Higham,
P.A., (1989). Believing details known to have been suggested. British
Journal of Psychology, 89, 265-283.
Overview
Reason for this study: Many studies
have investigated source monitoring errors and misinformation effects,
but the author feels that the results can be easily misinterpreted. In
the present study, he attempts to better explain the phenomenon of misinformation
effects, using slightly different methods than previous researchers.
Problems with interpreting misinformation
effects:
Demand Characteristics: *People
may believe the postevent information accurately represents the original
event. McCloskey and Zaragoza (1985) eliminated the suggested item as a
memory option to control for DC.
*People may choose the suggested
event b/c they assume if the experimenter prepared the information, then
it should be correct. Adding a warning before the test provides a good
control for this effect.
Source Guessing: People may
choose one of the options that indicate an event occurred, b/c they ‘know’
it happened somewhere, but don’t really know where. Control: Adding a ‘know’
option to eliminate source guessing.
The present study
Experiment 1:
-
56 subjects watched a video of an armed
robbery in a convenience store. They were told that their memory would
be tested for peripheral details of the crime scene.
-
Long Delay Condition: One hr.
after the crime, half of the subjects answered misleading questions. Then,
48 hours later they did a filler task, followed by the source monitoring
test 1 hr. after that.
-
Short Delay Condition: One hr.
after the crime, half of the subjects did a filler task. Then, 48 hours
later they answered the misleading questions, followed by the SM test 1
hr. after that.
-
Misleading questions were counterbalanced
with half mentioning a critical item not in the video and half mentioning
a non-critical object in the video or a critical item in neutral form.
These questions appeared in 2 versions.
-
Filler tasks were either a questionnaire
a/b social desirability or an orientation task which manipulated the level
of processing of non-critical items.
-
A warning was given (although very
slight) that some objects didn’t occur in the video or questions.
-
Subjects indicated ‘V’ for Video, ‘Q’
for Questions, ‘both’, ‘N’ for No memory, or ‘K’ for Know. Following ‘K’,
subjects guessed a possible source.
-
There were 16 objects on the recognition
test. (10 critical and 6 filler)
Delay conditions summary:
Short delay: 48 hrs. b/w video and
questions & 1 hr. b/w questions and test (p. 276)
Long delay: 1 hr. b/w video and
questions & 48 hrs. b/w questions and test.
Results:
-
Overall, subjects indicated that objects
suggested in the misleading questions were more likely to have been seen
in the video than were control objects.
-
Suggested objects were more likely
to be contributed to ‘both’ (instead of ‘video only’) than were control
objects.
-
There was an item by suggestion interaction
and delay by suggestion interaction, indicating that as delay increased,
so did the size of the misinformation effect for ‘video only’ responses.
-
The ‘know’ responses were non-significant
(It appears that the misinformation
effect in this expt. was due to false conscious recollection of the suggested
objects. Even if people have an accurate memory of the source, they can
still come to believe that they encountered an item from a different source.)
-
If an object was presented in the questions,
subjects were more likely to remember the source than if the object was
presented in the video.
(Poor memory for the video probably
contributed to the misinformation effect.)
Experiment 2
Because the delay conditions were
confounded, a new experiment was employed.
Discrepancy detection (DD):
The ability to detect a discrepancy b/w the event information and the post-event
misinformation. If DD is high, then misinformation effect (ME) would be
small and vice-versa. So, if the interval b/w video and question is longer,
DD would be less likely to occur yielding a greater ME. In Expt. 1, if
the V-Q delay was longer, then the Q-T delay was shorter. So, we really
don’t know which delay did what! Expt. 2 was necessary to determine if
the effect of ‘video only’ responses was a result of the change to the
V-Q delay or the Q-T delay. (See Table 2, p. 276)
-
46 subjects watched the video and answered
misleading questions 30 min. later.
-
The test was administered one hour
after completion of the questions.
-
All the materials were the same as
in Expt. 1.
Results
-
Overall, subjects seemed to rate the
suggested items as having been seen more often than rating the control
items as having been seen.
-
There was no significant effect for
‘know’ responses, so this supports expt. 1 in that subjects are displaying
misinformation effects due to false recollections.
-
The difference b/w the effect of suggestion
on ‘video only’ responses in Expt. 2 and Expt. 1/short delay was non-significant.
Therefore, increasing the delay from question to test caused the effect
of suggestion, not the decreased delay from video to question.
-
As in Expt. 1, subjects had better
memory for items presented in the questions than in the video.
General Discussion
-
People most likely made source errors
caused by false recollection of seeing objects that were only read about
in the questions.
This study helps to eliminate possible
misinterpretations of the misinformation effect, confirming the occurrence
of false recollection, rather than source guessing.
-
Higham makes a comparison with sleeper/false
fame experiments in which subjects chose mutually exclusive sources. However,
in the present study, many chose ‘both video and questions.’
Higham discusses opposition instructions,
which can cause subjects to choose a category indicating that an item was
never encountered. He states that this can cause an "underestimation of
the proportion of misled participants who come to believe that they saw
a suggested detail" (p. 282).
-
False recollection of a detail in an
event still occurs despite subjects’ knowledge that it could have been
encountered in post-event information.