This paper is based on a study about post-event information and the way it leads to misinformation
Abstract
This paper is based on a study about post-event information and the way it leads to misinformation. The study looks at the influence caused by forms of sentence surface on misinformation effects. In this paper the study is based on viewing a given film clip. In this case, the participants received a narrative on a post-event clip. The narrative described the film event fully. In the post-even narrative, critical sentences were presented either in the form of a statement or in the form of a question. The presentation contained misinformation rather than questions that reflected the original event. At the time of the final recall test, the participants were told that all relevant information that was presented within the post-event narratives was hardly in the actual event. They were therefore requested not to report such information. The experiment compared to previous findings to demonstrate that post-event information that is presented in the form of affirmative statement could produce the misinformation effect. It was also found that post-event information that is presented in the form of a question would always increase the ability of individuals to recall the correct information as well as promote the reduction of false recalls. This usually happens irrespective of whether the information contrains misleading or event studied items. The main findings of the study brought about a ruling that post-event information that is presented in the form of a question generates a condition that is similar to the testing effect.
Introduction
Many people who witness events often get information about such events even before having any chance to retrieve the fine details of the events. Typically, post-event information appear in various contexts in everyday life. Specific information about given events, which could include a specific name or even an object, are bound to reappear later in different occasions or in different places. Post-event information is also likely to take various forms. These different forms are said to generate the effect associated with tempering the accuracy of details about the original event. Loftus was the first individual to structure a standard experiment that was meant to examine the way memory reports about an event could be affected by post-event information effects. The paradigm in this case involved the basic phases. In this case participant started by experiencing an event, which could include watching a video or even some sequence of slides that depicts a relatively complex event. The second phase would then be a case whereby the participants receive some verbal information, which contains various suggestions that are misleading as far as the event details are concerned. To achieve this, participants have to be tested on their memory about the event. Misinformation effect is said to occur whenever the reports of the participants include more details that are suggested erroneously than the reports from the participants who hardly gets the misleading information. Typically, post-event information appears within the contexts of post-event questions or even the case of post-war narratives concerning the event witnessed. Misinformation effects mainly occur when such post-event narratives contain misinformation that wrongly reflects the event. It could as well occur when post-event questions are initiated with false presuppositions. In such cases, the misinformation could be contradictory specifically to the original events.
Deviating slightly from the work of Loftus, the daily experience by individuals are associated with cases of distorted information, which go to an extent of altering the way people think of events they had already witnessed. In everyday life, there are some cases when people are required to have their memory as sharp as possible. This requirement is hardly fulfilled in many cases due to the effect of memory distortion through distorted information. Even the case of eyewitnesses can hardly be perfectly upheld given that people look for further opinions about the events they have already witnessed. In such cases, the misinformation ends up being contradictory to the already witnessed events that should be held as original. This case could be reflected to the case of Loftus whereby participants witnessed a green car. The participants later received some misleading information. They ware for instance asked the way the car was fast but with reference to a different color such a white. The misleading question also included an aspect of misleading information about the actual location. The question in this case was reframed as, “How fast was the white sports car going when it passed the barn while traveling along the country road” Loftus, 1975)? This question brought in the aspect of color and location distortion. The car was green but the question referred to a white car. Again, there was hardly any barn in actual event. While the participants remember the details of the original event accurately, they were very likely to include false details such as white color and a barn, when the color was actually green and there was hardly any barn respectively.
Literature Review
Past studies has largely tackled the case of post-event information and how it affects the ability of people in remembering the original events. The work of Loftus is very remarkable in this case. In his work, he used various experiments to indicate the way people could report on past events that they readily witnessed. These experiments were initiated with the use of control experiments in which some participants would report on the original events without having received any misleading information (Loftus, 1975). On the second information, the participants would receive misleading information about the past events, in which the information is meant to reflect some aspects that were hardly in the original event. In the case of the control experiment, participants are more likely to remember and report the complete event without any problems relating to wrong details about the original event.
Conversely, the case of misleading information made the participants to falsely remember some additional details that were not in the original event. The participants include such details from their ideology and thoughts that they could have skipped some information by mistake. According to Frost (2000), people are more likely to include false information when asked to report about an event that they had originally witnessed given that the reporting is done after at least one week. The individuals in this case are found to face a problem of contradictory misinformation, which is generated by the misleading information they receive from other individuals about the same event. It is also found that whenever the false information appears to be the direct focus of a question asked by a second party, the distortion of the original information depends on the nature of the question. Usually, a presupposed question generates a greater influence on people’s notion about an event that they had originally witnessed. Given a case whereby participants of an experiment are asked whether they saw a bus in a film they were watching, this case may generate insignificant influence on their later reports about the original film event. In contrast, a presupposed question such as whether the individual saw some people boarding the bus could affect the individual memory about the event in which the likelihood of reporting falsely about the film would be relatively high (Loftus, 1975). Contradictory information about the film event eliminates the effect of misinformation and causes a spillover effect in which people would become increasingly resistant to some other misleading information CITATION Lof75 l 1033 (Loftus, 1975).
memory of the event (Loftus, 1975). Similarly, blatantly contradictory misinformation about the event not only eliminated the misinformation effect, but also caused the ‘spillover’ effect in which participants became more resistant to other misinformation (Loftus, 1979).
Despite the abundance of work on the misinformation effect, previous studies have not examined the influence of sentence types presented in post-event narratives. In everyday life, post-event information can appear in different forms, such as an affirmative statement (It is a stop sign), a question (Is it a stop sign?), or a negative statement (It is not a stop sign). The purpose of this study was to investigate the influences that post-event information presented in a question form may have on the misinformation effect. It is important to note that we used post-event narratives containing misinformation in this study, instead of indirect questions containing false presuppositions, as those used by previous studies. Although the post-event narrative in the current procedure contained questions, participants only received the post-event narrative and did not have to answer questions about the original event before the final memory test.
We compared the misinformation effect produced by post-event information presented as affirmative statements and as questions. A question might be processed as an affirmative statement. For example, a question could be misremembered as a statement, which is referred to as the statement bias (Pandelaere & Dewitte, 2006). If the statement bias also occurs in the post-event misinformation paradigm, post-event information about an event detail presented in a question form (Is it a stop sign?) should produce a misinformation effect equivalent to that presented in a statement form (It is a stop sign). In addition to this hypothesis, there is another possible prediction. Post-event information in a question form may draw participants’ attention to the critical item. Furthermore, a question may create an effect similar to the testing effect, in which taking a memory test improves later retention (see Roediger & Karpicke, 2006b for a review). Participants may try to retrieve the critical item from the original event. If this is the case, then the misinformation effect should be reduced and correct recall should be increased for post-event information in a question form. This memory enhancement effect would occur not only for questions about the studied (correct) item, but even for questions about the misleading item.
The misinformation effect can be caused by a reduced ability of participants to remember event details, source confusion errors, or response biases. Loftus and colleagues proposed that the misinformation effect occurs when misleading information overwrites original information (Loftus, 1981; Loftus, Miller, & Burns, 1978) and, as a result, permanently erases the original information from memory. However, other studies assert that while misinformation renders original details less accessible, they are still in memory. The misinformation effect may occur when an individual has difficulty in gaining access to the original information during a test as compared with the relative ease of access of more recent misinformation (Christiaansen & Ochalek, 1983). Another possibility is that information from different sources becomes integrated or confused. Participants may incorrectly attribute their memory of the post-event misinformation to their memory of the original event (Lindsay, 1990). Furthermore, response biases can also cause the misinformation effect; for example, participants who do not perceive the critical information of the original event but do notice the misinformation would believe that the misinformation occurred in the original event. Similarly, demand characteristics may play an important role. Participants may reason that misinformation must be correct and go along with the suggestion because the experimenter provided it (McCloskey & Zaragoza, 1985a, 1985b).
To rule out the influence of response biases and demand characteristics in this study, we adopted Lindsay’s (1990) opposition test. In this procedure, participants are told that any relevant information presented in the post-event narrative is not in the original event and should not be reported on the test. Participants’ ability to remember the source of misleading details is set in opposition to the tendency to report those suggestions. Because participants are told not to report the misleading details from the post-event narrative, the opposition test effectively eliminates the influences of response biases and demand characteristics.
EXPERIMENT 1
Experiment 1 followed the standard post-event misinformation paradigm and manipulated the forms of sentences presented in the post-event narrative. We used three formats of sentences: an affirmative statement with misleading information, a question with misleading information, and a question with the studied (correct) information. In the final cued-recall test, participants were told that any answer mentioned in the narrative was wrong and should not be reported.
Method
Participants and design
Participants included 72 undergraduate students from our University who participated voluntarily to fulfil part of their course requirements. They were non-psychology majors, taking an Introduction to Psychology course. This experiment was a 2 × 3 mixed design. The within-subjects factor was type of item (control item vs. specific item) presented in the post-event narrative, and the between-subjects factor was type of sentence that contains the specific item (affirmative statement with misleading items vs. question with misleading items vs. question with studied items).
Materials
The study material was a 6-min film clip about the morning routine of a college student. Two versions of the film were created. The second version was created by replacing the critical items in the original version through editing to make sure that the two versions were identical except for eight critical items. These critical items were a type of cleaning tool (a mop or a sweeper), brand of chewing gum (Airwaves or Doublemint), brand of cleansing cream (Dove or Johnson), type of tea (Rooibos tea or milk tea), type of textbook (psychology or statistics), colour of the cap of eyedrops (white or yellow), type of vehicle (scooter or bicycle), and newspaper name (United News vs. China Daily). Half of the participants in each sentence condition received each version of the film.
The post-event information describing the target events was read by a male voice and recorded through a computer. The length of recording was about 3 min 30 s. A male student described in first person the detailed events depicted in the film (e.g., When the alarm clock rang this morning, I still felt very sleepy. Therefore, I stayed in bed for a while before I got up and turned off the alarm clock…. I reached for the toothpaste beside the Dove cleansing cream and squished a little of it on the toothbrush…). All the materials were in Chinese and English translations of the original Chinese were presented in this report. There were eight critical sentences in the post-event narrative, each containing a control item or a specific item. The four control items were category names of the specific items (e.g., cleansing cream). The four specific items were either a studied item (e.g., Johnson cleansing cream) or a misleading item (e.g., Dove cleansing cream), depending on the sentence type. We created two versions of the post-event narrative for each version of the film to counterbalance the control and specific items. Thus, for each sentence type, there were four versions of the post-event narrative. There were three types of sentences that contain the specific item. The first type was an affirmative statement with misleading items (e.g., I reached for the toothpaste beside the Dove cleansing cream). The second type was a question with misleading items (e.g., I reached for the toothpaste beside the cleansing cream. Was it Dove?). Finally, the third type was a question with studied (correct) items (e.g., I reached for the toothpaste beside the cleansing cream. Was it Johnson?). Other sentences that did not contain the critical item were identical for all versions of the post-event narratives.
The memory test was a cued-recall test containing eight questions about the eight critical items. All participants received the same cued-recall questions. The eight questions about four control items and four specific items were presented randomly. An example of the cued-recall question would be What brand of the cleansing cream beside the toothpaste was it?
Procedure
Participants were tested individually and randomly assigned to one of the three sentence conditions. Each experimental session consisted of four phases. Participants viewed the film clip and then listened to the recording describing the events in the film. After a delay period during which participants filled in an unrelated questionnaire, participants took the cued-recall test.
Participants were asked to watch carefully a 6-min film clip about a morning routine of a male college student. After that, participants were told that they were going to listen to a diary written by the character in the film. The diary described what the character had done during that morning. Participants were told to listen carefully and that their memory for the diary would be tested. This was to ensure that participants paid full attention to the recording. After listening to the post-event information, participants were asked to fill in a 20-min questionnaire unrelated to the experiment.
In the final phase, participants had to answer questions about the eight critical items in the film. The questions were presented one by one in random order in the centre of a computer screen. Participants wrote down their answers on an answer sheet. The next question appeared after participants wrote down their answers. The instructions given were similar to those used in the study by Lindsay (1990). Participants were given the following instructions: ‘Some answers were not mentioned in the narrative; others were mentioned, but inaccurate. Thus, the answers to the questions were either not mentioned at all or wrong. In other words, the correct answer would never come from the diary. Please answer the questions based on what you remembered from the film’. On completion of the experiment, the participants were instructed not to discuss the experiment with any other potential participants.
Results
Table 1 presents the mean recall rates of misleading items and studied items for different sentence types. He calculated the number of misleading items and studied items recalled by the participants in each of the post-event item/sentence type condition and then the number was divided by the total number (4) of critical item presented. To examine the effect of post-event information and sentence type on the recall of misleading items, a 2 (control item vs. specific item) × 3 (affirmative statement with misleading items vs. question with misleading items vs. question with studied items) ANOVA was performed on the recall rate of misleading items. Both main effects of item type and sentence type were significant, F (1, 69) = 8.11, MSE= .015, p < .01, ηp2= .11; and F (2, 69) = 12.00, MSE= .015, p < .001, ηp2= .26, respectively. The interaction was also significant, F (2, 69) = 6.94, MSE= .015, p < .01, ηp2= .17. We then conducted separate t-tests to examine the misinformation effect for each sentence type. A misinformation effect occurred when participants recalled more misleading items after receiving the specific items than the control items. The analyses showed a misinformation effect only for the affirmative statement with misleading items, M= 23 versus M= .06, t (23) = 3.76, p < .01, and not for the other two sentence types, t (23) < 1 for both.
Table 1. Mean recall rates (standard deviation) of misleading and correct items as a function of post-event item/sentence type in Experiments 1 and 2
Post-event item/Sentence type Type of items recalled
Misleading item Correct item
Experiment 1
Control item .06 (.02) .73 (.04)
Statement with misleading items .23 (.04) .61 (.04)
Control item .05 (.02) .64 (.05)
Question with misleading items .07 (.03) .78 (.04)
Control item .03 (.02) .68 (.03)
Question with studied items .02 (.01) .84 (.04)
Experiment 2
Control item .05 (.02) .71 (.05)
Statement with misleading items .21 (.04) .70 (.05)
Control item .05 (.02) .66 (.05)
Question with misleading items .08 (.04) .82 (.05)
This study also investigated the effect of post-event information on the recall of correct items. A 2 (control item vs. specific item) × 3 (affirmative statement with misleading items vs. question with misleading items vs. question with studied items) ANOVA showed a significant interaction effect, F (2, 69) = 6.84, MSE= .043, p < .01, ηp2= .17, and marginally significant main effects of item type and sentence type, F (1, 69) = 3.64, MSE= .043, p = .061, ηp2= .05; and F (2, 69) = 2.54, MSE= .038, p = .087, ηp2= .07. To examine whether there was a misinformation effect on the recall of correct items for each sentence type, separate t-tests were performed. A misinformation effect occurred when participants recalled fewer correct items after receiving the specific items than after receiving the control items. Follow-up t-tests showed that the affirmative statement with misleading items revealed a significant misinformation effect. Participants recalled more correct items after receiving the control items than after receiving the specific items, however, it was not statistically reliable, M= .73 versus M= .61, t (23) = 1.77, p = .09. The opposite pattern was found for both the question with misleading items and the question with studied items. Participants recalled more correct items after receiving the specific items than after receiving the control items, M= .78 versus M= .64, t (23) = 2.07, p < .05; and M= .84 versus M= .68, t (23) = 4.29, p < .001, respectively.
Table 2 shows percentages of participants who recalled 1, 2, 3, or 4 misleading items. More than 70% of the participants who received misleading affirmative statements recalled at least one misleading item, whereas the percentage dropped to a quarter when the participants received the misleading questions. The proportion was even lower when the questions contained studied items.
Table 2. Percentages of participants who recalled 1, 2, 3, or 4 misleading items in Experiments 1 and 2
Post-event item/Sentence type Number of misleading items recalled
Experiment 1
1 2 3 4
Control item .25 0 0 0
Statement with misleading items .54 .13 .04 0
Control item .21 0 0 0
Question with misleading items .21 .04 0 0
Control item .13 0 0 0
Question with studied items .08 0 0 0
Experiment 2 Control item .21 0 0 0
Statement with misleading items .29 .21 .04 0
Control item .21 0 0 0
Question with misleading items .25 .04 0 0
Discussion
This experiment found that post-event information presented in an affirmative statement form produced the misinformation effect. The post-event misinformation increased participants’ recall of misleading items and impaired their recall of correct items. This pattern of results did not appear when post-event information was presented in a question form. Post-event information presented in a question form, regardless of whether it contained a misleading item or a studied item, increased the recall of correct items. Moreover, the percentage of participants who recalled at least one misleading item dropped significantly from the affirmative statement condition to the question condition, especially when the question contained studied items.
In this experiment, the affirmative statement and question that contained misleading items had a different sentence structure. The misleading item was embedded within the affirmative statement (e.g., I reached for the toothpaste beside the Dove cleansing cream), whereas the misleading item was in a separate sentence in the question condition (e.g., I reached for the toothpaste beside the cleansing cream. Was it Dove?). Thus, for the question sentence type, participants were very likely to detect the discrepancy between the misleading item and the original item because of the salience of the misleading item. Studies have showed that discrepancy detection increased participants’ resistance to the misinformation effect (e.g., Blank, 1998; Loftus, 1979; Tousignant, Hall, & Loftus, 1986). Experiment 2 was designed to rule out this explanation for the reduced misinformation effect found in the question condition.
EXPERIMENT 2
The purpose of Experiment 2 was to rule out an alternative explanation of Experiment 1 based on the salience of the misleading item in the question form. Experiment 2 compared the question sentence type with a matched statement sentence type in which the misleading item was not embedded in a sentence but in full focus (i.e., I reached for the toothpaste beside the cleansing cream. It was Dove). If it was the question form, and not the salience of the misleading item that reduced the misleading effect, then Experiment 2 should replicate the finding of Experiment 1 that recall of misleading items reduced and correct recall increased in the question condition.
Method
Participants and design
Participants were 48 undergraduates from our University. They were non-psychology majors and received payment for their participation in the study. This experiment was a 2 × 2 mixed design. The within-subjects factor was type of item (control item vs. specific item) presented in the post-event narrative, and the between-subjects factor was type of sentence that contained the specific item (affirmative statement vs. question).
The rest of the method was the same as that of Experiment 1, except for the following difference. All the affirmative statements were changed to match the sentence structure of the question sentence type (e.g., I reached for the toothpaste beside the cleansing cream. Was it Dove?). This was done by removing the last Chinese word from a question. In Chinese language, the word is added to the end of a sentence to form a yes/no question. Thus, the statement and question sentence types were exactly the same except for the last word.
Results
Table 1 presents the mean recall rates of misleading items and studied items for different sentence types. To examine the effect of post-event information and sentence type on the recall of misleading items, a 2 (control item vs. specific item) × 2 (affirmative statement vs. question) ANOVA was performed on the recall rate of misleading items. Both main effects of item type and sentence type were significant, F (1, 46) = 9.07, MSE= .023, p < .01, ηp2= .16; and F (1, 46) = 3.95, MSE= .024, p < .053, ηp2= .08, respectively. The interaction was also significant, F (1, 46) = 4.03, MSE= .023, p < .05, ηp2= .08. We then conducted separate t-tests to examine the misinformation effect for each sentence type. Only the affirmative statement [M= 21 vs. M= .05, t (23) = 3.16, p < .01], and not the question [t (23) < 1] showed the misinformation effect.
As for the effect of post-event information on the recall of correct items, a 2 (control item vs. specific item) × 2 (affirmative statement vs. question) ANOVA was performed. Although both main effects and the interaction effect did not reach statistical significance, follow-up t-tests showed that participants recalled more correct items after receiving the misleading question than after receiving the control items, M= .82 versus M= .66, t (23) = 2.28, p < .05. No difference in correct recall was found between the misleading statement and the control items, t (23) < 1. Table 2 shows percentages of participants who recalled 1, 2, 3, or 4 misleading items. More than half of the participants who received misleading affirmative statements recalled at least one misleading item, whereas the percentage dropped to 29% when the participants received the misleading questions. These results were consistent with those of Experiment 1.
Discussion
Experiment 2 matched the sentence structure of the question and affirmative statement. In both sentence types, the misleading item was not embedded in a sentence but in full focus. The main finding that post-event information presented in a question form reduced the misinformation effect and increased correct recall was replicated in Experiment 2. Thus, Experiment 2 ruled out the explanation based on the salience of the misleading item. In other words, it was the question form, and not the discrepancy detection that reduced the misleading effect.
GENERAL DISCUSSION
Experiment 1 used the standard post-event misinformation paradigm and manipulated the forms of sentences presented in the post-event narrative. The critical information was presented in one the following formats of sentences: an affirmative statement with misleading information, a question with misleading information, and a question with the studied (correct) information. Experiment 2 matched the sentence structure of the question and the affirmative statement. In both sentence types, the misleading item was not embedded in a sentence but in full focus. Both experiments replicated previous findings (Lindsay, 1990) that post-event information presented in an affirmative statement form produced the misinformation effect. The post-event misinformation increased participants’ recall of misleading items and impaired their recall of correct items. This pattern of results did not appear when we presented post-event information in a question form. Post-event information presented in a question form, regardless of whether it contained a misleading item or a studied item, did not produce the misinformation effect; rather, it increased the recall of correct items.
Although questions have been shown to sometimes be more easily misremembered as statements than vice versa (Pandelaere & Dewitte, 2006), the present study found that misinformation presented as a question did not produce the misinformation effect, as an affirmative statement did. When the post-event information was presented in an affirmative statement form, participants incorporated this new information into their memory, which overwrote the original memory details or made these memory details less accessible. Post-event information in a question form might not be easily integrated with the original memory. Furthermore, post-event information in a question form seems to create an effect similar to the testing effect (Roediger & Karpicke, 2006a), which refers to the phenomena that the act of retrieving an item from memory (testing) produces better memory for that item relative to restudy the item. After watching the film, the participants were given questions about the critical item in the event. These questions had an effect on sentence focus and drew participants’ attention to the information that the questioner wanted, that is, the critical items (Graesser, Singer, & Trabasso, 1994). More importantly, when the participants tried to answer the question and to retrieve the relevant information, a condition was created that was similar to that which produces the testing effect. Consequently, presenting questions about the event details immediately after the original events had occurred enhanced correct recall and reduced false recall in the final memory test.
During the final cued-recall test, we informed participants that any relevant information presented in the post-event narrative was not in the original event and that the