Journals of Gerontology: Psychological Sciences

cite as: J Gerontol B Psychol Sci Soc Sci, 2020, Vol. 75, No. 1, 96–103
doi:10.1093/geronb/gbz057
Advance Access publication May 10, 2019

Special Section: Making Use of Null Age Effects

Misinformation Effect in Aging: A New Light with

Equivalence Testing

Downloaded from https://academic.oup.com/psychsocgerontology/article/75/1/96/5487940 by guest on 29 February 2024

Marine Tessoulin, PhD,1,* Jean-Michel Galharret, MS,2 Anne-Laure Gilet, PhD,1 and
Fabienne Colombel, PhD1
1
Univ Nantes, Univ Angers, Laboratoire de Psychologie des Pays de la Loire (LPPL - EA 4638), Nantes, France. 2CNRS,
Laboratoire de Mathématiques Jean Leray (LMJL- UMR 6629), Nantes, France.
*Address correspondence to: Marine Tessoulin, PhD, Université de Nantes, Laboratoire de Psychologie des Pays de la Loire (LPPL 4638), Chemin
de la Censive du Tertre, BP 81227, 44312 Nantes Cedex 3, France. E-mail: [email protected]

Received: December 22, 2018; Editorial Decision Date: April 26, 2019

Decision Editor: Angela Gutchess, PhD

Abstract
Objectives: To better characterize the formation of false memories in older adults, we conducted a study using a French
adaptation of the misinformation paradigm from Loftus, Levidow, and Duensing (1992). We aimed to show higher false
memory production in older than in younger adults.
Method: One hundred and four younger adults (18–30 years) and 104 older adults (70–95 years) took part in the study.
Participants were presented with a misinformation paradigm through the viewing of a short video followed by a question-
naire containing misinformation about the film. After a short delay (45 min), they performed a recognition task.
Results: Contrary to our hypothesis, the results analyzed with a Welch t test did not reveal a greater misinformation effect
in older adults than in younger adults. Results were reanalyzed using the equivalence test which indicated that younger and
older adults are statistically equivalent and not statistically different.
Discussion: The equivalence test helped to clarify the contradictory results of the literature. Furthermore, such results show
the interest to reconsider misinformation effect in aging.
Keywords: Aging, Equivalence test, False memories, Misinformation effect
  

Our frenetic digital world makes it possible to disseminate widely than young adults. So, are older people more recep-
a multitude of information very quickly via a wide variety tive to misleading information? In light of this type of study,
of social networks, with an ever-increasing amount of er- one may ask what kind of memory trace of misleading in-
roneous information. This fake news or alternative facts formation is retained and whether older people are more
abound on the web and have become a routine phenom- inclined to integrate misleading information.
enon. For example, during the 3 months preceding the To memorize is to reconstruct reality rather than re-
U.S. presidential election in 2016, an analysis by Buzzfeed member information veridically. A new memory is a cre-
(Silverman, 2016) reported that the 20 most popular fake ation that implies reality but also our beliefs, knowledge
news items were shared 87,11,000 times while the 20 most about the world, and old memories. Thus, memories are
important political facts were shared 73,67,000 times. In a vulnerable to distortions (e.g., Dehon, 2012; Schacter,
recent study, Guess, Nagler, and Tucker (2019) investigated Guerin, & St Jacques, 2011). Distortions are a normal phe-
Facebook sharing of fake news and examined the charac- nomenon that can occur in adults and are amplified in older
teristics of those who shared it. They observed those over adults (e.g., Lövdén, 2003; Schacter, Koutstaal, & Norman,
65 years disseminated these alternative facts much more 1997). Studying distortions consists of considering the

© The Author(s) 2019. Published by Oxford University Press on behalf of The Gerontological Society of America. All rights reserved. 96
For permissions, please e-mail: [email protected].
Journals of Gerontology: PSYCHOLOGICAL SCIENCES, 2020, Vol. 75, No. 1 97

adequacy between the original event and the memory of a recognition test after a variable delay. Today, the score
the event. One way to study the malleability of memory is calculation method is based on errors at the recognition
to investigate false memories. The term “false memories” test, considering both the number of errors after receiving
refers to wrongly but sincerely remembering events distant neutral and misleading information. A misinformation ef-
from reality (Roediger & McDermott, 1995). Thus, false fect is obtained when the errors are more numerous after
memories contain some distortion of real events. receiving misleading information than after receiving neu-
Two main paradigms are used to study the produc- tral information (Roediger & Geraci, 2007). Thus, the mis-
tion of false memories in a laboratory setting. On the one information effect is considered as a type of false memories
hand, in the Deese-Roediger-McDermott paradigm (DRM; obtained under the specific conditions of this paradigm.
Deese, 1959; Roediger & McDermott, 1995), participants Several explanations have been developed to account
study lists of words (e.g., slipper, sock, heel) semantically for the misinformation effect observed in younger adults

Downloaded from https://academic.oup.com/psychsocgerontology/article/75/1/96/5487940 by guest on 29 February 2024

associated with one nonpresented word (shoe), called the (e.g., Volz, Stark, Vaitl, & Ambach, 2019), and are helpful
critical lure. When participants later perform a recall and/ in understanding the performances of older adults. The
or a recognition task, they often falsely recall or recognize first explanation is from the activation-monitoring theory
the critical lure, claiming it was in the studied word lists. (Roediger, Balota, & Watson, 2001) suggesting that the
When it comes to aging and false memories research, results production of false memories occurs via an activation of
obtained with the DRM paradigm show that aging is as- encoding traces associated with a failure in the monitoring
sociated with an increased production of false memories process which is responsible for identifying the acqui-
(e.g., Dehon, 2006; Norman & Schacter, 1997; Thomas & sition source of the information at the time of retrieval.
Sommers, 2005). This proposition fits well with our knowledge about cog-
On the other hand, false memories may also be studied nitive processes in aging. Many studies show that source
using the misinformation paradigm created by Loftus, monitoring (e.g., Johnson, Hashtroudi, & Lindsay, 1993),
Miller, and Burns (1978) that investigates the influence of that is, context of information (place, moment, modality
misleading information on memory (see Loftus, 2005, for a learning) is impaired in older adults (e.g., Dehon & Brédart,
review). The misinformation paradigm, used in the present 2004). In the misinformation paradigm, during the recog-
study, has the advantage of being closer to a real situation in nition test, older adults could recollect misleading informa-
which we witness a particular scene, than other paradigms tion but fail to recollect and monitor the source of this
such as the DRM. In a more recent version of their par- information. Thus, it may be difficult to reject misleading
adigm (Loftus, Levidow, & Duensing, 1992), participants information which does not come from the original event.
first watched a short film clip (1 min 25 s) depicting a po- Roediger and Geraci (2007) explored the effects of several
litical rally in which the main character is threatened by variables in younger (Mage = 19.45, SD = 1.43) and older
opponents. A blue car tries to run rushes him over, and (Mage = 75.00, SD = 6.30) adults on the misinformation
the main character is struck with a club. After viewing the effect such as the number of presentations of misleading in-
film clip, participants answered 10 questions about the formation (1 or 3) and the type of test (recognition test or
film with “yes” or “no” responses. Some participants were source-memory test). They showed that when misleading
exposed to misinformation while others were not. For ex- information is presented only once and is followed by a
ample, a misinformation question was “Did you see the recognition test, a greater misinformation effect is observed
two thugs in black shirts who threatened the victim just in older adults than in younger adults. However, the misin-
before the white vehicle tried to run him over?” Finally, formation effect is reduced in aging when participants re-
authors studied whether participants accepted misleading ceived a source-memory test, suggesting that older adults
information as original information, and their capacity to can reduce their production of false memories when they
recollect original information about the film. The authors are encouraged to focus on the source of information.
considered the correct answers for neutral questions about Difficulties in the source monitoring processes may thus
the film and misleading questions, comparing the accu- contribute to greater susceptibility in the misinformation
racy of participants who were not exposed to misleading effect observed in older adults. Note that their production
information and participants who were exposed to mis- of false memory remains higher than those of young adults
information. Memory of the original event is negatively in all the cases.
affected when participants incorrectly report misleading A second explanation refers to the fuzzy trace theory
postevent information in the recognition test (Loftus, (e.g., Brainerd & Reyna, 2002; Reyna & Brainerd, 1995).
2005). Whatever the variants which succeeded to the par- According to this theory, two types of mental representations
adigm (e.g., Frost, 2000; Huff & Umanath, 2018), in a emerge when events are encoded: a global representation,
typical misinformation experiment, participants first study helpful to preserve a gist memory trace (i.e., general con-
an event; they watch a film or slides that relate a story. tent of information), and a specific representation, useful to
In the second part, they are presented with neutral or preserve a verbatim memory trace (i.e., specific features of
misleading postevent information in the form of a narra- information) but which fades faster than the gist memory
tive or questions. In the final part, participants complete trace. In the misinformation paradigm, the original event
98 Journals of Gerontology: PSYCHOLOGICAL SCIENCES, 2020, Vol. 75, No. 1

and the misleading information share a global meaning. postevent questions or neither. One unexpected finding
Verbatim makes the original information unique and dif- was that older adults were less susceptible than younger
ferent from the misleading information. Both gist and adults to the misinformation effect. The authors assumed
verbatim retrieval of misleading information support the that the final memory test (a source monitoring test) could
production of false recognition, while verbatim retrieval explain the greater misinformation effect in younger adults.
of original information rules out false recognition. It is ac- Indeed, the source-recognition test encourages participants
knowledged that older adults tend to base their answers on to examine specific details and promotes source informa-
global representation of events (Brainerd & Reyna, 2015). tion identification in older adults. According to the authors,
Without access to the verbatim, it is very difficult to distin- without this type of test, older adults’ answers would be
guish the original event from the misleading event. If older based on familiarity and misleading information would
adults have no access to the verbatim of the original event, be as familiar as the original information. Additionally,

Downloaded from https://academic.oup.com/psychsocgerontology/article/75/1/96/5487940 by guest on 29 February 2024

during the recognition test they cannot select the correct the results show a negative relationship between confi-
information and thereby avoid misleading information. dence and memory performance which differed with age,
For example, Marche, Jordan, and Owre (2002) studied with greater confidence related to more errors for young
the effects of original information learning on the misinfor- adults, which may also partly explain the results obtained.
mation effect in younger adults (Mage ≈ 23) and older adults Nevertheless, Wylie et al. (2014), in their meta-analysis,
(Mage ≈ 73). They presented the original information once showed that most studies highlighted a greater misinforma-
(one-trial condition) or as many times as necessary to learn tion effect for older adults compared with younger adults
it (learning condition). Three weeks later and immediately (e.g., Hess, Popham, Emery, & Elliott, 2012; Holliday et al.,
prior to final testing, participants were asked questions 2011; Roediger & Geraci, 2007 [Experiment 1]).
that contained misleading information. In the learning con- In the present study, our main objective was to inves-
dition (verbatim memory for original and for misleading tigate the misinformation effect in older adults. Given the
events is poorer in older than in younger adults), older large discrepancy in the studies, we decided to simplify
adults are more susceptible to misinformation effect than the misinformation paradigm as much as possible. Our
younger adults because the latter are less willing to accept methodology was in line with that used by Roediger and
the verbatim representations of misinformation. Younger Geraci (2007) as we adopted their rigorous and demanding
adults experienced feelings of contrast when forced to criteria, considering the number of errors after receiving
choose between original memory and memory for mis- neutral and misleading information. Based on the meta-
leading information, a verbatim memory trace being avail- analysis of Wylie et al. (2014), we hypothesized a greater
able for both information types in younger adults. Marche misinformation effect in older adults compared to younger
et al. (2002) concluded that misinformation errors depend adults.
on the degree of initial learning.
The literature on false memories in aging is less con-
sensual with the misinformation paradigm than with the Method
DRM paradigm. Several studies have not found significant
age differences (e.g., Dodson & Krueger, 2006) or reverse Participants
effects where younger adults are more susceptible to mis- One hundred and four younger adults (18–30 years,
information than older adults (e.g., Auslander, Thomas, M = 23.50, SD = 3.42, 64 females) and 104 older re-
& Gutchess, 2017; Gabbert, Memon, Allan, & Wright, tired adults (70–95 years, M = 76.09, SD = 5.30, 74
2004; Marche et al., 2002, one-trial condition). For ex- females) were recruited for the experiment. Sample size
ample, in the Marche et al. study, when original informa- was chosen to be large enough to find between small
tion is presented once (original verbatim memory was also and moderate effects (G*Power calculation: d = 0.35,
poorer for older than younger adults), misinformation α = .05, power = .8, yields n1 = n2 = 102 and critical value
effect is greater in younger adults, possibly because they of t = 1.65 (Faul, Erdfelder, Lang, & Buchner, 2007)).
have better verbatim memory for misleading information The data-stopping rule involved continuing until the
which is presented closer in time to the memory test. Their full sample size was obtained for our main analysis. All
results highlight the discrepancy in the results obtained participants were recruited from associations and clubs
from studies on the misinformation effect depending on in Nantes (France). They were all native French speakers.
the encoding conditions. Recently, Auslander et al. (2017) They received no compensation for their participation.
explored the misinformation effect in older (62–86 years, All participants gave their informed consent prior to their
Mage = 74.40, SD = 6.87) and younger (18–26 years, inclusion in the study. All older participants obtained a
Mage = 19.79, SD = 2.00) adults who were asked, imme- score equal to or higher than 27 (M = 28.82, SD = 1.11) on
diately after viewing a film, to complete a postevent ques- the French version of the Mini-Mental State Examination
tionnaire that included misinformation. Participants had (MMSE; Derouesné et al., 1999), confirming the integ-
to determine whether the information was present; in the rity of their global cognitive functioning. The French ver-
video, in the postevent questions, both the video and the sion of the Geriatric Depression Scale (GDS; Bourque,
Journals of Gerontology: PSYCHOLOGICAL SCIENCES, 2020, Vol. 75, No. 1 99

Blanchard, & Vézina, 1990) and the French version of the participants performed the final recognition test. To con-
Beck Depression Inventory (BDI-II, Beck, Steer, & Brown, clude, they were thanked for their participation and fully
1998) revealed no signs of depression in the older adults debriefed.
(M = 6.69, SD = 5.61) and in the younger adults (M = 6.93,
SD = 6.39). The two groups were matched on Education
(t (206) = 1.86, p = .062, Cohen’s d = 0.258), that is, years Data Analysis
of education from the age of 6 (Myears = 13.72, SD = 2.22 A Welch’s t test was conducted to analyze correct
in the younger adults, Myears = 12.84, SD = 4.28 in the recognitions. Next, an equivalence test was run to statis-
older group). tically prove an absence of age-related difference. Proving
a significant difference between two groups A and B is a

Downloaded from https://academic.oup.com/psychsocgerontology/article/75/1/96/5487940 by guest on 29 February 2024

well-known problem. Let us consider that A and B have
Material
respective Gaussian distributions, N (µA, σA) and N (µB, σB).
A French version of the misinformation paradigm (Mahé, In this context, we test the null hypothesis that the means
Corson, Verrier, & Payoux, 2015) was proposed. The µA and µB are equal or equally so that µ = 0 where µ = µA
first part of the paradigm consisted in presenting a short µB. A Student’s t test was used when σA = σB, otherwise
-
film (6 min) taken from the film “Z” (Perrin, Rachedi, & Welch t test was used. Rejecting the null hypothesis allows
Gavras, 1969) on a computer screen. Ten minutes later, the researcher to claim that there is a significant difference
participants were presented orally with a questionnaire in between A and B with a statistical size α set by default at
which false information about the film was introduced: .05. Nevertheless, not rejecting the null hypothesis does not
four questions contained neutral information, four prove that groups A and B are equivalent. The null and
questions suggested information present in the film, and the alternative hypotheses have an asymmetric rule in sta-
four questions suggested misleading information. Finally, tistical hypothesis testing. So, to prove the equivalence of
after a delay of 40 min, a recognition test was proposed. the two groups, it must be stated as an alternative hypoth-
For the 12 questions, the participants had to choose the esis in the testing process. Without any further information
correct answer between two responses: (a) a response cor- on µ, it is not possible to provide a relevant test statistic
responding to the information watched in the film (the cor- whose distribution would be known under the null hypoth-
rect answer), or (b) a response either corresponding to the esis. Schuirmann (1987) defined the equivalence of A and B
misinformation information suggested in the questionnaire when µ is in “an indifference zone” around 0, denoted by
presented after the film or to new, plausible information. [d1, d2], d1 < d2 and used the TOST (Two One-Sided Tests)
For example, in one of the scenes, participants watched a procedure to test the equivalence. The null hypothesis of
man accost the main character (part one). Then, they had the TOST is H0: µ > d2 or µ < d1, and the alternative is H1:
to answer the following question (part two) if they were d1 ≤ µ ≤ d2. The first step is to compute the p-values p1 and
presented with neutral information: “The main character p2 of the two one-sided tests corresponding, respectively, to
is accosted by someone. What color is the person’s jacket?” H01: µ < d1 against H11: µ ≥ d1 and H02: µ > d2 against H12: µ
or the question: “The main character is accosted by a ≤ d2. The TOST p value is then p = max (p1, p2). This pro-
woman. What color is the woman’s jacket?” if they were cedure ensures that the TOST has a significant level α (for
presented with misleading information. Finally, during the details, see Wellek, 2010).
recognition test, they had to answer the question: “Who The equivalence test and the Student’s or Welch’s t tests
accosts the main character?” by choosing between two focus on assumptions that are not entirely reciprocal; con-
answers: “a man” or “a woman”. They had to answer to sequently, it is legitimate to be interested in the consistency
all the questions in order. of their conclusions. Table 1 summarizes the four possible
conclusions described in Lakens, Scheel, and Isager (2018).
The conclusions of the tests do not contradict each other
Procedure when the two groups are Statistically Equivalent and Not
Participants were tested individually in a single session Statistically Different or when the two groups are Not
that lasted about 1 hr and 30 min. They were aware that Statistically Equivalent and Statistically Different. The two
they had agreed to participate in a study on memory. After groups can also be Statistically Equivalent and Statistically
obtaining informed consent, participants were invited to Different. In this case, the parameter μ can be close to 0 (so
watch the film. Then, older adults completed the MMSE. in the interval [d1, d2]), but in the same time if the sample
Meanwhile, younger adults had to copy a drawing to keep size N is large enough, the Student’s t test can conclude that
active. Ten minutes after watching the film, all participants μ is statistically different to 0. When the two groups are
were presented orally with the misinformation question- Not Statistically Equivalent and Not Statistically Different,
naire about the film and had to write down their response we cannot conclude. This particular case demonstrates the
to each question. After this part, information about ed- nonreciprocity of the two tests.
ucation and mood was collected. Finally, 45 min after Lakens (2017) notes that large sample size is required to
the introduction of the misinformation questionnaire, get an acceptable power (commonly more than .80) when
100 Journals of Gerontology: PSYCHOLOGICAL SCIENCES, 2020, Vol. 75, No. 1

Table 1. Illustration of the Terminology Used by Lakens, Scheel, and Isager (2018)

Conclusion of the Student’s or Welch’s t tests

Reject h0 Not reject h0

Conclusion of the Reject H0 Statistically Equivalent and Statistically Statistically Equivalent and Not
Equivalence test Different (SE and SD) Statistically Different (SE ans NSD)
Not reject H0 Not Statistically Equivalent and Not Statistically Equivalent and Not
Statistically Different (NSE and SD) Statistically Different (NSE and NSD)

Note: h0 denotes the null hypothesis in the Student’s or Welch’s t test (i.e. h0: μ = 0 against h1: μ ≠ 0) and H0 denotes the null hypothesis in the equivalence test (i.e.,

Downloaded from https://academic.oup.com/psychsocgerontology/article/75/1/96/5487940 by guest on 29 February 2024

H0: µ > d2 or µ < d1, and the alternative is H1: d1 ≤ µ ≤ d2) where d1, µ, d2 have been defined above.

d1 or d2 is close to 0. All the analyses were done with R To improve inference about the absence of an age-
software (R Core Team, 2017). related effect on the misinformation effect, we reanalyzed
our data using an equivalence test (Lakens, 2017; Lakens,
McLatchie, Isager, Scheel, & Dienes, 2018; Lakens, Scheel,
Results & Isager, 2018). We chose two previous studies allowing us
to define the smallest and the largest size of the misinforma-
Correct Recognitions tion effect: Auslander et al. (2017) highlighted a significant
First, the accuracy for the neutral questions was examined. misinformation effect for younger adults over older adults,
The correct responses provided for neutral questions at the whereas Roediger and Geraci (2007) showed a greater mis-
recognition test reveal the baseline of participants’ abilities information effect in older adults. Let µ = µolder – µyounger the
at an episodic memory task. There is no statistical differ- misinformation difference mean between older adults and
ence between the older (M = 3.35, SD = 0.77) and younger younger adults. Based on the Auslander and colleagues’
adults correct answers (M = 3.46, SD = 0.72) to neutral study, we can consider that the smallest size of interest is
questions, Welch t test showed t(206) = 1.11, p = .26, d1 = −0.431: they indicated misinformation effect with 50
d = 0.147. Unfortunately, we could not apply the equiva- younger adults and 37 older adult samples. Similarly, based
lence test on these data since neither Roediger and Geraci on the Roediger and Geraci study, we can set the biggest
(2007) nor Auslander et al. (2017) provided a suitable his- size of interest d2 = 0.581 (they built this result on 24 for
toric Misinformation effect. In order to characterize the both sample sizes). The selected sample sizes (n1 = n2 = 104)
misinformation effect, we followed the method proposed should have been large enough to detect this smallest effect
by Roediger and Geraci (2007), which considered the errors size with a t test.
at the recognition test for the neutral and the misleading For the equivalence test, we consider H0: μ < d1 (−0.43)
condition. The rate of errors for the neutral condition or μ > d2 (0.58) against H1: d1 ≤ μ ≤ d2. The first one-sided
may indicate a failure to retrieve the original information test indicates that we can reject effects as small as or smaller
(watched in the film) in memory and serves as a baseline than d1 = −0.43, t(205.99) = 1.65, p = .0498. Furthermore,
for the difficulties in retrieving the original information. It the second one-sided test shows we can reject effects as
is essential to know in what extent the participants have large as or larger than d2 = 0.58, t(206) = −5.64, p < .001.
a memory of the original event (Marche et al., 2002) be- These results indicate that younger and older adults are sta-
fore studying the integration of misleading information in tistically equivalent and not statistically different regarding
memory. In our opinion, this method is based on the most their susceptibility to misinformation.
rigorous and demanding criteria. To test for an age-related
difference in the misinformation effect, we thus computed
the score by subtracting errors in the neutral condition Discussion
from those in the misinformation condition, for younger The aim of the current study was to examine the misin-
and older adults separately. formation effect in aging. First, we examined the accuracy
A Welch t test was then conducted to test our hypothesis of memories for neutral questions in younger and older
of a larger misinformation effect in older than in younger adults to have a baseline of participant’s capacity at an ep-
adults. Contrary to our prediction, our results do not in- isodic memory test. We found no statistical difference in
dicate a significant misinformation effect for older adults correct recognitions between younger and older adults.
(M = 0.88, SD = 1.29) over younger adults (M = 1.14, Our results do not match those obtained by other authors
SD = 1.28), t(206) = −1.46, p = .146, d = −0.20. Thus, in sta- (Karpel, Hoyer, & Toglia, 2001; Roediger & Geraci, 2007)
tistical terms, according the Neyman–Pearson principle, the who found a decline in correct recognitions in older adults.
observed data provide no strong evidence against the null However, our results are in line with those obtained by
hypothesis (i.e., no misinformation effect); in other words, some authors who found that older adults were as good as
we have insufficient evidence to reject the null hypothesis. younger adults at recognizing an event (Cohen & Faulkner,
Journals of Gerontology: PSYCHOLOGICAL SCIENCES, 2020, Vol. 75, No. 1 101

1989; Loftus et al., 1992). These results are not surprising well as younger adults on the correct answers for neutral
given that participants performed a recognition task. They questions) 1 hr after viewing the film. Therefore, we suggest
had to choose between two answers provided in the test. that they have a strong memory trace of the misleading in-
This type of task involves decision-making processes to se- formation (which was presented 45 min before the recog-
lect the right answer and reject the wrong answer. This is a nition test) too. With reference to the fuzzy trace theory, it
less-demanding task than a recall task in which participants is thus possible that the verbatim trace of the original and
must use retrieval strategies to produce a response. As the misleading information was as strong in younger adults
mentioned by Fraundorf, Hourihan, Peters, and Benjamin as in older adults. Hence, younger and older adults could
(2019), the age-related difference in recognition tasks is be confronted with the same difficulty during the recogni-
unclear in the literature, while it is well known that older tion test; they must choose between two pieces of informa-
adults are weaker than younger adults on a recall task. In tion that they have encoded and for which they have an

Downloaded from https://academic.oup.com/psychsocgerontology/article/75/1/96/5487940 by guest on 29 February 2024

their meta-analysis, the authors show decreased perfor- available verbatim trace: the information in the film and
mance in recognition tasks in older adults, but note the het- the misleading information. Consequently, during the rec-
erogeneity of performances, depending on multiple factors. ognition task, older adults may have used their verbatim
In addition, our older participants were active seniors who trace of the original event that prevented them accepting
demonstrated a high level of general cognitive ability (high misleading information. Our results are compatible with
MMSE scores and higher levels of education). Finally, keep this theoretical account and cannot be attributed to a
in mind that in the case of the misinformation paradigm, general failure in memory ability which may contribute to
high memory performance on neutral questions is required an increased misinformation effect in aging (Karpel et al.,
to study the misinformation effect. If the original informa- 2001). With reference to the activation-monitoring theory,
tion is not well encoded, it becomes less relevant to study in this study, older and younger participants may have
the effect of exposure to misleading information on the succeeded in applying the monitoring process. Thus, both
original information recognition. Obtaining high scores older and younger adults were able to identify the origin
to neutral questions in the two populations ensured that of the information and to correctly reject the misleading
scores on misleading questions are not due to the strength information not in the original event.
of initial encoding of information. One cannot also rule out the possibility that research
Second, we studied the recognition errors committed suffers from a limited publication of null results (Campbell
by younger adults and older adults after receiving neutral & Gustafson, 2018). The greater attraction towards re-
and misleading information. Contrary to our hypothesis, porting positive results together with the common disin-
a greater misinformation effect was not obtained in older terest in nonsignificant findings may also cause publication
adults compared to younger adults with a Welch t test. bias (e.g., Franco, Malhotra, & Simonovits, 2014; Giner-
Analyses conducted with the equivalence test showed that Sorolla, 2012; Reysen, 2006) resulting in a loss of immense
younger and older adults are statistically equivalent and not resources. Moreover, nonsignificant findings are too often
statistically different regarding their susceptibility to misin- misinterpreted in research; nonsignificant p value being
formation. This result is consistent with some other studies regularly considered as a support for the null hypothesis.
investigating the misinformation effect in aging (Auslander However, nonsignificant p value indicates that the test
et al., 2017; Gabbert, et al., 2004; Marche et al., 2002). cannot reject the null hypothesis, because of an absence of
Auslander et al. (2017), who obtained a greater mis- effect or because of an insufficient power of the test to de-
information effect in younger adults, suggested that the tect an effect. These new tests soon to be widespread fol-
type of test (a source-monitoring test) could have made lowing Lakens’s recommendations (Lakens, McLatchie,
the recognition of original information easier. Indeed, this Isager, Scheel, & Dienes, 2018) enable researchers to test for
type of test could help to discern whether the information the presence/absence of null effects. This approach will thus
was encoded during the film viewing (Phase 1) or during facilitate new studies regarding age-related effects particu-
exposure to the misleading information (Phase 2). In our larly when the literature yields mixed results. By allowing
study, participants did not receive a source-memory test, researchers to test for the absence of an age-related effect,
but it is possible that older adults identified the origin of equivalence tests as well as a Bayesian approach may con-
information as successfully as younger adults. According sequently serve the publication of new meta-analyses that
to Marche et al. (2002), who found a greater misinfor- include all studies and not only those reporting significant
mation effect in younger adults, it is the degree of initial age-related differences. Further research using only these
learning which influences the differences in misinformation new statistics are thus needed to explore the conditions of
effect in younger and older adults. Older adults are more appearance of the misinformation effect in aging in order
or less likely to report misleading information compared to get a wider, more complex and accurate picture of the
to younger adults depending of their memory trace of phenomenon by improving the inferences about null effects
the original event and the misleading information. In our as well as the reliability of the findings (Lakens, McLatchie,
study, older adults were as good as younger adults in Isager, Scheel, & Dienes, 2018; Lakens, Scheel, & Isager,
recognizing the original information (they performed as 2018). More generally, equivalence tests as well as a
102 Journals of Gerontology: PSYCHOLOGICAL SCIENCES, 2020, Vol. 75, No. 1

Bayesian approach could become valuable new tools for Dehon, H. (2006). Variations in processing resources and resistance
psychologists to better understand cognition in adulthood to false memories in younger and older adults. Memory, 14,
and aging. 692–711. doi:10.1080/09658210600648456
Finally, this study shows that older adults are not dramat- Dehon, H. (2012). Illusory recollection: The compelling subjec-
ically impaired on certain cognitive tasks. Although aging tive remembrance of things that never happened. Insights
is accompanied by various increasing cognitive difficulties, from the DRM Paradigm. Psychologica Belgica, 52, 121–149.
doi:10.5334/pb-52-2-3-121
it is important to consider the processes that are preserved.
Dehon, H., & Brédart, S. (2004). False memories: Young and older
Thus, under the conditions of our study, there is no evi-
adults think of semantic associates at the same rate, but young
dence to suggest that the elderly adults are more suggestible
adults are more successful at source monitoring. Psychology and
to or influenced by false information than younger adults.
Aging, 19, 191–197. doi:10.1037/0882-7974.19.1.191
These results have a practical application, helping to define

Downloaded from https://academic.oup.com/psychsocgerontology/article/75/1/96/5487940 by guest on 29 February 2024

Derouesné, C., Poitreneau, J., Hugonot, L., Kalafat, M., Dubois, B., &
the conditions under which older adults are as efficient as
Laurent, B. (1999). Le Mini-Mental State Examination (MMSE):
younger adults and question the credibility of an elderly eye- Un outil pratique pour l’évaluation de l’état cognitif des patients
witness to a crime or accident. While recommendations are par le clinicien. Version française consensuelle [The Mini-Mental
published to ensure testimonies are as accurate as possible State Examination (MMSE): A practical tool to evaluate the
(Schuster, 2007) in adults, it would be appropriate to design cognitive state of patients for the clinician. Consensual French
conditions under which the testimonies of elderly people, free Version]. La Presse Médicale, 28, 1141–1148.
of any pathological process, could be made more reliable. Dodson, C. S., & Krueger, L. E. (2006). I misremember it well: Why
older adults are unreliable witnesses. Psychonomic Bulletin &
Review, 13, 770–775. doi:10.3758/bf03193995
Funding Faul, F., Erdfelder, E., Lang, A. G., & Buchner, A. (2007). G*Power
No funding. 3: A flexible statistical power analysis program for the social, be-
havioral, and biomedical sciences. Behavior Research Methods,
39, 175–191. doi:10.3758/BF03193146
Conflict of Interest Franco, A., Malhotra, N., & Simonovits, G. (2014). Publication
bias in the social sciences: Unlocking the file drawer. Science,
None reported.
345(6203), 1502–1505. doi:10.1126/science.1255484
Fraundorf, S. H., Hourihan, K. L., Peters, R. A., & Benjamin, A. S.
(2019). Aging and recognition memory: A meta-analysis.
References Psychological Bulletin. 145, 339–371. doi:10.1037/bul0000185
Auslander, M. V., Thomas, A. K., & Gutchess, A. H. (2017). Frost, P. (2000). The quality of false memory over time: Is
Confidence moderates the role of control beliefs in the con- memory for misinformation “remembered” or “known”?
text of age-related changes in misinformation susceptibility. Psychonomic Bulletin & Review, 7, 531–536. doi:10.3758/
Experimental Aging Research, 43, 305–322. doi:10.1080/0361 bf03214367
073X.2017.1298960 Gabbert, F., Memon, A., Allan, K., & Wright, D. B. (2004). Say it
Beck, A. T., Steer, R. A., & Brown, G. (1998). Beck Depression to my face: Examining the effects of socially encountered mis-
Inventory–II [manual]. Paris: ECPA. information. Legal and Criminological Psychology, 9, 215–227.
Bourque, P., Blanchard, L., & Vézina, J. (1990). Étude psychométrique doi10.1348/1355325041719428
de l’échelle de dépression gériatrique [Psychometric study of the Giner-Sorolla, R. (2012). Science or art? How aesthetic standards
Geriatric Depression Scale]. Revue Canadienne du Vieillissement/ grease the way through the publication bottleneck but under-
Canadian Journal on Aging, 9, 348–355. doi:10.1017/s07 mine science. Perspectives on Psychological Science, 7, 562–571.
14980800007467 doi:10.1177/1745691612457576
Brainerd, C. J., & Reyna, V. F. (2002). Fuzzy-trace theory and false Guess, A., Nagler, J., & Tucker, J. (2019). Less than you think:
memory. Current Directions in Psychological Science, 11, 164– Prevalence and predictors of fake news dissemination on
169. doi:10.1111/1467-8721.00192 Facebook. Sciences Advances, 5, aau4586. doi:10.1126/sciadv.
Brainerd, C. J., & Reyna, V. F. (2015). Fuzzy-trace theory and aau4586
lifespan cognitive development. Developmental Review, 38, Hess, T. M., Popham, L. E., Emery, L. & Elliott, T. (2012). Mood,
89–121. doi:10.1016/j.dr.2015.07.006 motivation and misinformation: Aging and affective state
Campbell, H., & Gustafson, P. (2018). Conditional equivalence influences on memory. Neuropsychology, Development and
testing: An alternative remedy for publication bias. PLoS ONE, Cognition: Section B, Aging Neuropsychology and Cognition,
13, e0195145. doi:10.1371/journal.pone.0195145 19, 13–34. doi:10.1080/13825585.2011.622740
Cohen, G., & Faulkner, D. (1989). Age differences in source forgetting: Holliday, R. E., Humphries, J. E., Milne, R., Memon, A., Houlder, L.,
Effects on reality monitoring and one eyewitness testimony. Lyons, A., & Bull, R. (2011). Reducing misinformation effects
Psychology and Aging, 4, 10–17. doi:10.1037//0882-7974.4.1.10 in older adults with cognitive interview mnemonics. Psychology
Deese, J. (1959). On the prediction of occurrence of particular and Aging, 27, 1191–1203. doi:10.1037/a0022031
verbal intrusions in immediate recall. Journal of Experimental Huff, M. J., & Umanath, S. (2018). Evaluating suggestibility to ad-
Psychology, 58, 17–22. doi:10.1037/h0046671 ditive and contradictory misinformation following explicit error
Journals of Gerontology: PSYCHOLOGICAL SCIENCES, 2020, Vol. 75, No. 1 103

detection in younger and older adults. Journal of Experimental Reyna, V. F., & Brainerd, R. C. J. (1995). Fuzzy-trace theory: An
Psychology: Applied, 24, 180–195. doi:10.1037/xap0000138 interim synthesis. Learning and Individual Differences, 7, 1–75.
Johnson, M.-K., Hashtroudi, S., & Lindsay, D. S. (1993). doi:10.1016/1041-6080(95)90031-4
Source monitoring. Psychological Bulletin, 114, 3–28. Reysen, S. (2006). Publication of nonsignificant results: A survey of
doi:10.1037/0033-2909.114.1.3 psychologists’ opinions. Psychological Reports, 98, 169–175.
Karpel, M. E., Hoyer, W. J., & Toglia, M. P. (2001). Accuracy and qual- doi:10.2466/pr0.98.1.169-175
ities of real and suggested memories: Nonspecific age differences. Roediger, H. L., & Geraci, L. (2007). Aging and the misinformation
The Journals of Gerontology, Series B: Psychological Sciences and effect: A neuropsychological analysis. Journal of Experimental
Social Sciences, 56, 103–110. doi:10.1093/geronb/56.2.P103 Psychology: Learning, Memory, and Cognition, 33, 321–334.
Lakens, D. (2017). Equivalence tests: A practical primer for t-tests, doi:10.1037/0278-7393.33.2.321
correlations, and meta-analyses. Social Psychological and Roediger, H. L., Balota, D. A., & Watson, J. M. (2001). Spreading

Downloaded from https://academic.oup.com/psychsocgerontology/article/75/1/96/5487940 by guest on 29 February 2024

Personality Science, 8, 355–362. doi:10.1177/1948550617697177 activation and arousal of false memories. In H. L. Roediger,
Lakens, D., McLatchie, N., Isager, P. M., Scheel, A. M., & Dienes, Z. J. S. Nairne, I. Neath, & A. M. Surprenant (Eds.), The nature
(2018). Improving inferences about null effects with Bayes of remembering: Essays in honor of Robert G. Crowder (pp.
factors and Equivalence tests. The Journals of Gerontology, 95–115). Washington, DC: American Psychological Association.
Series B: Psychological Sciences and Social Sciences, 75, 45–57. doi:10.1037/10394-006
doi:10.1093/geronb/gby065 Roediger, H. L., & McDermott, K. B. (1995). Creating false
Lakens, D., Scheel, A. M., & Isager, P. M. (2018). Equivalence memories: Remembering words not presented in lists. Journal
testing for psychological research: A tutorial. Advances in of Experimental Psychology: Learning, Memory, and Cognition,
Methods and Practices in Psychological Science, 1, 259–269. 21, 803–814. doi:10.1037/0278-7393.21.4.803
doi:10.1177/2515245918770963 Schacter, D. L., Guerin, S. A., & St Jacques, P. L. (2011). Memory dis-
Loftus, E. F. (2005). Planting misinformation in the human mind: tortion: An adaptive perspective. Trends in Cognitive Sciences,
A 30-year investigation of the malleability of memory. Learning 15, 467–474. doi:10.1016/j.tics.2011.08.004
and Memory, 12, 361–366. doi:10.1101/lm.94705 Schacter, D. L., Koutstaal, W., & Norman, K. A. (1997). False
Loftus, E. F., Levidow, B., & Duensing, S. (1992). Who remembers memories and aging. Trends in Cognitive Sciences, 1, 229–236.
best? Individual differences in memory for events that occurred doi:10.1016/S1364-6613(97)01068-1
in a science museum. Applied Cognitive Psychology, 6, 93–107. Schuirmann, D. J. (1987). A comparison of the Two One-Sided Tests
doi:10.1002/acp.2350060202 Procedure and the Power Approach for assessing the equiva-
Loftus, E. F., Miller, D. G., & Burns, H. J. (1978). Semantic inte- lence of average bioavailability. Journal of Pharmacokinetics
gration of verbal information into a visual memory. Journal of and Pharmacodynamics, 15, 657–680. doi:10.1007/bf01068419
Experimental Psychology: Human Learning and Memory, 4, Schuster, B. (2007). Police lineups: Making eyewitness identification
19–31. doi:10.1037//0278-7393.4.1.19 more reliable. National Institute of Justice Journal, 258, 2–10.
Lövdén, M. (2003). The episodic memory and inhibition accounts Silverman, C. (2016, November 16). This analysis shows
of age-related increases in false memories: A consistency check. how viral fake election news stories outperformed
Journal of Memory and Language, 49, 268–283. doi:10.1016/ real news on Facebook. Retrieved June 3, 2019, from
S0749-596X(03)00069-X https://www.buzzfeednews.com/article/craigsilverman/
Mahé, A., Corson, Y., Verrier, N., & Payoux, M. (2015). viral-fake-election-news-outperformed-real-news-on-facebook
Misinformation effect and centrality. European Review of Applied Thomas, A. K., & Sommers, M. S. (2005). Attention to item-
Psychology, 65, 155–162. doi:10.1016/j.erap.2015.03.001 specific processing eliminates age effects in false memories.
Marche, T. A., Jordan, J. J., & Owre, K. P. (2002). Younger Journal of Memory and Language, 52, 71–86. doi:10.1016/j.
adults can be more suggestible than older adults: The influ- jml.2004.08.001
ence of learning differences on misinformation reporting. Volz, K., Stark, R., Vaitl, D., & Ambach, W. (2019). Event related-
Canadian Journal of Aging, 21, 85–93. doi:10.1017/ potentials differ between true and false memories in the misin-
S0714980800000660 formation paradigm. International Journal of Psychophysiology,
Norman, K. A., & Schacter, D. L. (1997). False recognition in 135, 95–105. doi:10.1016/j.ijpsycho.2018.12.002
younger and older adults: Exploring the characteristics of Wellek, S. (2010). Testing statistical hypotheses of equivalence
illusory memories. Memory & Cognition, 25, 838–848. and noninferiority (2nd ed.). New-York: Chapman and Hall/
doi:10.3758/BF03211328 CRC.
Perrin, J. (Producer), Rachedi, A. (Producer), & Gavras, C. (Director). Wylie, L. E., Patihis, L., McCuller, L. L., Davis, D., Brank, E. M.,
(1969). Z [Motion Picture]. France, Algérie: Valoria Films. Loftus, E. F., & Bornstein, B. H. (2014). Misinformation effects
R Core Team (2017). R: A language and environment for statis- in older versus younger adults: A meta-analysis and review.
tical computing. Vienna, Austria: R Foundation for Statistical In M. P. Toglia, D. F. Ross, J. Pozzulo, & E. Pica (Eds.), The
Computing [Computer software manual]. Retrieved from Elderly Eyewitness in Court. New York: Psychology Press.
https://www.R-project.org/ doi:10.4324/9781315813936