PropertyValue
is nif:broaderContext of
nif:broaderContext
is schema:hasPart of
schema:isPartOf
nif:isString
  • Fifty-four volunteers (34 female) participated in the current study. Ages ranged from 19 to 63 years (M = 26.2, SD = 7.9). The study was marketed as an eyewitness experiment and the participants were mainly students recruited from different departments at Lund University. All participants gave written informed consent and also agreed to have the interaction audio recorded. The study was approved by the ethical committee of the Department of Psychology of Lund University and has been conducted according to the principles of the Declaration of Helsinki. In short, the procedure of the experiment contained five steps: first the participants were shown a film of a kidnapping, and then they filled out a questionnaire about events in the film. After this their answers were manipulated using sleight-of-hand. A discussion about their manipulated answers followed in an attempt to create choice blindness. Finally, a second questionnaire was administered in order to test the effect of the potential choice blindness. A more detailed account follows. The participants were tested singly and the experiments started off by showing each participant a four-minute film on a 42” screen. The film was shot from a first-person perspective and shows the kidnapping of a young woman from a bus stop. Two men disembark from a car and approach the woman, then make her unconscious using chloroform. In the commotion the older, glasses-wearing man pulls out a gun, whilst the younger, cap-wearing man drags the woman into their car before they drive off from the scene. The film has been used in previous eyewitness research [49–51]. After the film was shown, the participants filled out a two-paged statement questionnaire, containing 20 statements about the film (e.g. The older man wears glasses). The questionnaire was influenced by the design used by Hall et al. [4]. Six of the 20 statements were target items (Table 1) that were manipulated, discussed with all participants, and the basis of the data analysis. The remaining 14 were filler items. For half of the participants items 9, 15 and 20 (version one) were manipulated, and for the other half items 6, 12 and 17 (version two) were manipulated. Every other participant was assigned to version one, and every other to version two. Thus, the different versions functioned as controls to one another. (See S1 and S2 Files for details) Table data removed from full text. Table identifier and caption: 10.1371/journal.pone.0173606.t001 Statements and their corresponding questions in the two questionnaires. aThe correct answer to the statements is “disagree.”bThe correct answer is “agree.” The participants were instructed to answer each statement by marking a 96 mm long continuous scale, where a mark made to the left signified “Disagree”, a mark made to the right signified “Agree”, and a mark made somewhere in the middle of the scale signified “Uncertain.” If the participants had no recollection of a particular statement whatsoever, the mark was thus made on the absolute midpoint of the scale. Several previous choice blindness studies have also used continuous scales for ratings [3, 4, 14, 18, 52]. After the participants completed filling out the first sheet of the statement questionnaire, they handed it over to experimenter A, and continued with the second sheet of the questionnaire. Whilst the participants were busy filling out the second sheet, experimenter A manipulated one statement answer on the first sheet. In order to perform the manipulations the questionnaires were cleverly constructed using reattachable stickers (Fig 1). Meanwhile, experimenter B timed the filling out of the two sheets. Figure data removed from full text. Figure identifier and caption: 10.1371/journal.pone.0173606.g001 A staged demonstration of how the manipulations were carried out. (A) The participant fills out the statement questionnaire, by indicating level of agreement on a scale. (B) Experimenter A pulls away the sticker with the participant’s answers. (C) The experimenter fills out a new scale, which was hidden underneath the sticker, and manipulates some of the answers. (D) The new, manipulated questionnaire. There was no predetermined rule for the size of the manipulations across the scale. Instead, the manipulations were made so that the markings were to signify a different answer along the scale, i.e. “Agree”, “Disagree” and “Uncertain” as described above, than the answer originally given by the participants, therefore making some of the manipulations sizeable. Also, each manipulation was made with the intent of creating a believable pattern of responses, mimicking the response pattern of each participant. For example, if a participant frequently marked towards the poles of the scale, the manipulations used this same pattern. Manipulations ranged from 32.0 mm to 96.0 mm and the mean distance of the manipulations was 55.7 mm (SD = 17.5) on the 96.0 mm scale. Incorrect answers were also manipulated along the scale, in the direction of the correct answer. While the second sheet was also manipulated in the same manner as the first, the participants were busy chatting with experimenter B. All in all three statement answers were manipulated. Experimenter B then discussed the same ten predetermined statements with all participants. For every predetermined statement, the participants were asked to describe their recollection of what they had seen on the film, and justify this. During the discussion notes were kept by experimenter A, and a sound recording was also made to support these. After this discussion, the participants were given the memory questionnaire to fill out. The purpose of the memory questionnaire was to examine whether the manipulations made in the statement questionnaire would affect the participants’ recollection of what they had witnessed on the film. It consisted of 15 questions, six of which were related to the six target items of the statement questionnaire. This questionnaire had no scale. For eight of the questions the participants chose between the alternatives “Yes”, “No” and “Don’t know.” In several studies of the misinformation effect [19, 26] a forced choice design is used in the memory questionnaire, where the participants must choose either”Yes” or”No” for the question probing the misinformation given earlier. This is a poor reflection on reality, where people are seldom forced to give their recollection of an event if they do not have one. Therefore, we chose to give our participants the chance to say that they did not know. The remaining seven questions were answered in free response, where the participants also had the choice to say they did not know. The two questionnaires hence had different designs. There was no credible reason for the participants to fill out a questionnaire identical to the one they had just filled out and discussed a few minutes ago. There was also a risk that the participants would strive to give the same answer in both questionnaires, rather than try to remember what they had witnessed on the film. The scale in the statement questionnaire was therefore not graded in any way, thus making it unlikely for the participants to think of their answers in the dichotomous yes-or-no-terms used in the memory questionnaire. To further curb this consistency endeavor, the participants were also explicitly asked to give their recollection of the film for the final questionnaire. To facilitate a comparison between the two questionnaires, and thereby examining if a misinformation effect had occurred, the answers in the statement questionnaire were categorized. The participants were not aware of this categorization when they gave their original answers. To give them a fair chance to detect that the manipulation did not reflect their original opinion, the categories were verbalized during the discussion by the experimenter, for example”The older man wears glasses… you disagree with that.” Finally, in order to find out whether they had suspected that any of their answers on the statement questionnaire had been manipulated experimenter A asked the participants, in accordance with other choice blindness studies [2, 4, 10, 13, 18], four detection questions: “How did this feel?” “Did you think of anything in particular during the experiment?” “Did you think of anything in particular when the two of you discussed the statements?” and “Did you notice anything strange about the questionnaires?” After this the true purpose of the study was revealed. During the experiment there was no delay between the different steps, and no filler task was used. The participants were categorized as either non-detectors or detectors. Non-detectors were those who at no point during the experiment expressed any kind of suspicion concerning the manipulations. These participants, however, could still correct the manipulated answers, but without suspecting that the answers were not their own. The inclusion of these participants in the subsequent analysis is in accordance with previous choice blindness studies [3, 4, 11, 12, 52, 53]. The participants expressed that they had misread the statement or that they had marked the wrong end of the scale: “I was sloppy when I filled this out. I’m sorry! I’m pretty sure he doesn’t hit her in the head.” These participants were then asked if they wanted to correct their marking to better match their recollection. Detectors, however, expressed suspicion concerning the manipulations. Concurrent detectors clearly stated that the manipulated answers were not their own during the discussion about the statements. Retrospective detectors were those who, during the four detection questions at the end of the experiment, expressed that their answers had been manipulated in some way. Both concurrent detectors and retrospective detectors were removed from the subsequent analyses. In order to quantify the size of the choice blindness manipulations across the scale, the difference between the markings originally made by the participants and the manipulated markings was measured in millimeters. To then examine whether the answers in the memory questionnaire were affected by the manipulations made in the statement questionnaire, the scale was divided into three equally sized parts. A marking made in the part to the left was categorized as “No”, a marking made in the middle part was categorized as “Don’t know”, and a marking made in the part to the right was categorized as “Yes.” These three categories were chosen because these were the three response alternatives in the memory questionnaire, thus allowing a comparison between the two questionnaires. The answers in the memory questionnaire were categorized in the following way: “Changed in the direction of the manipulation,” “Changed independently of a possible manipulation” and “Not changed.” The first category “Changed in the direction of the manipulation” meant that a choice blind participant’s answer in the memory questionnaire did not belong to the same category as that given in the statement questionnaire. For example, if the participant’s answer in the statement questionnaire was manipulated from “Disagree” to “Agree”, and the participant later gave the answer “Yes” or “Don’t know” in the memory questionnaire, the recollection would have been changed in accordance with the manipulation, as the expected answer would be “No.” The second category “Changed independently of a possible manipulation” also meant that the participant’s answer in the memory questionnaire did not belong to the same category as that given in the statement questionnaire. However, this category was used either because a choice blind participant had changed their answer in the opposite direction of the manipulation, or because a participant whose markings had not been manipulated answered differently in the two questionnaires. The third and final category “Not changed” meant that the same answers were given in both questionnaires. In the main statistical analysis of the transference between the two questionnaires, item 6, “A yellow car stops at the bus shelter” and item 9, “The older man jumps into the car and drives away” were discarded, in part because they were answered in free response in the memory questionnaire, and in part because these items were somewhat ambiguous in their nature. Categorizing the free response answers involved a great deal of interpretation, which was deemed too arbitrary. The ambiguity in the wording of the statements made the participants likely to reinterpret them as they justified markings they themselves had not made. After a reinterpretation of the statement, this no longer matched the corresponding question in the memory questionnaire. Including the two ambiguous items, however, did not change the overall misinformation effect outcome. That is, if all six items were included in the analysis a similar result, but with a somewhat smaller effect size, was still evident when comparing recollection of choice blind items with non-manipulated items, χ2 (2, n = 196) = 92.4, p < .001, Cramer’s V = 0.69, p < .001.
rdf:type