimenter composed the questions and somust know what was in the slides (McCloskey &Zaragoza, 1985). Participants low in self-esteemmay therefore select the misleading item because

TABLE 1

Reporting of original item on standard recognition test in

Experiment 1

High

misled

Low

misled

High non-

misled

Low non-

misled

Salient .52 .32 .60 .62

(.37) (.32) (.41) (.39)

Non-salient .60 .28 .66 .48

(.41) (.33) (.37) (.39)

Standard deviations included in parentheses.

4 SAUNDERS

Dow

nloa

ded

by [

Jo S

aund

ers]

at 0

3:34

13

Janu

ary

2012

they think that the experimenter wants them to,rather than because they have genuine memoryimpairment. To examine this issue, in Experiment2 the standard recognition test was replaced withthe modified recognition test, which places theoriginal item in competition with a novel itemrather than the misleading item. If participantswith low self-esteem have genuine memory im-pairment, then they will select the original itemless often than participants high in self-esteem. Ifparticipants low in self-esteem are susceptible toexternal pressure, then no misinformation effectwill be found.

EXPERIMENT 2: MISINFORMATIONEFFECTS USING THE MODIFIED

RECOGNITION TEST

Method

Participants and design. A total of 100 under-graduate psychology students (50 male, 50 female,M age�23.7) participated in exchange for coursecredit. The study had a 2 (self-esteem: high orlow)�2 (misled: misinformation or no mis-information)�2 (misled item: salient and non-salient) mixed design with repeated measures onthe latter factor.

Materials and procedure. Participants arrived atthe laboratory and were greeted by a femaleexperimenter. Participants first completedthe Rosenberg self-esteem questionnaire andwere then divided into two groups based on amedium split of their scores producing a high self-esteem group (M�� 26) and low self-esteemgroup (M�B 26). The materials and procedurewere the same as for Experiment 1 with theexception of the recognition test in which thecritical question referring to the misinformationeffect gave participants a choice between theoriginal item and a new item.

Results and discussion

To examine whether there were any differences inthe reporting of the new item on the recognitiontest between high and low self-esteem groups aone-way ANOVA was conducted on the salientitems. It revealed no effect of self-esteem, F(3,96)�.22, ns. A similar one-way ANOVA was

conducted on the non-salient items, which re-vealed no effect of self-esteem, F(3, 96)�.11, ns.

To determine whether there were any misin-formation effects present a series of independentt-tests were conducted between the misled andnon-misled conditions. They revealed that for thehigh self-esteem group there was no difference inthe reporting of the new item for salient,t(48)��.42, ns, and for non-salient items,t(48)�.73, ns. For the low self-esteem group nodifferences were detected between the misled andnon-misled conditions for salient, t(48)��.40,ns, and non-salient items, t(48)��.20, ns.

Participants in the high self-esteem conditionreported the salient new item (M�.20) less oftenthan the low self-esteem participants (M�.24)although this difference did not reach signifi-cance, t(48)��.42, ns, see Table 2. Participantsin the low self-esteem condition reported the non-salient new item (M�.32) less often than parti-cipants in the high self-esteem condition(M�.34) although this difference did not reachsignificance, t(48)��.20, ns.

Memory for the non-target items was alsoexamined. In the high self-esteem conditionreporting of salient details was .65 (SD�.26)and non-salient details was .50 (SD�.28). In thelow self-esteem condition reporting of salientdetails was .64 (SD�.23) and non-salient detailswas .50 (SD�.26). A 2 (detail: salient and non-salient)�2 (self-esteem: high or low) mixedANOVA was performed and revealed an effectof detail, F(1, 98)�14.22, pB.001, h2�.13, butno interaction between detail and self-esteem,F(1, 98)�.00, ns. To determine whether there wasany differences in reporting of details betweengroups, a series of independent t-tests wereconducted. They revealed no differences betweenthe groups in the reporting of salient, t(98)�.14,ns, and non-salient details, t(98)�.12, ns.

Participants low in self-esteem were not foundto be more suggestible than participants high in

TABLE 2

Reporting of new item on modified recognition test in

Experiment 2

High

misled

Low

misled

High non-

misled

Low non-

misled

Central .20 .24 .24 .28

(.32) (.36) (.36) (.36)

Peripheral .34 .32 .38 .34

(.37) (.35) (.44) (.35)

Standard deviations included in parentheses.

MISINFORMATION EFFECT 5

Dow

nloa

ded

by [

Jo S

aund

ers]

at 0

3:34

13

Janu

ary

2012

self-esteem when the modified recognition testwas used. This suggests that the increased sug-gestibility of low self-esteem participants in Ex-periment 1 was due to the external pressureexerted by the experimenter rather than due togenuine memory impairment. In Experiment 1the standard recognition test was used, which issusceptible to demand characteristics. Partici-pants high in self-esteem were found not toexhibit a misinformation effect suggesting thatthey do not respond to demand characteristics.

One reason why misinformation effects maynot be apparent on the modified recognition testis that the test appears directly after the mis-information phase. Previous work using the mod-ified recognition test has found thatmisinformation effects do not emerge when ashort retention interval has been utilised (Belliet al., 1992; Bonto & Payne, 1991; Loftus et al.,1989; McCloskey & Zaragoza, 1985). When alonger retention interval is used (e.g., 3�7 days)misinformation effects emerge (Belli et al., 1992;Ceci et al., 1987; Toglia et al., 1992). Misinforma-tion effects may emerge at longer retentionintervals because the original memory trace hasweakened prior to the introduction of misinfor-mation. If low self-esteem participants’ misinfor-mation effect in Experiment 1 was due to genuinememory impairment then they should exhibit alarger misinformation effect on the modified testwhen longer retention intervals are employed.

EXPERIMENT 3: MISINFORMATIONEFFECTS USING DELAYS IN THEMODIFIED RECOGNITION TEST

Method

A total of 100 undergraduate psychology students(50 male, 50 female, M age�22.1) participated inexchange for course credit. The study had a 2(self-esteem: high or low)�2 (misled: misinfor-mation or no misinformation)�2 (misled item:salient and non-salient) mixed design with re-peated measures on the latter factor.

Materials and procedure. Participants arrived atthe laboratory and were greeted by a femaleexperimenter. Participants first completed theRosenberg self-esteem questionnaire and werethen divided into two groups based on a mediumsplit of their scores producing a high self-esteem

group (M�� 25) and low self-esteem group(M�B 25). The materials and procedure werethe same as for Experiment 2 with the exceptionthat a 3-day delay was inserted between theoriginal slide sequence and the misinformationphases.

Results

To examine whether there were any differences inthe reporting of the new item on the recognitiontest between high and low self-esteem groups aone-way ANOVA was conducted on the salientitems. It revealed no effect of self-esteem, F(3,96)�.17, ns. A one-way ANOVA was also con-ducted on the non-salient items and revealed noeffect of self-esteem, F(3, 96)�.20, ns.

To determine whether any misinformationeffects were evident after the delay, independentt-tests were conducted. They revealed that in thehigh self-esteem group there was no differencebetween the misled (M�.46) and non-misled(M�.40) conditions in the reporting of the newsalient item, t(48)�.51, ns. For non-salient itemsno differences were found between the misled(M�.60) and non-misled (M�.66) groups,t(48)�.56, ns. In the low self-esteem group nodifferences were detected between the misled(M�.48) and non-misled participants (M�.44),t(48)�.35, ns, and no differences were foundbetween the misled participants reporting of thenon-salient new item (M�.60) and the non-misled participants (M�.58), t(48)�.18, ns.

Participants in the high self-esteem conditionreported the salient new item (M�.46) less oftenthan the low self-esteem participants (M�.48)but this difference did not reach significance,t(48)��17, ns, see Table 3. No differenceswere detected between the low self-esteem groupin the reporting of the non-salient new item

TABLE 3

Recall of the new item on the modified recognition test for

Experiment 3

High

misled

Low

misled

High non-

misled

Low non-

misled

Central .46 .48 .40 .44

(.43) (.42) (.41) (.39)

Peripheral .60 .60 .66 .58

(.38) (.41) (.37) (.37)

Standard deviations included in parentheses.

6 SAUNDERS

Dow

nloa

ded

by [

Jo S

aund

ers]

at 0

3:34

13

Janu

ary

2012

(M�.60) in comparison to the high self-esteemgroup (M�.60), t(48)�.00, ns.

Memory for non-target items was also exam-ined. In the high self-esteem condition the recallof salient details was .55 (SD�.31) and non-salient details was .45 (SD�.30). In the low self-esteem condition recall of salient details was.60 (SD�.28) and non-salient details was .45(SD�.29). A 2 (detail: salient and non-sal-ient)�2 (self-esteem: high or low) mixed ANO-VA was performed and revealed a main effect ofdetail, F(1, 98)�9.55, pB.01, h2�.09, but nointeraction between detail and self-esteem, F(1,98)�.38, ns. To determine whether there wereany differences between the groups in the report-ing of salient and non-salient details a series ofindependent t-tests were conducted. They re-vealed no differences between the two group inthe reporting of salient details, t(98)��.78, ns,or non-salient details, t(98)�.00, ns.

Participants low in self-esteem were found tobe more susceptible to misleading post-eventinformation in Experiment 1 but this effect wasabolished in Experiments 2 and 3. This may bebecause low self-esteem participants are onlymisled when demand characteristics are not con-trolled for, or it could be due to a type 1 error inExperiment 1. To determine if this was the caseExperiment 1 was replicated using new materials.

EXPERIMENT 4: REPLICATION OFMISINFORMATION EFFECTS USINGTHE STANDARD RECOGNITION TEST

Method

Participants and design. A total of 100 under-graduate psychology (50 males, 50 females,M age�22.9) students participated in exchangefor course credit. The study had a 2 (self-esteem:high or low)�2 (misled: misinformation or nomisinformation)�2 (misled item: salient andnon-salient) mixed design with repeated measureson the latter factor.

Materials and procedure. Participants arrived atthe laboratory and were greeted by a femaleexperimenter. Participants first completed theRosenberg (1965) self-esteem questionnaire. Par-ticipants were then divided into two groups basedon a medium split of their scores, producing ahigh self-esteem group (M�� 27) and low self-esteem group (M�B 27). Participants then

watched a slideshow of an event depicting aman being shot in a drive-by shooting. Six slideswere shown and each slide was shown for 5seconds. Participants then received a question-naire containing 10 questions about the event,including questions about salient and non-salientitems. Salient and non-salient items were deter-mined in a pilot study whereby 20 participantswatched the slide show and then completed a freerecall test. This produced a list of salient and non-salient items. From this list two salient items andtwo non-salient items were used as a misleadingitem. The salient misleading questions were‘‘What hand was the gunman holding the blackgun?’’ when the gun was silver, and ‘‘What type ofshop did the brown car drive past?’’ when the carwas maroon. The non-salient misleading ques-tions were ‘‘What colour top was the victimwearing when he came out of number 17?’’ whenthe house number was 19, and ‘‘What colourtrousers was the victim wearing when he waslocking the blue door?’’ when the correct colourwas white. Three questions were also askedabout salient items and three questions askedabout non-salient items. Order of presentation ofsalient and non-salient questions was randomised.Participants in the non-misled condition workedon anagrams for 2 minutes. Participants thencompleted a name generation task for 5 minutes.Following completion of the distractor task parti-cipants were presented with a recognition task.Participants were asked questions about salientand non-salient items. One question related to thesalient misleading item and one to the non-salientmisleading item. Participants were given twochoices: the correct item and the misleadingitem. On completion participants were thanked,debriefed, and dismissed.

Results

To examine whether there were any differences inthe reporting of the original item on the recogni-tion test between high and low self-esteemgroups, a one-way ANOVA was conducted onthe salient items. It revealed an effect of self-esteem, F(3, 96)�4.12, pB.05, h2�.04, suggest-ing a difference in the reporting of the originalitem between groups. A one-way ANOVA wasalso conducted on the non-salient items andrevealed an effect of self-esteem, F(3, 96)�4.38,pB.01, h2�.04.

MISINFORMATION EFFECT 7

Dow

nloa

ded

by [

Jo S

aund

ers]

at 0

3:34

13

Janu

ary

2012

To determine whether any misinformationeffects were evident a series of independent t-testswere conducted. They revealed that in the highself-esteem group no differences were detected forsalient items between misled (M�.64) and non-misled participants (M�.68), t(48)��.35, ns.For non-salient items no differences were foundbetween the misled (M�.58) and non-misledparticipants (M�.62), t(48)��.39, ns. In thelow self-esteem group, misled participants re-ported the original salient item (M�.34) lessoften than non-misled participants (M�.66),t(48)��3.14, pB.01, and misled participantsreported the non-salient original item (M�.28)less often than non-misled participants (M�.60),t(48)��3.07, pB.01.

In the misled conditions participants in thehigh self-esteem condition reported the salientoriginal item (M�.64) more often than the lowself-esteem participants (M�.34), t(48)�2.75,pB.01, see Table 4. Participants in the low self-esteem condition reported the non-salient origi-nal item (M�.28) less often than participants inthe high self-esteem condition (M�.58),t(48)�3.03, pB.01.

Memory for the non-target salient and non-salient details was also examined. There werethree salient and three non-salient items. Recog-nition of salient items in the high self-esteemcondition was .71 (SD�.28) and recognition ofnon-salient items was .56 (SD�.31). In the lowself-esteem condition recognition of salient itemswas .66 (SD�.27) and recognition of non-salientitems was .49 (SD�.29). A 2 (detail: salient andnon-salient)�2 (self-esteem: high or low) mixedANOVA was conducted and revealed an effect ofdetail, F(1, 98)�14.54, pB.001, h2�.13, but nointeraction between detail and self-esteem, F(1,98)�.06, ns. To determine whether there wereany differences between the two groups inrecall of salient and non-salient details, indepen-dent t-tests were conducted. They revealed no

differences between the two groups in the report-ing of salient, t(98)�.48, ns, and non-salientdetails, t(98)�.47, ns.

GENERAL DISCUSSION

Previous research using interrogative suggestibil-ity has suggested that participants low in self-esteem are more suggestible to leading questionsin the GSS than participants high in self-esteem(Gudjonsson & Lister, 1984; Singh & Gudjonsson,1984). Four experiments are presented, whichexamined whether participants low in self-esteemare more suggestible to post-event suggestion andwhether this suggestibility effect is due to genuinememory impairment or whether it is due toenhanced suggestibility to demand characteristics.In Experiments 1 and 4 the standard recognitiontest was used, which places the original item incompetition with the misleading item. Partici-pants low in self-esteem were found to exhibitmore misinformation effects for salient and non-salient items than participants high in self-esteemwhile participants high in self-esteem did notexhibit misinformation effects at all. In Experi-ment 2 the modified recognition test was used,which places the original item in competition witha novel item. No differences in the levels ofmisinformation effects were found between highand low self-esteem participants. In Experiment 3the modified test was used with a longer retentioninterval to examine whether the modified test wasmasking any genuine memory impairment. Nomisinformation effects were evident in either thelow or high self-esteem conditions.

The findings of Experiments 1 and 4 suggestthat participants low in self-esteem are moresusceptible to post-event suggestion. When theoriginal item is in competition with the misleadingitem participants low in self-esteem are morelikely to choose the misleading item than partici-pants high in self-esteem. This is likely due to thedemand characteristics inherent in the recogni-tion test with participants low in self-esteemmore sensitive to them than participants high inself-esteem.

In Experiment 2 the modified recognition testwas used, which removes the misleading item asa choice on the final test, thereby removing thedemand characteristics. When the misleadingitem was not a choice on the recognition testno differences in the levels of misinformationeffects was found between participants high and

TABLE 4

Reporting of original item on standard recognition test in

Experiment 4

High

misled

Low

misled

High non-

misled

Low non-

misled

Salient .64 .34 .68 .66

(.42) (.35) (.38) (.37)

Non-salient .58 .28 .62 .60

(.34) (.36) (.39) (.38)

Standard deviations included in parentheses.

8 SAUNDERS

Dow

nloa

ded

by [

Jo S

aund

ers]

at 0

3:34

13

Janu

ary

2012

low in self-esteem. Experiment 3 supported thesefindings with no differences in the reporting ofthe new item found between high and low self-esteem participants even when a longer retentioninterval was used to find evidence of memoryimpairment. These findings suggest that partici-pants low in self-esteem are more sensitive todemand characteristics and external suggestionthan participants high in self-esteem. Whendemand characteristics are present participantslow in self-esteem are more susceptible to themthan participants high in self-esteem. Whendemand characteristics are removed from therecognition test, participants low in self-esteemperform as well as participants high in self-esteem. This suggests that the suggestibility effectdemonstrated by the participants low in self-esteem is not due to genuine memory impair-ment but is rather due to sensitivity to demandcharacteristics and external suggestion. The par-ticipants low in self-esteem are likely to lackconfidence in the reliability of their memories.When demand characteristics are present, such asthe possibility that the experimenter wrote thequestions and must know what was in the slides,the participants low in self-esteem are likely toput more stock in the external suggestion than intheir own memories.

Campbell (1990) has suggested that partici-pants low in self-esteem have more poorly ex-pressed notions of who and what they are.Campbell suggests that participants withthese more uncertain self-concepts would bemore susceptible to external stimuli. If this werethe case then this would account for the partici-pants low in self-esteem being more susceptible tomisleading post-event suggestion.

These findings suggest that the police shouldbe particularly careful when interviewing wit-nesses with low self-esteem. Witnesses with lowself-esteem may be sensitive to any post-eventsuggestion contained within questions and may bemore likely to report this post-event suggestionthan witnesses with high self-esteem. This en-hanced suggestibility effect is likely to be due towitnesses agreeing with the interviewer ratherthan due to genuine memory impairment.

In summary, participants low in self-esteem aremore suggestible to post-event suggestion thanparticipants high in self-esteem. However, thissuggestibility effect is due to sensitivity to de-mand characteristics and suggestion rather thandue to genuine memory impairment on the part ofthe participants. This suggests that the police

should be careful when interviewing witnesses

with low self-esteem, as the interviewee may be

particularly sensitive to any misleading informa-

tion contained in questions.

Manuscript received 14 October 2011

Manuscript accepted 19 October 2011

First published online 13 January 2012

REFERENCES

Bekerian, D. A., & Bowers, J. M. (1983). Eyewitnesstestimony: Were we misled? Journal of ExperimentalPsychology: Learning, Memory & Cognition, 9,139�145.

Belli, R. F. (1988). Color blend retrievals: Compromisememories or deliberate compromise responses?Memory & Cognition, 16, 314�326.

Belli, R. F. (1993). Failure of interpolated tests ininducing memory impairment with final modifiedtests: Evidence unfavourable to the blockinghypothesis. American Journal of Psychology, 106,407�427.

Belli, R. F., Windschitl, P. D., McCarthy, T. T., &Winfrey, S. E. (1992). Detecting memory impair-ment with a modified test procedure: Manipulatingretention interval with centrally presented eventitems. Journal of Experimental Psychology: Learn-ing, Memory & Cognition, 18, 356�367.

Bonto, M. A., & Payne, D. G. (1991). Role ofenvironmental context in eyewitness memory.American Journal of Psychology, 104, 117�134.

Campbell, J. D. (1990). Self-esteem and clarity of theself-concept. Journal of Personality and Social Psy-chology, 59, 538�549.

Ceci, S. J., Ross, D. F., & Toglia, M. P. (1987).Suggestibility of children’s memory: Psycholegalimplications. Journal of Experimental Psychology:General, 116, 38�49.

Chandler, C. C. (1989). Specific retroactive interferencein modified recognition tests: Evidence for anunknown cause of interference. Journal of Experi-mental Psychology: Learning, Memory & Cognition,15, 256�265.

Chander, C. C. (1991). How memory for an event isinfluenced by related events: Interference in mod-ified recognition tests. Journal of ExperimentalPsychology: Learning, Memory & Cognition, 13,115�125.

Chandler, C. C., & Gargano, G. J. (1998). Retrievalprocesses that produce interference in modifiedforced-choice recognition tests. Memory & Cogni-tion, 26, 220�31.

Christiaansen, R.E., & Ochalek, K. (1983). Editingmisleading information from memory: Evidence forthe coexistence of original and post-event informa-tion. Memory & Cognition, 11, 467�475.

Gudjonsson, G. H., & Clarke, N. K. (1986). Suggest-ibility in police interrogation: A social psychologicalmodel. Social Behaviour, 1, 83�104.

MISINFORMATION EFFECT 9

Dow

nloa

ded

by [

Jo S

aund

ers]

at 0

3:34

13

Janu

ary

2012

Gudjonsson, G. H., & Lister, S. (1984). Interrogativesuggestibility and its relationship with perceptions ofself-concept and control. Journal of the ForensicScience Society, 24, 99�110.

Gudjonsson, G. H., & Sigurdsson, J. F. (2003). Therelationship of compliance with coping strategiesand self-esteem. European Journal of PsychologicalAssessment, 19, 117�123.

Lindsay, D. S. (1990). Misleading suggestions canimpair eyewitnesses’ ability to remember eventdetails. Journal of Experimental Psychology: Learn-ing, Memory, and Cognition, 16, 1077�1083.

Lindsay, D. S., & Johnson, M. K. (1989a). The reversedeyewitness suggestibility effect. Bulletin of thePsychonomic Society, 27, 111�113.

Lindsay, D. S., & Johnson, M. K. (1989b). The eye-witness suggestibility effect and memory source.Memory and Cognition, 17, 349�358.

Loftus, E. F. (1977). Shifting human color memory.Memory & Cognition, 5, 696�699.

Loftus, E. F., Donders, K., Hoffman, H. G., & Schooler,J. W. (1989). Creating new memories that arequickly accessed and confidently held. Memory &Cognition, 17, 607�616.

Loftus, E. F., Miller, D., & Burns, H. (1978). Semanticintegration of verbal information into a visualmemory. Journal of Experimental Psychology: Hu-man Learning & Memory, 4, 19�31.

McCloskey, M., & Zaragoza, M. (1985). Misleadingpostevent information and memory forevents: Arguments and evidence against memoryimpairment hypotheses. Journal of ExperimentalPsychology: General, 114, 1�16.

Pickel, K. L. (2007). Remembering and identifyingmenacing perpetrators: Exposure to violence andthe weapon focus effect. In R. C. L. Lindsay, D. F.Ross, J. D. Read, & M. P. Toglia (Eds.) The hand-book of eyewitness psychology: Vol. 2. Memory forpeople (pp. 339�360). Mahwah, NJ: Lawrence Erl-baum Associates.

Rosenberg, M. (1965). Society and the adolescent self-image. Princeton, NJ: Princeton University Press.

Saunders, J. (2009). Memory impairment in the weaponfocus effect. Memory & Cognition, 37, 326�335.

Singh, K. K., & Gudjonsson, G. H. (1984). Interrogativesuggestibility, delayed memory and self concept.Personality and Individual Differences, 5, 203�209.

Toglia, M. P., Ross, D. F., Ceci, S. I., & Hembrooke,H. (1992). The suggestibility of children’s memory:A social-psychological and cognitive interpretation.In M. L. Howe, C. J. Brainerd, & V. F. Reyna (Eds.),The development of long-term retention (pp. 217�241). New York, NY: Springer-Verlag.

Wright, D. B., & Stroud, J. N. (1998). Memory qualityand misinformation for peripheral and centralobjects. Legal and Criminological Psychology, 3,273�286.

10 SAUNDERS

Dow

nloa

ded

by [

Jo S

aund

ers]

at 0

3:34

13

Janu

ary

2012