Upload
kellie-wilding
View
213
Download
1
Embed Size (px)
Citation preview
The almost unanimous false memory of the first
World Trade Center impact
Tia G. B. [email protected]. of Communication
Faculty of HumanitiesAalborg University
Denmark
SARMAC VI 2005 2
Introduction
Footage of first impact was not published the first day (Pezdek, 2003)
But many claim to have seen
Greenberg (2004): Bush 2 out of 3 times
Pezdek (2003): Avr. 73% Americans
SARMAC VI 2005 3
How come
Not unprecedented Crombag et al. (1996):
El Al, 55-66% Ost et al. (2002): Diana,
20 of 45 queried
WTC impact1 differs Footage later
Ideas of how (mainly cf. Greenberg, 2004) Imagery Narrative Post-event information
Benigning Hesitates
Maligning Access to aids
SARMAC VI 2005 4
Present studies
1. Prevalence and patterns in a non-US sample2. Replication (partly)3. Collective memory examplified
Method Study 1
Re-test part of FBM study (1 day-9 months) 100 non-psych. students (65% female)
Study 2 Similar questionnaire once only (12 months) 84 psych. students (71% female)
SARMAC VI 2005 5
Event structure and visual media
First, the North Tower of WTC was hit. It could have been an accident. When the South Tower was hit as well, it became obvious that it was on purpose (terrorism). Later, the collapse of the towers greatly extended the catastrophy. There is much to suggest that visual images have a special effect. Thus, I should like to know when you first saw footage of these turning points.
Saw the North Tower get hit:Same day at ___ o'clock or the day after at ___ o'clock or __ days after.How sure are you about that (1-7; 1=pure guess, 7=absolutely certain) Saw the South Tower get hit:Same day at ___ o'clock or the day after at ___ o'clock or __ days after.How sure are you about that (1-7; 1=pure guess, 7=absolutely certain) Saw a tower collapse:Same day at ___ o'clock or the day after at ___ o'clock or __ days after.How sure are you about that (1-7; 1=pure guess, 7=absolutely certain)
SARMAC VI 2005 6
Prevalence
Study 1 (N=100) 90% gave First-Day North time Of the remaining 10:
2 commented 4 others corrected
Study 2 (N=84) 86% gave First-Day North time Of the remaining 12 (14%):
6+ were trekking etc…
5 Missing
5 Other day
90 Same day
3 Missing
9 Other day
72 Same day
SARMAC VI 2005 7
Confidence
Study 1 First-Day vs Other-Day
answers: not less confident
Study 1, FD’s only North vs other turning-
points: higher confidence
Study 2 ditto Mean FD=5.0; OD=4.0
Study 2 ditto Mean North=4.9;
South=4.5; Collapse=4.2
N M SD Diff.
FD 88 5.3 1.533ns
OD 5 4.0 2.345
N=82 M SD Diff. from North
North 5.3 1.493
South 4.9 1.679 t(81)=4.416***
Coll. 4.5 1.723 t(81)=5.263***
SARMAC VI 2005 8
Patterns?
First-Day cases retained for further analysis For these,
reality cannot produce North confidence so what might?
SARMAC VI 2005 9
Correlates of North Confidence (N varies) Study1 Study 2
Imagery Visual image clarity (delayed) .197 -.119
Vividness (Sept 12 2001) .144 na
Vividness (delayed) .197 .105
Confidencein other answers
South (delayed) .859** .803**
Collapse (delayed) .633** .710**
Source (delayed) .257* na
Activity (delayed) .172 na
Location (delayed) .199 na
Flashbulb variables
Emotionality (Sept 12 2001) -.021 na
Emotionality (delayed) .178 .068
Surprise (Sept 12 2001) -.117 na
Surprise (delayed) .198 .019
Consequentiality (Sept 12 2001) -.228* na
Consequentiality (delayed) -.037 -.062
Rehearsal (Sept 12 2001) .062 na
Rehearsal (delayed) .170 .024
SARMAC VI 2005 10
North Confidence contributors
Study 2 Signicant model by SC x CC
accounts for 65% SC contributes 33%, CC ns
Study 1 SC x CC x source confidence x consequentiality Adj. R2=.733 F4,76=55.955, p<.001 SC contributes 51%, Source
confidence 6%, rest are ns
SARMAC VI 2005 11
Estimate types 13 possible sequences 3 account for 90% of cases in
study 1 (N=83) Ditto study 2 (93%, N=67)
Sequences Study 1 Study 2 Type
N<S<C 33.7 % 43.3 % Narrative
N=S=C 30.1 % 25.4 % Same estimate for all
N=S<C 26.5 % 23.9 % Same then narrative
All other 9.6 % 7.5 %
Estimate types
9,6%26,5%
30,1%33,7%
All otherN=S<C
SameNarrative
SARMAC VI 2005 12
Summarizing
The error is close to unanimous The (nessessarily wrong) North time is held with more
confidence than the (possibly correct) South and Collapse times
Nevertheless, North time seems to parasitize other answers, in particular South time Pattern of confidence suggest that much North confidence
derives from confidence in other parts of the memory Pattern of estimates suggest that North time is rarely
assessed by individual recollection of that specific episode component
Memory problem or lack of specific categorisation?
SARMAC VI 2005 13
Study 3
Collective memory condensed: DR-TV’s summary 11/9 2003
”De frygtelige timer i New York” (The terrifying hours in New York) © DR (Danmarks Radio)
Used at SARMAC VI with permission Translations by me
Pretty English sacrificed for synchronicity and brevity
SARMAC VI 2005 14
SARMAC VI 2005 15
Analysis
Like President Bush, journalists have access to memory aids
Nevertheless
SARMAC VI 2005 16
Error 1
The second plane is (also) reported to hit the northern tower
SARMAC VI 2005 17
Error 2
This footage is from a later speech
SARMAC VI 2005 18
Error 3
This is the North Tower collapsing
SARMAC VI 2005 19
Discussion
Interpretations other than ’memory error’ are possible Speech error + editing demands
However Indistinct North-South categories enhance risk of
errors 1 and 3 Compatible observations from other programmes
For us, this is the available information
SARMAC VI 2005 20
Discussion & conclusion Components’ reception not represented individually Imagery not required Post-event inclusion, or a South clip mistaken for North
Conclusion The assumption of having seen impact1 on the first day
is very frequent and confidently held A mistake afforded by:
insufficient specificity of categories inadvertent misinformation by media normal news processing
Confidence veridicality problem remains
Acknowledgements
Intellectual input Dan Greenberg ROCOCO group (Aarhus Uni) CPU and MÆRKK groups (Aalborg Uni)
Technical assistance Keld Ringgaard, Lea Skov Treebak, Per Mouritzen
Travel grant Obel Foundation
SARMAC VI 2005 22
smidt ud
SARMAC VI 2005 23
General discussion and conclusion
SARMAC VI 2005 24
Conclusion
The claim to have seen impact1 on the first day is very frequent and confidently held
This mistake seems seems afforded by: insufficient specificity of categories inadvertent misinformation by media
Grasping for meaning, sacrisficing details
Two general theoretical perspectives Recollection vs. temporality vs. judgment News reception vs. natural cross-modal synchronicity
SARMAC VI 2005 25
First impact
SARMAC VI 2005 26
SARMAC VI 2005 27
SARMAC VI 2005 28
SARMAC VI 2005 29
SARMAC VI 2005 30
SARMAC VI 2005 31
Conclusion
SARMAC VI 2005 32
SARMAC VI 2005 33
Study 1 First-day-group (N=90) and
other-day-group (N=5) reported equal
vividness confidence emotionality rehearsal, etc.
when providing the estimate
as well as 12/9-01
Closest candidate for a difference: Confidence in North Tower impact
time t(91)=1.84; p<.069
N M SD
FD 89 5.2 1.5
OD 4 3.8 2.6
But OD-comments (p. 7)
SARMAC VI 2005 34
Results
Alle tager fejl - figur eller tabel
SARMAC VI 2005 35
Intro / theory
Greenberg (2004): Bush 2 out of 3 Pezdek (2003): 73%
Crombag et al. (1996): El Al, 55-66% Ost et al. (2002): Diana, 20 of 45 asked
Imagery Narrative Source monitoring errors and failures
SARMAC VI 2005 36
How come
Not unprecedented Crombag et al. (1996): El Al, 55-66% Ost et al. (2002): Diana, 20 of 45 asked
WTC impact1 differs Souce monitoring error but not reality monitoring error
Ideas of why (mainly cf. Greenberg, 2004) Imagery Narrative Post-event information
SARMAC VI 2005 37
Intro / theory
Greenberg (2004): Bush 2 out of 3 Pezdek (2003): 73%
Imagery Narrative Post-event information
SARMAC VI 2005 38
Study 1 & 2
Study 1 Re-test of FBM study
(1 day-9 mts) 100 non-psych. students
(65% female)
Study 2 Once only
(12 mts) 84 psych. students
(x % female)
Questions: Error frequency and confidence? Patterns?
Method: Subset of flashbulb memory questionnaire
SARMAC VI 2005 39
Study 1
Questions: How prevalent in a foreign (to US) sample? Hints of why:
Follow the narrative? High imagery?
Method: Incl. in flashbulb study re-test after 9 mts
Original questionnaire was within 24 hours 100 non-psych. students (65% female)
SARMAC VI 2005 40
The errorPrevalence 89% gave North Tower time
on 11/9 2001 Of the remaining 11:
2 commented (missing) 1 described (missing) 4 corrected (3 OD)
Thus: Almost unanimous
t(91)=1.84; p=.069 But cf. OD-correcters
N M SD
FD 89 5.2 1.5
OD 4 3.8 2.6
Confidence
Konfidensdimsen dur ejpga. rettede data.
SARMAC VI 2005 41
Sequences
Sequences N Type
N<S<C 28 Narrative
N=S=C 25 Same estimate for all
N=S<C 22 Same then narrative
All other 8
SARMAC VI 2005 42
man kunne sætte korrelationen med syd og kollaps her
husk også at bemærke, at de 33% narrativ jo ikke er meget, givet min madding
Anatomy of the erroneous answers – answer types
Most common sequences
38,9%
27,8%
33,3%
All other estimates
Same answer for all
Following narrative
SARMAC VI 2005 43
Study 2
Aim: Replication
(Partly: early variables not available)
Method: Delay 12 months Footage questions from study 1
Embedded in questionnaire almost similar to re-test questionnaire in study 1
84 psych. students (71% female)
SARMAC VI 2005 44
The error
4,8%
9,5%
85,7%
Blank or ambiguous
Other day
Same day
Claims to have seen North Tower impact 11/9/01
5,0%
5,0%
90,0%
Blank or several ans
Other day
Yes
Prevalence 86% reported seing North
Tower impact on 11/9 2001 90% in study 1
t(77)=2.03; p<.05 FD > OD ns in study 1
N M SD
FD 71 5.0 1.8
OD 8 3.6 2.2
Confidence
SARMAC VI 2005 45
study 2
SARMAC VI 2005 46
Results
Footage errors and confidence
SARMAC VI 2005 47
Hypothesis
Stong confidence is related to …?
SARMAC VI 2005 48
Results
Confidence Confidence relateret til?
SARMAC VI 2005 49
study 1 & 2
SARMAC VI 2005 50
North Confidence ContributorsStudy 1 Regression analysis SC x CC x
source confidence x pers. conseq. x reception distinct.
Adj. R2=.737; F5,74=45.202, p<.001
SC contributes 50%, Source confidence 7 %, rest are ns.
Study 2 Signicant model by SC x CC
accounts for 65,2%. SC contributes 33%, CC ns.
SARMAC VI 2005 51
Further analysis of errors
Relative confidence Hvordan kan her være 85 når der kun var 83 før?!
Thus, pts. are more confident in
their (erroneous) time given for their first seeing the North Tower impact
than for their (possibly correct) times given for first seeing the other two turning point.
r (N=85) South Collaps
North .854** .633**
(N=85) M SD
North 5.29 1.487
South 4.85 1.687
Collapse 4.42 1.775
t(84)=4.685** t(84)=5.649**
Tia Hansen:
men dette gør jo typeopdelingen overflødig, for hvordan skulle den kunne redegøre for dette?
Skal denne slide gemmes til efter typeanalyser?
Tia Hansen:
men dette gør jo typeopdelingen overflødig, for hvordan skulle den kunne redegøre for dette?
Skal denne slide gemmes til efter typeanalyser?
SARMAC VI 2005 52
estimattyper, konfidens og korrelater Study 1 Mean NC does no differ between the three estimate types SC x source conf (hvad predikerer de to tilsammen?) x CC (og
de tre?) predikerer næsten det hele af same og same/narrativ, men kun ¼ af narrativ.
Study 2
SARMAC VI 2005 53
relativ konfidens og konk
SARMAC VI 2005 54
Relative confidence
Study 1
N = 82 M SD t
South Confidence 4.85 1.679t(81)=4.416 ***
North Confidence 5.28 1.493
t(81)=5.263 ***Collapse Confidence 4.50 1.723
Study 2
N = 67 M SD t
South Confidence 4.53 1.863t(65)=2.754 **
North Confidence 4.92 1.842
t(65)=4.224 ***Collapse Confidence 4.15 1.994
SARMAC VI 2005 55
Relative confidence
Study 1
N = 82 M SD Difference
South Confidence 4.85 1.679t(81)=4.416 ***
North Confidence 5.28 1.493
t(81)=5.263 ***Collapse Confidence 4.50 1.723
Study 2
N = 67 M SD t
South Confidence 4.53 1.863t(65)=2.754 **
North Confidence 4.92 1.842
t(65)=4.224 ***Collapse Confidence 4.15 1.994
SARMAC VI 2005 56
Further analysis of errors
Relative confidence Hvordan kan her være 85 når der kun var 83 før?!
Thus, pts. are more confident in
their (erroneous) time given for their first seeing the North Tower impact
than for their (possibly correct) times given for first seeing the other two turning point.
r (N=85) South Collaps
North .854** .633**
(N=85) M SD
North 5.29 1.487
South 4.85 1.687
Collapse 4.42 1.775
t(84)=4.685** t(84)=5.649**
SARMAC VI 2005 57
study 3
SARMAC VI 2005 58
How come
Not unprecedented Crombag et al. (1996):
El Al, 55-66% Ost et al. (2002): Diana,
20 of 45 queried
Likely reasons Imagery Narrative
WTC impact1 differs Footage later
additional possible reason: Post-event information
Greenberg observes: Benigning: Hesitates Maligning: Access to
aids
SARMAC VI 2005 59
Present study
1. Prevalence and patterns in a non-US sample
2. Replication (partly)
3. Collective memory examplified
SARMAC VI 2005 60
Study 1 & 2
Method Subset of flashbulb memory questionnaire
Study 1 Re-test part of FBM study (1 day-9 months) 100 non-psych. students (65% female)
Study 2 Similar questionnaire once only (12 months) 84 psych. students (71% female)
SARMAC VI 2005 61
ConfidenceFirst-Day vs Other-Day estimates
N M SD Difference between groups
Study 1 First-Day 88 5.3 t(91)=1.723, ns
Other-Day 5 4.0
Study 2 First-Day 71 5.0 t(78)=1.558, ns
Other-Day 9 4.0
North confidence vs South and Collapse confidence
N M SD Different from North
Study 1 North
South
Collapse
Study 2 North
South
Collapse
SARMAC VI 2005 62
Confidence
Study 1 11/9 estimates seem more
confident than Other-Day estimates, but
t(91) = 1.723, ns
Study 2 Ditto t(78) = 1.558, ns
St. 1 N M SD
FD 88 5.3 1.5
OD 5 4.0 2.3
St. 2 N M SD
FD 71 5.0 1.9
OD 9 4.0 2.3
SARMAC VI 2005 63
North Confidence contributors
Study 2 Signicant model by SC x CC
accounts for 65% SC contributes 33%, CC ns
Study 1 SC x CC x source confidence x consequentiality Adj. R2=.729 F4,79=56.949, p<.001 SC contributes 55%, Source
confidence 6%, rest are ns
SARMAC VI 2005 64
Correlates of North Time Confidence (N varies) Study1 Study 2
Imagery? Visual image clarity (delayed) .197 -.119
Vividness, reception (Sept 12 2001) .144 na
Vividness, reception (delayed) .197 .105
Confidence in other answers?
Source (delayed) .257* na
Activity (delayed) .172 na
Location (delayed) .199 na
South time (delayed) .859** .803**
Collapse time (delayed) .633** .710**
Flashbulb variables, traditional?
Emotional intensity (Sept 12 2001) -.021 na
Emotional intensity (delayed) .178 .068
Surprise (Sept 12 2001) -.117 na
Surprise (delayed) .198 .019
Pers. consequent. (Sept 12 2001) -.228* na
Pers. consequentiality (delayed) -.037 -.062
Flashbulb variables, ’ecological’?
Rehearsal (covert) (delayed) .170 .024
Reception distinct. (Sept 12 2001) -.273* na
Reception distinctivity (delayed) -.102 .238*
SARMAC VI 2005 65
Relative confidenceM SD Different from North C.
Study 1 (N=82)
North Confidence 5.28 1.493
South Confidence 4.85 1.679 t(81)=4.416 ***
Collapse Confidence 4.50 1.723 t(81)=5.263 ***
Study 2 (N=67)
North Confidence 4.92 1.842
South Confidence 4.53 1.863 t(65)=2.754 **
Collapse Confidence 4.15 1.994 t(65)=4.224 ***