Printer Friendly

Effects of Dress and Speech Style on Credibility of Co-Witness and Misinformation Acceptance.

While watching a person give an eyewitness account on the news or in the courtroom, we may not consider how their attire may impact how we perceive their testimony. However, there are factors that sway our implicit opinions of and confidence in these witnesses. Attribution theory (Kelley, 1972) states that people will make inferences about a person while considering the appropriateness of his or her language, behavior, and appearance in a particular social context. For example, Ruva and Bryant (2004) manipulated the age and speech style (powerless or powerful) of a witness providing testimony in a trial transcript. Powerless speech includes pauses, hesitations, and filler words such as "I think" and "maybe" as opposed to powerful speech that is void of disfluencies or qualifications. Powerful speech is generally viewed as more credible than powerless speech (Ruva & Bryant, 2004; Schmidt & Brigham, 1996). Ruva and Bryant found that a 22-year-old speaking in a powerless way was seen as less credible than a 6-year-old speaking in the same way. A 6-year-old speaking powerfully exceeds general expectations, so more credibility is attributed to a child speaking this way. An adult (22-yearold) speaking in a powerful way is consistent with expectations, so credibility is rated as typical for that age.

Appearance also plays a role in how we perceive people (for review, see Lennon, Johnson, Noh, Zheng, Chae, & Kim, 2014). Implicit personality theory (Schneider, 1973) says we naturally make inferences and fill gaps in our knowledge when we do not entirely understand something, so people will infer personality traits from appearance clues. Clayton, Lennon, and Larkin (1987) found that perception of personality traits was linked with appearance clues, such as physical characteristics, style of dress, etc. People tend to use style of dress to make inferences about likelihood of drinking, smoking, and using profanity (Workman, Arseneau, & Ewell, 2004) as well as a woman's sexuality (Elliot, Greitemeyer, & Pazda, 2013; Montemurro & Gillen, 2013). Style of dress or fashion may also lead to judgment of other characteristics of a person and allows others to guide their behavior towards that person accordingly (Kaiser, 1983). Individuals are more confident in those who are professionally dressed (Rehman, Nietert, Cope, & Kilpatrick, 2005; Rufa'i, Oyeyemi, Oyeyemi, Bello, & Ali, 2015).The present study explored the effects of style of dress of the co-witness to determine if there are differences in credibility and confidence in a witness depending on how they are dressed. Attribution theory would also predict that a person who sounds confident but is not well-dressed would stand out and receive a boost in credibility because of this perceived inconsistency.

The misinformation effect occurs when a witness of an event includes information in his or her testimony that is congruent with other sources, such as other witnesses, but this information is inaccurate (for review, see Loftus, 2005). The misinformation effect is relevant in this experiment because incorrect information provided by a more credible witness should more often be accepted as true. Time has its impact on the misinformation effect, in that the more time given between the event and the introduction of post event misinformation, the more susceptible an individual is to misinformation effects (Tousignant, Hall, & Loftus, 1986). When more time passes between the event and misinformation, the event memory is weaker and a discrepancy between misinformation and the original memory is harder to detect, causing the individual to accept the misinformation. This is known as the discrepancy detection principle, which states if differences between memory and misinformation are not immediately detected, there is a greater susceptibility to misinformation effects (Loftus, Miller, & Burns, 1978; Tousignant et al., 1986). When the discrepancy is not detected source misattribution can occur; witnesses may incorrectly remember that the new information came from the original event (McCloskey & Zaragoza, 1985; Zaragoza & Lane, 1994).

Loftus, Donders, Hoffman and Schooler (1989) found that people are as confident in their misinformation reports as they are in their own memory reports. This confidence in inaccurate statements can be especially harmful in criminal cases where the testimony plays an important role in the conviction of the suspect. Eyewitness identification is viewed as reliable evidence in these criminal cases. However, jurors can be ignorant of biases as well as the possibility of the misinformation effect (Durham & Dane, 1999; Estrada-Reynolds, Gray, & Nunez, 2015; Steblay, Besirevic, Fulero, & Jimenez-Lorente, 1999).

In the present study we manipulated speech style (powerless, powerful) and dress (casual, professional) of a co-witness to ascertain their effects on witness credibility and in turn susceptibility to misinformation presented by that witness. Inaccurate information provided by a co-witness should be more readily believed if that cowitness is perceived to be credible. The existing literature explains that individuals may attribute specific characteristics to others based on dress. Although the effects of dress on witness credibility have not been studied specifically, the pattern of results found in previous research is consistent with the prediction that a professionally dressed witness would be viewed as more credible than a casually dressed witness. The manipulation of speech style was incorporated to attempt to replicate Ruva and Bryant's (2004) findings that witnesses who spoke powerfully were rated as more credible than witnesses who spoke powerlessly. We expected an interaction of speech style and dress on credibility. Ruva and Bryant (2004) also found an interaction between age and speech style on witness credibility. Age set an initial expectation for credibility that was either verified or called into question by the speech style the witness used. In our experiment we used dress instead of age to set an initial expectation. We hypothesized that the difference in credibility between participants in the powerful speech group and the powerless speech group would be greater for the group where the co-witness appeared casually dressed than for the group where the co-witness appeared in professional dress. Essentially, we expected speech style to have more of an effect on the casual group than the professional group. This interaction is expected due to attribution theory (Kelley, 1972) and the findings of Ruva and Bryant (2004). If a casually dressed witness is speaking in a powerful voice, it would exceed expectations, therefore that witness would be rated as more credible. Additionally, if a professionally dressed witness is speaking in a powerless voice, it would not meet expectations and his testimony would be considered less credible. We expected the speech style to have more of an effect on the casually dressed witnesses because previous research has shown greater differences in credibility on the expected-to-bepowerless witness (Ruva & Bryant, 2004). The predictions for misinformation acceptance follow the same patterns. We expected more misinformation acceptance in the professional dress condition than the casual dress condition and in the powerful speech condition than the powerless speech condition. We also expected the interaction to show a greater effect of speech style in the casual dress condition.



Participants in this study were 147 undergraduate students from a private liberal arts and sciences college in southwestern Pennsylvania. Participants included 90 women and 57 men between the ages of 18 and 28 (M = 19.34 SD = 1.48). The sample was predominately Caucasian at 93.9% (Black or African American 3.4% and Other 2.7%). Ten participants' data were removed from the analyses because of computer error, 15 for recognizing the confederate, six for being more than three standard deviations from the mean for misinformation reported, and one for answering his phone during the study. All participants were recruited using convenience sampling; all participation was voluntary. Students were recruited during their class times by providing their name, email address, and time slot preference on a sign-up sheet. Some professors offered extra credit for participation. Emails were sent to each participant to remind them of their upcoming appointment for the study.


All participants watched the same video of a male pedestrian getting hit by a car in a parking lot. The video included; a group of students walking through a parking lot, a car nearly backing into an oncoming car, and the male pedestrian walking out in front of the car from behind a van and getting hit. Each participant was assigned to watch one of four eye-witness testimony videos. Each video was of a confederate who was reading a script that contained five pieces of misinformation about the accident. All videos contained the same pieces of misinformation. Participants were randomly assigned to a dress (professional or casual) and speech style (powerful or powerless) yielding four conditions in a 2 x 2 independent groups factorial design. A filler task of simple mathematical equations was given. A cued recall questionnaire regarding the accident included five specific questions incorporating the misinformation, as well as another 10 general questions about the accident video. Demographic items, including age, sex, ethnic background, and major were asked at the end of the study for the purpose of describing the sample.

Co-Witness Videos. A confederate was video recorded while he read a script about the accident video. For dress, he was either wearing a white dress shirt and blue tie for the professional condition or a light grey sweatshirt for the casual condition. For the speech style conditions, he was either talking powerfully or powerlessly (see Appendix A). There were four possible videos that a participant could have received.

Filler Task. A series of simple math questions were asked in order to serve as a filler task. The math questions were designed to prevent rehearsal of video information, but they were simple enough that they did not stress the participant.

Memory Questions. Fact-based questions (see Appendix B) were asked about the case to ensure the participant thoroughly watched and was engaged in the video. They were also used to determine the number of questions answered incorrectly due to the acceptance of misinformation.

Witness Credibility Scale. This scale was taken from Brodsky, Griffin, and Cramer's (2010) description of the instrument with two adjective pairs from Ruva and Bryant (2004) added to the end. Participants were asked to rate the co-witness on 22 characteristics. The questions required an answer on a 10-point Likert scale. For example, an assessment for kindness would be from 1 (extremely unkind) to 10 (extremely kind) and an assessment for truthfulness would be from 1 (extremely untruthful) to 10 (extremely truthful).


During recruitment, participants were informed that they would be assigned to video record or type their responses on a computer. However, all participants typed their responses as the possibility of being video recorded was a form of deception used to ensure that the participants believed the co-witness was a previous participant himself. Participants arrived at the assigned room during the time slot that they previously signed up for and were separated into private rooms. They were instructed to sign in to the Qualtrics project on a desktop computer. Participants began by reading the informed consent document and clicking "Agree." All participants watched the same video of a man getting hit by a car. Next, they were told that they had been randomly assigned to type their responses to a free recall prompt instead of having their responses video recorded. Upon completion of the free recall, participants were randomly assigned to watch one of four possible videos of a co-witness giving misinformation. This was an act of deception; participants were told that they would be watching another participant give his or her recount of the car accident video. However, the other participant was a confederate who was following a script. Next, participants completed the filler task to limit rehearsal of information followed by the completion of the cued recall questionnaire. The witness credibility scale was then completed by all participants. Finally, participants completed the demographic items. Following demographics, each participant was provided with a link to receive extra credit if offered by his or her professor. Delayed debriefing was utilized to ensure deception was maintained during the study and was delivered through email.


A 2x2 between-subjects Factorial MANOVA was used to calculate main effects of dress and speech style on misinformation reported and credibility of the co-witness. The variable of credibility was split into five factors; trustworthiness, likeability, confidence, knowledgeability, and overall credibility. These factors were determined from past factor analyses that were completed in previous studies (Brodsky et al., 2010; Ruva & Bryant, 2004). The 2 x 2 Factorial MANOVA was also used to calculate any potential interactions between dress and speech style.

The results showed that the overall credibility of the co-witness was significantly higher when he spoke powerfully (M = 6.56, SD = 1.44) rather than powerlessly (M = 5.88, SD = 1.19); F(1,146) = 9.10, p = .003, [[eta].sub.p.sup.2] = 060. There was no difference in overall credibility when the co-witness dressed professionally (M = 6.34, SD = 1.32) versus casually (M = 6.02, SD = 1.36); F (1,146) =.33, p =.57. There was a significant interaction between dress and speech style for overall credibility; F(1,146) = 4.69, p =.032, [[eta].sup.2.sub.p] =.032. As shown in Figure 1, the powerful speech resulted in more credibility than the powerless speech only in the casual dress condition.

The results also indicated that the confidence of the co-witness was significantly higher when he spoke in a powerful manner (M = 7.58, SD = 1.45) than in a powerless one (M = 6.45, SD = 1.60); F(1,146) = 17.22, p <.001, [[eta].sub.p.sup.2] = 11. There was no difference in confidence when the cowitness dressed professionally (M = 7.26, SD = 1.43) versus casually (M = 6.63, SD = 1.76); F(1,146) = 2.05, p =.15. There was also no interaction between dress and speech style for confidence (F(1,146) = 2.79, p =.097. Table 1 contains descriptive statistics for interactions on credibility subscale scores.

The results also showed that the trustworthiness of the co-witness was rated significantly higher when he spoke in a powerful manner (M = 5.52, SD = 2.06) rather than a powerless one (M = 4.73, SD = 1.55); F(1,146) = 6.80, p =.01, [[eta].sub.p.sup.2] =.045. There was no difference in trustworthiness when the co-witness was dressed professionally (M = 5.21, SD = 1.94) versus casually (M = 4.94, SD = 1.72); F(1,146) =.05, p =.82. There was also no interaction between dress and speech style for trustworthiness; F(1,146) = 1.77, p =.19.

The results also indicated that the co-witness was rated as significantly more knowledgeable when he spoke powerfully (M = 5.84, SD = 1.84) rather than powerlessly (M = 5.17, SD = 1.35); F(1,146) = 6.33, p =.013, [[eta].sub.p.sup.2] =.042. There was no difference in knowledge-ability when the co-witness was dressed professionally (M = 5.60, SD = 1.71) versus casually (M = 5.33, SD = 1.51); F(1,146) =.072, p =.79. There was a significant interaction between dress and speech style for knowledgeability; F(1,146) = 4.56, p =.034, [[eta].sub.p.sup.2] =.031. This interaction follows the same pattern as the interaction for overall credibility.

The results also showed that the co-witness was not rated as significantly more likeable when he spoke powerfully (M = 7.28, SD = 1.49) rather than powerlessly (M = 7.18, SD = 1.75); F(1,146) =.18, p = .67. There was no difference in likeability when the co-witness was dressed professionally (M = 7.26, SD = 1.39) versus casually (M = 7.19, SD = 1.87); F(1,146) <.001, p =.99. There was no significant interaction between dress and speech style for likability F(1,146) = 3.22, p =.075.

Results indicate the amount of misinformation reported was significantly higher when the co-witness was dressed professionally (M = 0.70, SD =.74) rather than when he was dressed casually (M =.44, SD = .58); F(1,146) = 5.87, p =.017, [[eta].sub.p.sup.2] =.039. There was no difference in misinformation acceptance between the powerful (M =.58, SD =.70) and the powerless group (M =.56, SD =.65); F(1,146) =.081, p =.78. There was also no interaction between dress and speech style for misinformation acceptance; F(1,146) =.054, p =.82 (Table 2).

The results indicated that the total number of correct answers was not higher when the co-witness was dressed professionally (M = 9.43, SD = 1.94) rather than when he was dressed casually (M = 9.16, SD = 1.88); F(1,146) =.49, p =.49. There was also no difference in total number of correct answers when the co-witness spoke powerfully (M = 9.29, SD = 1.88) versus powerlessly (M = 9.30, SD = 1.94); F(1,146) =.074, p =.79. There was also no interaction between dress and speech style for total number of correct answers; F(1,146) = 1.12, p =.29; see Table 2.


The purpose of our study was to see if the way a person dressed and how they spoke had an impact on how credible they were perceived to be. As predicted from previous findings (Ruva & Bryant, 2004; Schmidt & Brigham, 1996), results indicated a higher rating of overall credibility, confidence, knowledgeability, and trustworthiness when the co-witness was speaking powerfully rather than when he was speaking powerlessly, which we hypothesized. This finding supports implicit personality theory (Schneider, 1973) because participants used the co-witness's speech style to draw their own conclusions about the co-witness's credibility. Attribution theory also contributes to the explanation of the effect of speech style. Participants may have expected the co-witness, who was purportedly a fellow undergraduate participant, to be nervous speaking in front of a camera, so powerless speech may have been expected because of nervousness. Thus, when our co-witness spoke in a powerful manner, participants may have been more likely to rate him as more confident and credible because it went against what was expected in that situation (Kelley, 1972; Ruva & Bryant, 2004). Contrary to our hypotheses and past research (Rehman et al., 2005; Rufa'i et al., 2015), overall credibility and its subscales did not differ by dress. The interactions between dress and speech style for knowledgeability and overall credibility were significant. The powerful speech resulted in greater credibility and knowledgeability than the powerless speech only in the casual dress condition. These interactions support attribution theory (Kelley, 1972) and the findings of Ruva and Bryant (2004) because the participants rated the co-witness to be more credible and knowledgeable when he exceeded the expectations that they put on him after viewing his attire.

We examined whether dress and speech style affected the amount of misinformation that was reported. Logically, if dress and speech style can affect the perceived credibility of the witness, then those effects should carry through to affect participants' adoption of misinformation provided by the co-witness. Unexpectedly, the pattern of results found for credibility did not appear for misinformation reported, perhaps due to the order in which the measures were taken. The Witness Credibility Scale was presented to participants after they had answered questions about the eyewitness event for the final time. The co-witness's speech style and dress were expected to have implicit effects on perceived credibility, which would then have implicit effects on how his testimony was received. The Witness Credibility Scale asks participants to make their implicit impressions explicit, which is not something witnesses are typically asked to do. Participants rated the co-witness on 22 traits, forcing them to think more deeply about the witness. We delayed the Witness Credibility Scale so that completing it could not influence participants' responses to questions about the eyewitness event. If participants had completed the Witness Credibility Scale earlier, we might have found different results for misinformation reported.

As initially predicted, the amount of misinformation that was reported by participants was significantly higher when the co-witness was dressed professionally than when he was dressed casually, which supports attribution theory. Speech style did not produce any significant differences in the amount of misinformation that was reported nor was the interaction between dress and speech style significant. The professional dress of the co-witness may have seemed more out of place than the casual dress because of the social context of our experiment (Kelley, 1972). The casual setting of our experiment led participants to be more likely to report what the co-witness said when dressed professionally rather than casually (Clayton et al., 1987; Kaiser, 1983; Ruva & Bryant, 2004).

This incorrect information could have made its way into the participants' reports in two ways. If the participant was unsure of the detail from his or her own memory of the event, then he or she may have accepted the co-witness's testimony (Loftus, 2005). Source misattribution, which is when a person mistakes the source of the information that was reported, could also have contributed to our results. When participants accepted misinformation from the co-witness, they might have assumed the information came from the video however, it came from the co-witness (Zaragoza & Lane, 1994).

Misinformation regarding the number of people walking in a group in the beginning of the video was not reported by any of the participants. This result could have been due to the primacy effect, which is a part of the serial position curve. The primacy effect is when an individual better recalls the first few pieces of information that they have received than the pieces of information that were given in the middle of the sequence (Murdock, 1962). Whether because of the primacy effect or other factors affecting the salience of the actors, participants could have considered this bit of misinformation obvious. With this misinformation presented at the beginning of his testimony, participants may have been less receptive to other information given by the co-witness than they would have otherwise, resulting in an overall low rate of misinformation acceptance. Our intention for including misinformation was to go beyond credibility ratings to see if accuracy of testimony would be affected. However, future research could do more by manipulating misinformation to see whether the presence of said misinformation interacts with dress and/or speech style.

The rate of misinformation reporting was very low, possibly because of the small amount of time between the event and the post-event information and the post-event information and the recall (Mudd & Govern, 2004). Given this short timing between the event video and cowitness video, rates of discrepancy detection may have been high. With the short delay between the co-witness video and final recall, it is possible that the participants remembered what the co-witness said and explicitly remembered that he said it, so source misattribution was likely fairly low. Therefore, much of the misinformation that was reported may have been explicitly accepted as accurate because participants did not recall those particular details from their own memories of the event. The low rates of misinformation reporting could also be attributed to the free recall that we had participants complete before viewing the co-witness' statement. The free recall was used to support the deception that the cowitness video was of an actual participant. However, having the participants write down what they saw before hearing the co-witness' recall might have reinforced their own recollection of the event (Wang, Paterson, & Kemp, 2014). Given the timing between phases of the experiment and the fact that participants performed free recall of the event before watching the co-witness video, it is not surprising that rates of misinformation reported were low.

The present study demonstrated that speech style and dress can affect witness credibility and these effects are consistent with attribution theory. Inconsistencies in findings between the credibility ratings and misinformation reporting may be an artifact of the order in which the measures were presented. Effects on misinformation reporting were limited and small, perhaps due to the short intervals between phases of the experiment. Future studies might increase these intervals to decrease the likelihood of discrepancy detection and weaken source memory for details of the co-witness' testimony.


Brodsky, S. L., Griffin, M. P., & Cramer, R. J. (2010). The Witness Credibility Scale: An outcome measure for expert witness research. Behavioral Science and the Law, 28, 892-907.

Clayton, R., Lennon, S. J. & Larkin, J. (1987). Perceived fashionability of a garment as inferred from the age and body type of the wearer. Home Economics Research Journal, 15(4), 237-247.

Durham, M. D., & Dane, F. C. (1999). Juror knowledge of eyewitness behavior: Evidence for the necessity of expert testimony. Journal of Social Behavior & Personality, 14(2), 299-308.

Elliot, A., Greitemeyer, T., & Pazda, A. (2013). Women's use of red clothing as a sexual signal in intersexual interaction. Journal of Experimental Social Psychology, 49, 599-602.

Estrada-Reynolds, V. C., Gray, J. M., & Nunez, N. (2015). Information integration theory, juror bias, and sentence recommendations captured over time in a capital trial. Applied Cognitive Psychology, 29(5), 713-722. doi: 10.1002/acp.3155

Kaiser, S. B. (1983). Toward a contextual social psychology of clothing: A synthesis of symbolic integrationist and cognitive theoretical perspectives. Clothing and Textiles Research Journal, 2(1), 1-9.

Kelley, H. H. (1972). Attribution in the social interaction. In E. E. Jones, D. E. Kanouse, H. H. Kelley, R. E. Nisbett, S. Valins, & B. Weiner (Eds.), Attribution: Perceiving the causes of behavior (pp. 1-26). Morristown, NJ: General Learning Press.

Lennon, S. J., Johnson, K. K. P., Noh, M., Zheng, Z., Chae, Y., & Kim, Y. (2014). In search of a common thread revisited: What content does fashion communicate? International Journal of Fashion Design, Technology and Education, 7(3), 170-178. doi: 10.1080/17543266.2014.942892

Loftus, E. F. (2005). Planting misinformation in the human mind: A 30-year investigation of the malleability of memory. Learning & Memory, 12, 361-366.

Loftus, E. F., Donders, K., Hoffman, H. G., & Schooler, J. W. (1989). Creating new memories that are quickly accessed and confidently held. Memory & Cognition, 17(5), 607-616.

Loftus, E. F., Miller, D. G., & Burns, H. J. (1978). Semantic integration of verbal information into a visual memory. Journal of Experimental Psychology: Human Learning and Memory, 4(1), 19-31.

McCloskey, M., & Zaragoza, M. (1985). Misleading postevent information and memory for events: Arguments and evidence against memory impairment hypotheses. Journal of Experimental Psychology: General, 144(1), 1-16.

Montemurro, B., & Gillen, M. (2013). How clothes make the woman immoral: Impressions given off by sexualized clothing. Clothing and Textiles Research Journal, 31(3), 167-181.

Mudd, K., & Govern, J. M. (2004). Conformity to misinformation and time delay negatively affect eyewitness confidence and accuracy. North American Journal of Psychology, 6(2), 227-238.

Murdock, B. B., Jr. (1962). The serial position effect of free recall. Journal of Experimental Psychology, 64(5), 482-488.

Rehman, S. U., Nietert, P. J., Cope, D. W., & Kilpatrick, A. O. (2005). What to wear today? Effect of doctor's attire on the trust and confidence of patients. The American Journal of Medicine, 118, 1279-1286.

Rufa'i, A. A., Oyeyemi, A. Y., Oyeyemi, A. L., Bello, R. S., & Ali, M. U. (2015). Physiotherapists attire: Does it affect patients' comfort, confidence and overall patient-therapist relationship? International Journal of Physiotherapy, 2(5), 677-684.

Ruva, C. L. & Bryant, J. B. (2004). The impact of age, speech style, and question form on perceptions of witness credibility and trial outcome. Journal of Applied Social Psychology, 34(9), 1919-1944.

Schmidt, C. W., & Brigham, J. C. (1996). Jurors' perceptions of child victim-witnesses in a simulated sexual abuse trial. Law and Human Behavior, 20(6), 581-606.

Schneider, D. J. (1973). Implicit personality theory: A review. Psychological Bulletin, 79(5), 294-309.

Steblay, N., Besirevic, J., Fulero, S. M., & Jimenez-Lorente, B. (1999). The effects of pretrial publicity on juror verdicts: A meta-analytic review. Law and Human Behavior, 23, 219-235. doi:10.1023/A:1022325019080

Tousignant, J. P., Hall, D., & Loftus, E. F. (1986). Discrepancy detection and vulnerability to misleading postevent information. Memory & Cognition, 14(4), 329-338.

Wang, E., Paterson, H., & Kemp, R. (2014). The effects of immediate recall on eyewitness accuracy and susceptibility to misinformation. Psychology, Crime & Law, 20(7), 619-634.

Workman, J., Arseneau, N., & Ewell, C. (2004). Traits and behaviors assigned to an adolescent wearing an alcohol promotional T-shirt. Family and Consumer Sciences Research Journal, 33(1), 498-516.

Zaragoza, M. S., Lane, S. M. (1994). Source misattributions and the suggestibility of eyewitness memory. Journal of Experimental Psychology: Learning, Memory, and Cognition, 20(4), 934-945.


Co-Witness Scripts

Below are the scripts that the confederate performed in the co-witness videos that were used in our study. Misinformation is included in both scripts. The words and phrases that are in brackets indicate the correct information.

Powerful Speech Script

At the beginning of the video, I saw a group of two [four] people walking together in the parking lot behind the library in the afternoon. After that, a woman dressed in a black top entered from the left and walked toward me. Then, a dark colored car entered from the left and a red [silver] car nearly backed into the darker car. The darker car beeped twice, then went around the red [silver] sedan. Once the dark car drove past the red [silver] car, the red car backed up out of its parking spot. I noticed whenever the red [silver] car backed up, it was a Chevy. The red [silver] Chevy then drove forward. After that, a guy walked out from behind a tan truck [van] talking on his cell phone [reading] and was wearing a green jacket and immediately got hit by the red [silver] car. The car that hit the pedestrian immediately stopped, and the driver got out and checked to see if the man was okay [the driver did not stop]. Throughout the video, I noticed four female and two male individuals, along with five cars in total.

Powerless Speech Script

At the beginning of the video, I'm pretty sure I saw a group of two [four] people walking together in the parking lot behind the library. I think it was about mid-afternoon. After the group separated, I think a woman dressed in a black top, I'm not too sure though, entered from the left, and walked toward me. Um, after that, a dark colored car entered from the left and I think I saw a red [silver] car nearly backed into the darker car. Well, I couldn't really hear too well, but I'm pretty sure the darker car beeped twice, then went around the red [silver] sedan. Once the dark car drove past the red [silver] car, the red [silver] car backed up out of its parking spot. The red [silver], maybe Chevy, sedan then drove forward. Um, after that, I think a guy walked out from behind a tan truck [van]. He might have been talking on his cell phone [reading] and wearing a green jacket and was immediately hit by the red [silver] car. I think the car that hit the pedestrian immediately stopped, and the driver probably got out and checked to see if the man was okay [the driver did not stop]. I'm not too sure how many people I saw in the video, but I might have saw four girls and two guys, along with five cars in total.


Memory Questions

Approximately how many people did you see in the video?

How many cars were present in the video?

What color was the car that hit the pedestrian?

What color jacket was the pedestrian wearing?

What was the pedestrian doing whenever he was hit by the car?

What type of vehicle did the pedestrian walk out from behind?

How many times did the dark colored car beep whenever the other car was backing up?

What color was the parked car on the right side of the screen?

How many people were walking in a group in the beginning of the video?

How many males are present in the video?

What brand was the car that hit the pedestrian?

Did the driver get out of the car to check if the pedestrian was okay?

What color was the woman's top that entered the frame around the left corner?

Based on the context of the video, what time of day did the incident take place in?

Where did the incident take place?

Author Note: Kaylee L. Gojkovich, Nicole L. Reitz, Victoria N. Monstrola, Madison M. Campbell, & Mark G. Rivardo, Department of Psychological Science, Saint Vincent College.

Kaylee L. Gojkovich, Nicole L. Reitz, Victoria N. Monstrola, Madison M. Campbell, & Mark G. Rivardo Saint Vincent College

Author info: Correspondence should be sent to: Dr. Mark Rivardo, Department of Psychological Science, Saint Vincent College. E-mail: mark.rivardo @

Caption: FIGURE 1. Dress x speech style interaction for mean overall credibility of the co-witness.
TABLE 1 Means (and Standard Deviations)
for Dress x Speech Style Interactions for
Credibility Subscales


Subscale                    Powerful       Powerless
Confidence                  7.56(1.37)     6.92(1.43)
Trustworthiness             5.39(2.17)     5.00(1.63)
Knowledgeability*           5.65(1.86)     5.55(1.54)
Likability                  7.09(1.47)     7.47(1.29)

                              Powerful      Powerless

Confidence                   7.62(1.59)     6.12(1.64)
Trustworthiness              5.73(1.90)     4.53(1.48)
Knowledgeability *           6.15(1.80)     4.91(1.14)
Likability                   7.59(1.51)     6.97(2.01)

*Dress x speech style interaction was significant;
F(1,146) = 4.56, p = .034, tf = .031.

TABLE 2 Means (and Standard Deviations) for Dress x
Speech Style Interactions for Misinformation Reported
and Total Correct

                              Powerful      Powerless

Misinformation Reported       .70(.76)       .71(.72)
Total Correct                9.55(1.97)     9.29(1.92)


                              Powerful      Powerless
Misinformation Reported       .40(.58)       .46(.58)
Total Correct                8.88(1.67)     9.31(1.98)

Please Note: Illustration(s) are not available due to copyright restrictions.
COPYRIGHT 2019 North American Journal of Psychology
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2019 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Gojkovich, Kaylee L.; Reitz, Nicole L.; Monstrola, Victoria N.; Campbell, Madison M.; Rivardo, Mark
Publication:North American Journal of Psychology
Date:Dec 1, 2019
Previous Article:The Theory of Ownership Price and Fear of Loss.
Next Article:Is Avoiding Meat Prosocial? Field Research on Meat and Vegan Influence on Prosocial Behavior.

Terms of use | Privacy policy | Copyright © 2020 Farlex, Inc. | Feedback | For webmasters