
##4001269 ABSTRACT . The authors performed 2 experiments investigating the influence of the belief that a vehicle crashed on witnesses ' estimates of the vehicle 's speed . In Experiment 1 , participants saw a video of a civilian car being driven , after which they were assigned to 1 of 2 conditions . The 1st group was told that the vehicle subsequently crashed ; the 2nd group was not told that the vehicle crashed . The results indicted no differences between the 2 groups on a number of factors , including estimates of the vehicle 's speed . Experiment 2 was identical except that the video showed a police car using flashing lights and sirens . Participants who had been told that the car had crashed overestimated speed , the likelihood of a crash , and the likelihood of someone being killed . Participants who were not told that the vehicle crashed estimated the speed of the vehicle accurately . Confidence in their estimates of speed was not significantly different between the 2 groups . Results are discussed with regard to police investigations of road accidents . <p> @ @ @ @ @ @ @ @ @ @ , police , speed , witness <p> THE POLICE MUST INVESTIGATE vehicle accidents and , if necessary , initiate prosecutions against drivers who have caused crashes . In such investigations the police are typically interested in determining the speed of a vehicle . The reason for this is that speed is one of the most important causes of car crashes . Shibata and Fukuda ( 1994 ) analyzed the data from 33,821 traffic accidents for various risk factors , including driving without a license , consumption of alcohol , speed , and the use of a seat-belt . Speed was the most important risk factor for fatalities . <p> The police rely on a number of sources , such as skid marks and the location and extent of damage to a vehicle , to determine the speed of a vehicle that has been involved in a collision . However , in the majority of cases the most important source of evidence is an eyewitness ( see Kebbell &; Milne , 1998 ) . Therefore , it is important for the police to know how much reliance they can place on eyewitness @ @ @ @ @ @ @ @ @ @ found that witnesses can be reasonably good . For example , a recent study involved mock-eyewitness estimates of the speed of civilian and police cars . Witnesses were remarkably accurate , typically estimating less than 3 miles per hour ( 5 kmph ) above or below the actual speed of the vehicle ( Barnecutt , Pfeffer , &; Creswell , 1999 ; Froyland , Ainsworth , &; Pfeifer , 1998 ; Marshall , 1966 ; Semb , 1969 ) . <p> Nevertheless , in real police investigations witnesses may fare less well . An extensive body of research shows that eyewitness memory is susceptible to distortion . The reconstructive nature of memory means that post-event information may be incorporated into memory and witnesses can be confident in inaccurate information ( Ayers &; Reder , 1998 ; Smith , Kassin , &; Ellsworth , 1989 ) . This possibility is illustrated in a study by Loftus and Palmer ( 1974 ) in which participants were shown a film of a car accident . Later they were asked , " About how fast were the cars going when they smashed into each other @ @ @ @ @ @ @ @ @ @ " smashed " with one of the following words : " collided , " " bumped , " " hit , " or " contacted . " Although the words all refer to the coming together of two objects , they differ in what they imply about the speed and force of impact . <p> Participants who received the " smashed " version estimated the speed at 40.8 mph ( 65.7 kmph ) , whereas those given the " contacted " version estimated the speed at 30.8 mph ( 49.6 kmph ) . Furthermore , participants who received the " smashed " version were more likely to answer yes to the question " Did you see any broken glass ? " even though there was no broken glass . In fact , leading questions can have an influence even if they are very subtle . Loftus and Zanni ( 1975 ) found that even a change in the wording of the question from " the " to " a " increased the proportion of witnesses who were misled ( e.g. , " Did you see a broken headlight ? " or @ @ @ @ @ @ @ @ @ @ . <p> Fortunately , the use of leading questions can be minimized , and many police officers are trained to interview eyewitnesses in a manner that reduces the use of such questions ( Kebbell , Milne , &; Wagstaff , 1999 ) . Nevertheless , a witness 's account may be biased by suggestive post-event sources of information that are not introduced by a police officer . In most instances , when a witness is interviewed concerning a vehicle , it is because that vehicle has been involved in some sort of collision . The very fact that a witness knows that the vehicle has crashed may encourage the witness to attribute the crash to the speed of the vehicle , and this may cause the witness to overestimate the speed of the vehicle prior to the crash . <p> Also , the fact that a vehicle has been involved in a crash may alter eyewitness beliefs about the way that the vehicle was driven . Although speed can be a major factor in causing crashes , other explanations are possible and for some , the driver is not responsible . @ @ @ @ @ @ @ @ @ @ side street without seeing an oncoming vehicle . Research by Hendricks et al . ( 2001 ) into 723 crashes involving 1,284 vehicles suggested that 15.1% were caused by such errors . <p> Alternatively , mechanical failures such as a tire blowout or brake failure appear to be responsible in 4% to 13% of cases ( Treat et al. , 1979 ) . Consequently , in this study , we wanted to determine whether the belief that a vehicle crashed would influence eyewitnesses ' estimates about how well that vehicle was driven . The implication is that drivers who were driving perfectly safely may be falsely remembered as driving dangerously simply because they were subsequently involved in a crash . EXPERIMENT 1 <p> To address the idea that knowledge that a vehicle crashed may potentially influence witnesses ' estimates of speed and the dangerousness with which a vehicle was driven , we conducted an experiment to manipulate the knowledge that a vehicle crashed . We hypothesized that witnesses who believed that a vehicle had crashed would estimate the speed of the vehicle to be higher than those who did not know @ @ @ @ @ @ @ @ @ @ that witnesses who believed that the vehicle had crashed would remember the way the vehicle was driven as being more dangerous . Method <p> Participants <p> Participants were 33 prospective undergraduate students ( 8 men and 25 women ) . The average age of participants was 22.45 years ( range 17-56 , SD = <p> 11.30 ) . <p> Materials <p> A 1-min video was shot from a stationary position at the side of a street to produce plausible security camera footage . The beginning of the film showed the street and then a car being driven down the street at a constant 49.7 mph ( 80.0 kmph ) toward and then past the camera . We obtained from police files photographs of a crashed car and a not-crashed car identical to the one in the video in terms of color , make , and model . The crashed vehicle had sustained heavy frontal damage consistent with hitting another vehicle . <p> Procedure <p> We tested the participants in groups of up to 6 . Upon arrival they were seated at a distance of approximately 2 m from a 21-in. television/video @ @ @ @ @ @ @ @ @ @ are about to see was filmed by a security camera . Please watch carefully . " The participants were then randomly assigned to one of two conditions . <p> Car not crashed . In this condition we asked participants to watch the film , after which they were shown photographs of a " not-crashed vehicle . " We then asked them to read an unrelated magazine article as a filler task for 5 min before completing a questionnaire . The questionnaire required that they estimate the speed of the vehicle and provide a confidence judgment of the accuracy of their estimate . We asked them to rate their confidence judgments on a 10-point Likert-type scale ranging from pure guess ( 1 ) to absolutely certain ( 10 ) . We asked them " How dangerous was the way that the vehicle was being driven ? " Responses were given on a 10-point Likert-type scale ranging from not at all dangerous ( 1 ) to extremely dangerous ( 10 ) . We asked them " How likely was a crash ? " and " How likely was the driver to kill someone @ @ @ @ @ @ @ @ @ @ scale ranging from not at all likely ( 1 ) to extremely likely ( 10 ) . We then showed the participants pictures of a not-crashed vehicle and told them that this was what the car looked like after it came to a halt . <p> Car crashed . This condition was identical to the not-crashed condition with the exception that after watching the film we told participants , " Immediately after this video was taken this vehicle was involved in a crash . Here are some pictures of the vehicle after the crash . " We then showed them pictures of the crashed vehicle and told them that this was what the car looked like after it came to a halt . <p> After taking part in this experiment , all participants were debriefed . Results <p> Table 1 shows the means and standard deviations for mock witnesses ' estimates of the speed at which the vehicle was traveling and their perceptions of how dangerously it was being driven . A t test revealed no significant difference between conditions in mock witnesses ' estimates of speed , t(31) = 0.33 @ @ @ @ @ @ @ @ @ @ of the vehicle , t ( 15 ) = 4.10 , p < .01 , and t ( 16 ) = 2.82 , p < .05 , respectively . <p> We found no significant differences between the two conditions in the confidence with which they reported the accuracy of their estimate of the vehicle 's speed , t(31) = 0.19 , ns . Similarly , we found no significant differences between the two conditions on ratings of " How dangerous was the way that the vehicle was being driven ? " t(30) = 0.41 , ns , " How likely was a crash ? " t(31) = 0.87 , ns , and " How likely was the driver to kill someone ? " t(31) = 0.69 , ns . Discussion <p> The results of this experiment contradicted our hypothesis that a belief that a vehicle crashed would increase witnesses ' perceptions of a vehicle 's speed . In fact , the results showed that participants in both conditions significantly underestimated the speed of the vehicle . The results also demonstrated that knowledge concerning the outcome of an event need not necessarily @ @ @ @ @ @ @ @ @ @ being driven . In this case , participants who believed that the car had subsequently crashed rated the vehicle as being driven as safely as participants who believed that the car had not been involved in an accident . We return to these issues in the general discussion . <p> However , these findings do not necessarily mean that knowledge that a vehicle crashed has no influence on eyewitness estimates . For instance , anecdotal evidence from police accident investigators suggests that witnesses grossly overestimate the speeds of police vehicles when flashing lights and sirens are used and a crash occurs . This seems plausible because , in most instances in real life or in films and television , flashing lights and wailing sirens are used when a vehicle is traveling at high speed . Consequently , the fact that flashing lights and sirens are so firmly associated with vehicles traveling at high speed , combined with the belief that a vehicle crashed , may cause eyewitnesses to overestimate a crashed vehicle 's speed and rate more negatively the way the vehicle was driven ( Ayers &; Reder , 1998 ) @ @ @ @ @ @ @ @ @ @ determine whether the belief that a police vehicle crashed while using flashing lights and sirens increases mock witnesses ' speed estimates . The object of this experiment was to manipulate the knowledge that a police vehicle crashed . We hypothesized that witnesses who believed the police vehicle had crashed would perceive the speed of that vehicle and other estimates of dangerousness to be higher than those who did not believe the vehicle had crashed . Method <p> Participants <p> Participants were 41 prospective undergraduate students ( 6 men and 35 women ) whose average age was 26.37 ( range 17-69 , SD = 15.09 ) . <p> Materials <p> We constructed a video identical to the previous one except that the vehicle used was a police model of the civilian vehicle and was shown with lights flashing and sirens sounding . Again , it was driven at a constant 49.7 mph ( 80.0 kph ) . We obtained from police files a photograph of a crashed car and a not-crashed car identical to the one shown in the video in terms of make , model , and color . For consistency @ @ @ @ @ @ @ @ @ @ photographs in both experiments . The crashed vehicle was in fact a police car ; however , in Experiment 1 the photographs were processed using Adobe Photoshop to remove all visible police markings . Thus , the damage to the police vehicle ( Experiment 2 ) and civilian vehicle ( Experiment 1 ) in the crashed conditions was identical , eliminating a potential confound between experiments . <p> Procedure <p> We used the same method as in the previous study with the exception that the vehicle used in the film was a marked police car with flashing lights and sirens . Thus , in one condition ( police car not crashed ) , participants watched a film in which a police car with flashing lights and a siren drove down a street . They were then shown a picture of a police car not crashed . In the second condition ( police car crashed ) participants watched the same video but were subsequently shown a picture of a crashed police vehicle . Results <p> Table 2 shows the means and standard deviations for mock-witness estimates of vehicle speed . Participants in @ @ @ @ @ @ @ @ @ @ in the not-crashed condition , t(39) = 2.89 , p < .01 . Also , whereas participants in the not-crashed group were , on average , very accurate in estimating the speed of the vehicle , t ( 16 ) = 0.41 , ns , the crashed group significantly overestimated the speed of the vehicle , t(23) = 4.58 , p < .001 . However , despite the fact that there were clear differences in the accuracy of their speed estimates , no significant differences were found between the two conditions for witness confidence in the accuracy of these estimates , t(39) = 0.55 , ns . <p> Considering participants ' perceptions of how dangerously the police vehicle was being driven , those in the crashed condition did not rate the vehicle as being driven more dangerously than those in the not-crashed condition , t(39) = 1.46 , ns . However , drivers were rated as being significantly " more likely " to crash , t(39) = 4.09 , p < .001 , and significantly " more likely to kill someone " by those in the crashed condition than by those @ @ @ @ @ @ @ @ @ @ < .001 . Discussion <p> In this experiment , the belief that a police vehicle with flashing lights and sirens had crashed caused witnesses in the crashed condition to overestimate the vehicle 's speed , estimating the vehicle to be traveling significantly faster than participants in the not-crashed condition did . Whereas participants in the not-crashed condition of this experiment were accurate in their estimations of the speed of the police vehicle , those in the crashed condition significantly overestimated the speed of the car . Nevertheless , those in the crashed condition were just as confident in the accuracy of their estimate as those in the not-crashed condition . In addition , participants in the crashed condition of this experiment reported that they thought the car was more likely to be involved in a crash and to cause a fatality than did participants in the not-crashed condition . GENERAL DISCUSSION <p> The results of the two experiments were mixed . The hypothesized effect of knowledge that a vehicle crashed causes witnesses to overestimate vehicle speed was not supported in Experiment 1 . This result suggests that for accidents involving civilian @ @ @ @ @ @ @ @ @ @ vehicle simply because they know it crashed . However , we found a significant effect in Experiment 2 . The obvious , and only , difference between the two experiments is that in Experiment 2 we used a police vehicle with flashing lights and sirens . One possible explanation of this result may be that witnesses in the crashed police car condition are aware that police drivers are highly trained and therefore unlikely to crash under normal conditions . Given this knowledge , mock witnesses may have reasoned that for a highly trained police driver to crash he or she must have been traveling at high speed . <p> Alternatively , or additionally , because witnesses are likely to associate police lights and sirens with high-speed pursuits , they may overestimate speed . Thus , the overestimate of speed is explicable within an activation-based framework in terms of source of activation confusion ( Anderson &; Bower , 1973 ; Reder &; Schunn , 1996 ) . The concept of " high speed " is activated by the knowledge that a vehicle crashed and the use of flashing lights and sirens . @ @ @ @ @ @ @ @ @ @ vehicle , the activation of this concept influences their memory for the speed of the vehicle and they give high estimates . Unfortunately , this is likely to be an unconscious process and witnesses can be unaware that this has occurred ( Kelley &; Jacoby , 1996 ) . <p> This latter point is supported by the confidence judgments . In the crashed police car condition , whereas mock witnesses significantly overestimated the speed of the police vehicle , they were no less confident in this condition than in the uncrashed police vehicle condition . This is an important finding because confidence is often used to infer accuracy ( Fox &; Walters , 1986 ; Leippe , Manion , &; Romanczyk , 1992 ) . For instance a witness who says , " I 'm absolutely certain the car was doing 100 miles per hour " is more likely to be believed that one who says , " I thought the car was doing 60 miles an hour but I 'm not sure . " The practical implication of this finding is that confidence in the speed of a vehicle can @ @ @ @ @ @ @ @ @ @ , if a number of witnesses provide estimates of a vehicle 's speed , an investigator should be careful about simply relying on the most confident witness . <p> Returning to the issue of speed estimates , mock witnesses in both conditions of Experiment 1 underestimated the speed of the civilian vehicle , whereas those in the uncrashed condition in Experiment 2 , with the police car using lights and sirens , were accurate , and the witnesses in the crashed condition overestimated the speed of the vehicle . Although accuracy was not the focus of this experiment and the fact that the stimuli was presented via videotape rather than witnessed in real life reduces our ability to generalize , these findings suggest that witnesses ' speed estimates can be influenced by factors other than the actual speed of the vehicle , can vary greatly , and should be treated with caution . <p> In Experiment 1 no difference was found between " How likely was a crash ? " and " How likely was the driver to kill someone ? " The absence of significant effects in Experiment 1 shows @ @ @ @ @ @ @ @ @ @ to crash and likely to kill someone because they believed the vehicle did crash later . However , in Experiment 2 significant differences were found between the two conditions . Participants in the crashed condition rated the driver as being more likely to crash and more likely to kill someone . This shows that witnesses ' reports concerning the way a vehicle was driven can be influenced by flashing lights and sirens and the belief that a vehicle crashed . Again , reasons for these reports are likely to be source of activation confusion . <p> Nevertheless , no significant effect was found between the crashed and uncrashed conditions concerning remembered dangerousness . Even flashing lights and sirens and belief that a vehicle crashed did not cause remembered dangerousness to increase . However , in our film the vehicle is simply driven down a straight road at a constant speed . Perhaps if the vehicle had been driven in a more ambiguous manner , for instance , if the vehicle had to swerve to avoid a pedestrian , then witnesses ' estimates of the perceived dangerousness of the way that the @ @ @ @ @ @ @ @ @ @ offer the potential for considerable future work . Research into the use of heuristics in recall is relevant here . This work shows heuristics are particularly likely to be invoked when ( a ) there is little time to engage in systematic analysis , ( b ) individuals are overloaded with information , and/or ( c ) there is little available information ( Macrae , Hewstone , &; Griffiths , 1993 ) . Potentially , police cars with flashing lights at night are more likely to command attention , and other cues are unavailable to determine speed . Hence , the distortions we have found here with police vehicles may be exacerbated by nightfall . <p> Similarly , eyewitnesses ' memory of the driving of vehicles other than police vehicles may be influenced by reconstructive memory processes and a belief that the vehicle crashed . For instance , many people have stereotypes concerning the driving of New York taxis that may mean blame is more likely to be attributed to the taxi driver in the event of a crash than to another driver involved in the crash . These issues demand @ @ @ @ @ @ @ @ @ @ have shown that belief that a vehicle crashed has no influence on witnesses ' reports of the way a civilian car is driven , but when a police car with flashing lights and sirens is supposedly involved in a crash , witnesses are likely to overestimate the speed , the likelihood of someone being killed , and the likelihood of a crash . The implication is that anecdotal reports from police officers who have been involved in accidents when they had lights and sirens on may well be correct . The lights and sirens and knowledge that a crash occurred may lead witnesses to overestimate speed . Of course , it should be kept in mind that in real cases police officers may minimize their speed to prevent disciplinary measures . Nevertheless , these results suggest that witnesses ' descriptions of crashes involving police vehicles using flashing lights and sirens should be treated with caution . <p> The authors thank the British Academy for the award of a postdoctoral fellowship to the first author that partly supported this work . We would also like to thank Edith Cowan University for the @ @ @ @ @ @ @ @ @ @ the Western Australian Police Service for their help in constructing the materials used for this experiment . <p> Address correspondence to Mark R. Kebbell , School of Psychology , James Cook University , Townsville , 4810 , Queensland , Australia ; mark.kebbell@jcu.edu.au ( e-mail ) . TABLE 1 Participants ' Estimates of Speed , Confidence , and Other Factors Concerning the Dangerousness of the Way a Civilian Vehicle Was Being Driven PREFORMATTED TABLE TABLE 2 Participants ' Estimates of Speed , Confidence , and Other Factors Concerning the Dangerousness of the Way a Police Vehicle Was Being Driven PREFORMATTED TABLE <p> 
##4001270 ABSTRACT . Longing can be defined as a blend of the primary emotions of happiness and sadness . These primary emotions are experienced very early by children , and the meanings of the words happiness and sadness are also known by children early in their verbal development . To find out at what age children are able to understand and use the more developed concept of longing , the authors interviewed 74 preschoolers ( 4- and 5-year-olds ) in Norway and Sweden about their experiences of longing . Chi-square analyses showed age and sex differences in knowledge of the concept , and some differences between categories of longing were also significant . Results showed that young children 's ability to understand and use the concept of longing appears to be limited and that girls seem to mature earlier in this respect than boys do . <p> Key words : children , emotional development , emotions , longing <p> LONGING has usually been defined as a blend or a mix of the primary emotions of love and sadness , and it is also considered an emotion of second-or higher-order @ @ @ @ @ @ @ @ @ @ , 1999 ) . It is also an emotion with a character of its own . In a study of emotion families , Shaver , Schwartz , Kirson , and O'Connor ( 1987 ) found longing to be the only emotion in one of the subclusters . The experiencing of mixed or opposite emotions comes relatively late in a child 's development . By the age of 9 , children acknowledge that one situation can provoke two opposite emotions , whereas younger children usually do not admit that more than one emotion at a time can be experienced ( Harris , 1993 ; Whitesell &; Harter , 1989 ) . <p> Age effects in training children to acknowledge mixed emotions were reported by Peng , Johnson , Pollock , and Glasspool ( 1992 ) , and 6- to 7-year-olds were found to benefit more from such training than 4- to 5-year-olds . Holm , Classon , Greaker , Karlsson , and Strmberg ( 2000 ) found that 6-year-old children have experienced several dimensions of longing . Because 4- and 5-year-olds can use the word longing , we wanted to find out @ @ @ @ @ @ @ @ @ @ light on the development of longing as a mixed emotion . In this study we wanted to discover whether 4- and 5-year-old children can understand , use , and describe the concept of longing and relate it to their own experiences . In studies of children 's experiences of the concept of longing , the term love has been replaced by the more appropriate term happiness ( Holm , 2001 ; Holm et al. , 2000 ) , and that is also the case in this study . Method <p> Participants and Procedure <p> Do 4- and 5-year-old children know what longing is ? Have they experienced it themselves ? How do they feel when they are longing ? What do they do ? Are they longing for something when they are at home ? In preschool ? Do they talk about their longing with someone ? These were questions that we asked 74 Norwegian and Swedish 4- and 5-year-old preschoolers in an attempt to find out if and how children in these age groups experience this particular emotion . Twenty-two Norwegian children ( 15 girls and 7 boys ) and @ @ @ @ @ @ @ @ @ @ were interviewed individually during ordinary days at their preschools . Thirty-one of the children were 4 years old , and 43 were 5 years old . <p> Thirty-two children listened to a story about a boy who was longing for his playmate ( a girl ) and talked a little about the story before the interview questions were posed ; 22 listened to the story afterwards ; and the story was not read at all to 20 of them . The interviews were written out and analyzed by all three authors , independent of each other , before the results were summarized . Results and Discussion <p> No effect of the story conditions " before " versus " after " versus " no story " were found . No nationality differences were found in any of the answers . Some children did not answer all the questions , and some answers had nothing to do with the question posed . The reasons for not answering all questions were mostly lack of interest ( wanting to go back to some activity ) or lack of understanding . <p> To find out if @ @ @ @ @ @ @ @ @ @ longing , we posed the question " Do you know what longing is ? " The answer was no from 38 children and yes from 28 . Twenty of these latter children said that it is longing for someone . An age and sex difference could be seen , namely , that the yes answers came from the 5-year-old girls from both countries in all but 5 cases . A chi-square analysis of the yes and no answers from the 5-year-old girls versus the rest of the sample , ? ( 1 , N = 66 ) = 22.05 , p < .001 , showed that the 5-year-old girls ' answers differed significantly from those of the rest of the children who answered the question . <p> This result might be expected because girls often are verbally and emotionally more mature than boys ( Brody &; Hall , 1993 ) . Girls may be more observant of their own feelings , something that can be seen in results regarding sex differences in homesickness ( Zimmerman &; Bijur , 1995 ) . Some age differences in the depth of understanding of emotions @ @ @ @ @ @ @ @ @ @ , Garrattoni , and Venturini ( 2000 ) and by Lagattuta , Wellman , and Flavell ( 1997 ) . <p> Sixteen of the children answered no to the question " Have you ever been longing ? " All of the remaining 58 who answered yes also told the interviewers what they had been longing for . Earlier research has shown that children usually describe emotional states by describing what induces it ( Cartron-Guerin &; Reveillaut , 1980 ) . Twenty-seven had been longing for family members , 8 for playmates , 11 for things , 7 for occasions , and 5 for animals . Of the whole sample , 47% had been longing for persons and 31% for nonpersons . <p> A chi-square analysis of differences among the five different categories of longing showed that longing for family members was significantly more common than longing for something else , ? ( 4 , N = 58 ) = 27.16 , p < .001 . The pattern for yes and no answers on this question was different from the answers to the first question . Most children obviously had been longing @ @ @ @ @ @ @ @ @ @ not know what longing was . <p> " How do you feel when you are longing ? " was answered with " I do n't know " by 21 of the children . Another 21 mentioned something negative like " not good , " " a little sad , " " no fun , " and " dreary . " Twelve mentioned something positive like " good " and " happy . " The difference between these groups was not significant . Five children had experienced both positive and negative feelings ( e.g. , " I am a little happy in the beginning , but then I become sad " and " When I am longing for my uncle I am happy , but not when I am homesick " ) . <p> On follow-up questions , 13 children answered that longing could be both happy and sad . Ten of them were girls ; 8 were 5-year-olds and 2 were 4-year-olds . This indicates an age-and sex-dependent growing awareness of the dual nature of longing ( see , e.g. , Holm , 1999 ) . Thirty-seven children ( half the @ @ @ @ @ @ @ @ @ @ one sad , one neutral , and one happy -- to help them decide what longing feels like . Fifteen children chose the happy face , and 15 chose the sad face . Seven chose the neutral face . Three children pointed to two or three faces , and they were excluded from the analysis . The age and sex differences among the three groups -- happy , sad , and neutral -- were not significant . <p> The answers to the question " What do you do when you are longing ? " showed that 38 children were active , most of them playing . Only 5 were passive , just waiting and doing nothing . Nine did not know or did not answer . A chi-square analysis showed that the difference among the groups was significant , ? ( 2 , N = 52 ) = 37.42 , p < .001 . Activity as a means of distraction from the emotion of homesickness has been reported by Thurber and Weisz ( 1997 ) and as a distraction from the emotion of longing by Holm et al . ( 2000 @ @ @ @ @ @ @ @ @ @ Are you longing for something when you are at home ? " Six were longing for parents and relatives , 9 for playmates , 9 for playing and toys , 2 for animals , whereas 15 of then said " Nothing " or " I do n't know . " The difference was significant , ? ( 4 , N = 41 ) = 11.08 , p < .05 . In answer to the question " Are you longing for something when you are at preschool ? " 6 children said they longed for parents and relatives , 7 longed for playmates , 4 wanted to play , 5 longed for home , and 17 responded that they longed for nothing or said " do n't know . " The difference was significant , ? ( 4 , N = 39 ) = 14.21 , p < .01 . A relatively large proportion of the children did not seem to be longing in these contexts or were not aware of such feelings . <p> Twenty-six children were also asked , " Do you talk to someone when you are longing ? @ @ @ @ @ @ @ @ @ @ ; 7 said playmates ; 6 said brothers , sisters , and relatives ; 1 said my pet and myself ; 2 children did not know . Almost all the children talked to some person , and there was a significant difference among the categories , ? ( 4 , N = 26 ) = 10.53 , p < .05 . <p> One conclusion we reached from the results of this study is that it may not be fruitful to interview such young children , as we did here , because sex-and age-limitations to the understanding of longing exist in these young children . The levels of verbal , cognitive , and emotional development in these age groups make the proportion of potential candidates for direct interviewing limited . Several of the children in this study answered only a few of the planned questions , and that limited the value of the statistical analyses . Perhaps interviews that take place in some other context than in the preschool may extract more answers and yield better results . <p> We did not find any differences between the countries , probably because these @ @ @ @ @ @ @ @ @ @ in countries that differ considerably may show other results ( e.g. , Jolley , Zhi , &; Thomas , 1998 ) . We did find an age-and sex-related result . The girls ' answers often were longer than the answers given by the boys . A comparison of the answers to the first two questions indicated that most children have experienced longing , but not all children seem to be able to use the concept or describe it verbally . <p> Part of this research was made possible by a grant from the research program The Reflecting Citizen , School of Education and Communication , Jnkping University . <p> Address correspondence to Olle Holm , School of Education and Communication , Jnkping University , Box 1026 , 551 11 Jnkping , Sweden ; olle.holm@hlk.hj.se ( e-mail ) . <p> 
##4001273 ABSTRACT . The authors examined how patriarchy , sexism , and gender influence Turkish college students ' attitudes toward women managers . Turkish undergraduate students ( N = 183 ) from Middle East Technical University completed questionnaires measuring attitudes toward women managers as well as attitudes toward the concepts of hostile and benevolent sexism and support for patriarchy . Participants were of upper-or middle-class Turkish backgrounds . The results showed that male participants exhibited less positive attitudes toward women managers than did female participants . In addition , participants who held more favorable attitudes toward patriarchy and who scored high on hostile sexism also held less positive attitudes toward women managers than those who held less favorable attitudes toward patriarchy and who scored low on hostile sexism . A regression analysis showed that support for patriarchy and hostile sexism was more important for explaining less favorable attitudes toward women managers than was benevolent sexism . <p> Key words : attitudes , benevolent sexism , gender differences , gender stereotypes , hostile sexism , patriarchy , women managers <p> PERCEPTIONS OF AND ATTITUDES TOWARD WOMEN as managers have received @ @ @ @ @ @ @ @ @ @ Dubno , 1985 ; Schein , 1975 ) and in other countries such as Nigeria ( Adeyemi-Bello &; Tomkiewicz , 1996 ) . Our purpose in the present study was to explore the issues of attitudes toward women in top managerial positions in Turkey . We examined how gender differences , patriachy , and hostile and benevolent sexism influence attitudes toward women managers in a sample of Turkish college students . <p> Gender stereotypes are responsible for workplace discrimination and for negative attitudes toward women as managers ( Eagly &; Mladinic , 1994 ; Heilman , 1995 ) . According to traditional Turkish gender roles , men are dominant , independent , competitive , and capable of leadership , and women are submissive , dependent , caring , and good at domestic tasks and child rearing ( Geis , 1993 ) . Gender stereotypes in a society are often responsible for prejudice and discrimination against women . Thus , the idea of patriarchy , which refers to a structural control by men over political , legal , economic , and religious institutions ( Goldberg , 1993 , cited in Glick &; @ @ @ @ @ @ @ @ @ @ and to behave in a discriminatory manner toward women in many areas of their lives . Johnson ( 1995 ) even suggested that the product of the patriarchal traditions of men 's right to control their women can be called patriarchal terrorism and involves systematic violence , economic subordination , threats , and other control tactics . <p> Sexism , an important social-psychological concept relevant to patriarchy , is created by patriarchy and may be an important factor in the exclusion of women from the workplace . Sexism is a negative attitude or discriminatory behavior based on the presumed inferiority or difference of women as a group ( Cameron , 1977 ) . As social psychologists argued , persistent sexism and patriarchy encourage a belief in the superiority of men and promote the idea that men deserve to control and receive more resources than women control or receive . In patriarchal and sexist cultures , the women are assigned to domestic roles , whereas the men are associated with career , authority , and leadership ( Eagly &; Mladinic , 1994 ) . <p> Glick and Fiske ( 1997 ) argued @ @ @ @ @ @ @ @ @ @ is a product of the patriarchal tradition combined with men 's dependence on women for sexual reproduction . Modern sexists might reject any sort of discrimination toward women , but they might have ambivalent affection for women . A woman can be seen as a saint or a strumpet . These two views create ambivalent sexism , termed hostile and benevolent sexism . <p> Glick and Fiske ( 1996 ) defined hostile sexism as traditional prejudicial attitudes toward women ( e.g. , anger , resentment , antipathy ) , characterizing women as inferior , and justifying male power and traditional gender roles . Hostile sexism can be accepted as a subtle way of measuring support for male dominance , competitive gender differentiation ( devaluing women ) , and hostile sexuality ( e.g. , anger at women for using sexual attraction to manipulate men ) . <p> Benevolent sexism " is a set of interrelated attitudes toward women that are sexist in terms of viewing women stereotypically and in restricted roles but that are subjectively positive in tone and also tend to elicit behaviors typically categorized as pro-social or intimacy-seeking " ( @ @ @ @ @ @ @ @ @ @ Protective paternalism ( protecting or helping women ) , complementary gender differentiation ( favoring women ) , and heterosexual intimacy ( feelings of personal need , intense affection for and admiration of women ) are the three underlying sources of benevolent sexism . A cross-cultural study on hostile and benevolent sexism in 19 nations by Glick et al . ( 2000 ) showed that in the countries that were highest in sexism ( Cuba , Nigeria , Turkey ) , individuals tended to reject hostile sexism but were less likely to reject benevolent sexism , whereas in countries that were more egalitarian ( England , Australia , the United States ) , individuals tended to reject both hostile and benevolent sexism . <p> In short , hostile and benevolent sexism might influence individuals ' attitudes toward women managers differently in Turkey . Glick , Dibold , Bailey-Werner , and Zhu ( 1997 ) found that men 's hostile sexism scores were correlated with negative evaluations of career-oriented women , whereas benevolent sexism scores correlated with favorable evaluations of homemakers . <p> Patriarchy , Sexism , and Women Managers in Turkey <p> @ @ @ @ @ @ @ @ @ @ role differences . The institutions of marriage and the family are highly patriarchal ( Kandiyoti , 1995 ) . Men are the dominant sex and are expected to control the home . The husband is culturally accepted as the ruler of the family and is regarded as the formal authority to whom the wife and children must ultimately respond . The husband 's role is authoritarian , and he assumes responsibility for maintaining the family structure by whatever means he feels are justified . The wife 's role is taking care of the family and remaining dependent on her husband to protect the family structure . In short , the Turkish people still generally value patriarchy ( e.g. , Kagitibasi , 1981 ) . <p> Sakalli ( 2002 ) demonstrated that sexism was an important factor in understanding prejudiced attitudes toward stigmatized groups , suggesting that sexist Turkish individuals were more likely to hold negative attitudes toward stigmatized groups such as homosexuals . In terms of an ambivalent sexism model , Sakalli showed that the concept of hostile sexism was an especially important influence on negative attitudes held toward stigmatized groups @ @ @ @ @ @ @ @ @ @ home in paid jobs since the 1950s ( zbay , 1995 ) , but their jobs have been mainly in the areas of education , health , and secretarial work , relevant to the gender role stereotype . Women also have had opportunities to work in middle and top managerial positions . According to the General Department of Women 's Status and Problems ( 1996 ) , the percentages of middle to top management positions , respectively , held by women were 80% for a chief position , 15% for a division director position , 3.7% for a department head , and 0.12% for a general director position . Kabasakal , Bocacigiller , and Erden ( 1994 ) found that the percentage of women employees in 64 organizations was 43% , with 26% in middle management and 3% in top management . Overall , consistent with other countries , all these numbers suggest that Turkish men are still seen as " first choice " for managerial positions . <p> There have been some studies on perceptions of women managers in Turkey . ifti ( 1979 , cited in General Department of @ @ @ @ @ @ @ @ @ @ Turkish men were seen as more devoted to their work , more reliable , more precise , and harder working than women managers . Women employees were seen as more compliant than were men employees . Some research in Turkey has shown that women 's lower representation in managerial positions has resulted from the definition of work based on gender , the organizational culture that has created a barrier for women 's advancement in their career ( Atabek , 1994 ) , and patriarchal stereotypes ( Berberoglu &; Mavis , 1990 , cited in General Department of Women 's Status and Problems , 1996 ) . <p> Consequently , all these studies have indicated that the managerial position is defined in terms of the masculine stereotype and therefore is seen as more suitable for men . This situation creates a barrier for women who want to advance . Therefore , attitudes toward women as managers need to be studied in detail , as it appears that there has not been any research conducted on attitudes toward women managers in Turkey . <p> The Purpose of the Present Study <p> In the @ @ @ @ @ @ @ @ @ @ a sample of Turkish college students . We focused on students because young Turkish students are tomorrow 's employees and managers . We hoped they would provide us with information about future conceptualizations of women managers in Turkey and about future behavior toward women managers , because attitudes influence behavior ( Ajzen &; Fishbein , 1977 ) . In addition , and different from earlier studies in the United States and other countries , we explored how hostile and benevolent sexism may influence attitudes toward women managers . On the basis of results of earlier studies on attitudes toward women managers and leaders , patriarchy , and hostile-benevolent sexism , we formed the following predictions : Consistent with earlier research suggesting that women have more positive attitudes toward women managers than do men ( Adeyemi-Bello &; Tomkiewicz , 1996 ; Heilman , Block , Martell , &; Simon , 1989 ; Schein , Mueller , &; Jacobson , 1989 ) , we expected Turkish women to show more positive attitudes toward women managers than Turkish men . We expected the participants who scored high on support for patriarchy to have more @ @ @ @ @ @ @ @ @ @ on support for patriarchy . Similarly , because hostile sexism can be accepted as a subtle form of patriarchy ( Glick &; Fiske , 1997 ) , we expected participants who scored high on hostile sexism to have more negative attitudes toward women managers than those who scored low on hostile sexism . We expected that patriarchy and hostile sexism would explain the negative attitudes toward women managers more than benevolent sexism would . Method <p> Participants and Procedure <p> The participants were 183 ( 92 men , 91 women ) Turkish undergraduate students from the Middle East Technical University who attended elective courses in general psychology and an Introduction to Social Psychology . Data for 4 participants were excluded because their questionnaires were incomplete . Mean ages were 21.28 years ( SD = 1.75 ) for the men and 20.67 years ( SD = 2.10 ) for the women . The participants were of the upper or middle class . They participated voluntarily and were given extra credit for completing the questionnaires . <p> Materials <p> To measure attitudes toward women as managers , we developed a new 18-item questionnaire @ @ @ @ @ @ @ @ @ @ was appropriate for Turkish culture and was also timely . All the items were rated on a 6-point scale ranging from 1 to 6 . Higher scores reflected positive attitudes toward women managers . <p> Factor analysis showed that the data explained 70% of the total variance . The first factor measured participants ' general approval and trust of women managers ( eigenvalue = 11.48 ) and accounted for 63.82% of total variance . A few sample items were " I support women to be managers " ; " I would not like to be directed by a women manager at my workplace " ; and " I trust women managers . " <p> The second factor measured participants ' attitudes toward the methods women managers use to reach management levels ( eigenvalue = 1.14 ) and accounted for 6.34% of the total variance . Sample items were " Women use their attractiveness to be promoted in their job " and " There are many tactics that women use to move up in their career . " Cronbach alphas for the whole scale , for the first , and for the @ @ @ @ @ @ @ @ @ @ respectively . We performed a test-retest analysis , which showed a high correlation between the first administration of the scale and the re-administration of it ( r = .90 ) . <p> To illustrate the convergent validity of the newly constructed scale , participants also completed the Women as Managers Scale ( WAMS ; Peters , Terborg , &; Taylor , 1974 ) . WAMS results were highly correlated with results on our scale ( r = .85 , p < .001 ) , suggesting that our scale is able to measure constructs similar to those measured by the WAMS . <p> We used the Ambivalent Sexism Inventory ( Glick &; Fiske , 1996 ) to measure sexism . The scale includes two main dimensions : hostile sexism ( e.g. , " Feminists are making entirely reasonable demands of men " ; " Most women interpret innocent remarks or acts as being sexist " ) and benevolent sexism ( e.g. , " Men are complete without women " ; " A good woman should be set on a pedestal by her man " ) . Cronbach alphas for the whole scale @ @ @ @ @ @ @ @ @ @ , and .78 , respectively . Participants indicated agreement with each item on a scale ranging from 1 ( indicating nonsexist ) to 6 ( indicating sexist attitudes toward women ) . <p> We developed a 4-item scale ( a = .85 ) to measure patriarchy . The items were " Women 's careers should not come before marriage and family " ( a = .88 ) ; " Family structure is influenced negatively if the woman works " ( a = .85 ) ; " Women should not find a job and work before taking consent of their husbands " ( a = .84 ) ; and " The most important responsibility of women is to look after their children and to take care of the home " ( a = .75 ) . <p> The scale explained 69.61% of the total variance . Participants indicated agreement with each item on a scale from 1 ( indicating low support for patriarchy ) to 6 ( indicating high support for patriarchy ) . The patriarchy data were used as an independent variable . The participants who scored lower than the median @ @ @ @ @ @ @ @ @ @ patriarchy ; the participants who scored above the median were considered to have high support for patriarchy . Results <p> We calculated the correlations between variables in the study , patriarchy , hostile sexism , benevolent sexism , and attitudes toward women managers . Hostile sexism ( r = -.64 , p < .01 ) , and patriarchy ( r = -.68 , p < .01 ) were significantly correlated with attitudes toward women managers . However , benevolent sexism was not significantly correlated with attitudes toward women managers ( r = -.11 , ns ) . <p> We used median splits so that patriarchy and hostile sexism could be treated as independent variables . The participants who scored lower than 2.50 in support of patriarchy or 3.5455 in support of hostile sexism were accepted as low in support of patriarchy and hostile sexism . Those who scored higher than the aforementioned medians were accepted as high in support of patriarchy and hostile sexism . We then performed a three-way multiple analysis of covariance ( MANCOVA ) , using benevolent sexism as the covariant , to examine , by gender , @ @ @ @ @ @ @ @ @ @ ; Mdn = 2.50 ) and hostile sexism ( low vs. high ; Mdn = 3.5455 ) influenced attitudes toward women managers . The results of the MANCOVA showed that the main effects of patriarchy , F ( 1,161 ) = 19.36 , p < .05 , ? = .10 ; hostile sexism , F ( 1,161 ) = 16.37 , p < .05 , ? = .09 ; and gender , F ( 1,161 ) = 13.86 , p < .05 , ? = .07 , were statistically significant . <p> The participants who scored high in support of patriarchy held less positive attitudes toward women managers ( M = 3.87 , SD = 0.96 ) than did others who scored low in support of patriarchy ( M = 5.08 , SD = 0.74 ) . Participants who scored high in hostile sexism also held less positive attitudes toward women managers ( M = 3.90 , SD = 1.00 ) than did those who scored low in hostile sexism ( M = 5.09 , SD = 0.70 ) . In addition , the men held less favorable attitudes toward women @ @ @ @ @ @ @ @ @ @ than did the women ( M = 5.10 , SD = 0.71 ) . See Table 1 for the complete statistics . <p> To examine the third hypothesis , we performed a multiple regression analysis . The regression analysis showed that , taken together , support for patriarchy , hostile sexism , and benevolent sexism accounted for 58% of the variance in attitudes toward women managers , F ( 3,166 ) = 78.79 , p < .01 . As expected , support for patriarchy and hostile sexism was more important for explaining less favorable attitudes toward women managers than was benevolent sexism ( see Table 2 ) . Discussion <p> The purpose of the present study was to explore attitudes of Turkish college students toward women managers . We explored the effects of gender differences , patriarchy , and hostile and benevolent sexism . Consistent with earlier studies in other countries ( e.g. , Dubno , 1985 ; Farmer &; Waugh , 1999 ) , male participants held less positive attitudes toward women managers than did female participants , perhaps because of the women 's psychological identification with their own career @ @ @ @ @ @ @ @ @ @ other researchers , women are less sexist than men ( Glick &; Fiske , 1996 ; Swim , Aikin , Hall , &; Hunter , 1995 ) and show more tolerance toward stigmatized groups such as homosexuals ( Sakalli , 2002 ) . Consequently , it was not suprising that the women held more positive attitudes toward women managers than did the men . Thus , our first hypothesis was supported . <p> Our second hypothesis was also supported . Participants who had high scores on support for patriarchy held less positive attitudes toward women managers than did those with low scores on that issue . And participants who scored high on hostile sexism held less positive attitudes toward women managers than did those who scored low on hostile sexism . As expected , our regression analysis showed that support for patriarchy and hostile sexism were more important for explaining less favorable attitudes toward women managers than was benevolent sexism . The result was consistent with the findings of Glick et al . ( 1997 ) that hostile sexism , but not benevolent sexism , predicted negative attitudes toward career-oriented women @ @ @ @ @ @ @ @ @ @ power , the economic subordination of women , and traditional gender roles ( Glick &; Fiske , 1997 ) , participants who accepted patriarchal ideology were naturally against women working in managerial positions . <p> Benevolent sexism also represents sexist beliefs , but it includes behaviors typically categorized as prosocial or intimacy seeking ( Glick &; Fiske , 1996 ) . Because benevolent sexism covers protective paternalism ( protecting , helping women ) , complementary gender differentiation ( favoring women ) , and heterosexual intimacy ( feelings of personal need , intense affection for and admiration of women ) , the correlation between benevolent sexism and attitudes toward women managers was not significant in the present study ( r = -.11 ) . <p> As suggested by Glick et al . ( 2000 ) , individuals who live in sexist cultures such as Cuba , Nigeria , and Turkey might not recognize benevolent sexism as a form of sexism , even though they immediately recognize hostile sexism , and the result might be a nonsignificant relationship between benevolent sexism and attitudes toward women managers . However , people who live in @ @ @ @ @ @ @ @ @ @ are more likely to recognize benevolent sexism as a form of sexism and reject it along with hostile sexism ( Glick et al. , 2000 ) . <p> Consequently , there may be a significant correlation between sexism and attitudes toward women managers in other countries that are less sexist . In other words , studies conducted in less sexist countries might have different results from the present study ; benevolent sexism may be strongly relevant to negative attitudes toward women managers in less sexist countries . <p> In the future , researchers should focus on how hostile and benevolent sexism might influence attitudes toward women managers in other cultures . Such studies and their application to real life might help individuals , both women and men , to recognize the paradox of the concept that " women are wonderful , but still we ca n't hire or promote them . " Individuals might learn the distinction between hostile and benevolent sexism and the importance of benevolent sexism to gain acceptance ( rather than resistance ) in low-status groups and sexist cultures . <p> Address correspondence to Nuray Sakalli-Ugurlu , Psychology @ @ @ @ @ @ @ @ @ @ ( Orta Dogu Teknik niversitesi ) , 06531 , Ankara , Turkey ; nurays@metu.edu.tr ( e-mail ) . TABLE 1 Means and Standard Deviations for Attitudes Toward Women Managers PREFORMATTED TABLE TABLE 2 Regression Analysis Summary for Patriarchy , Hostile Sexism , and Benevolent Sexism Predicting Attitudes Toward Women Managers PREFORMATTED TABLE <p> 
##4001274 ABSTRACT . To investigate the relationship between worry and sense of humor , the author administered the Worry Domains Questionnaire ( WDQ ; F. Tallis , M. Eysenck , &; A. Matthews , 1992 ) and the Multidimensional Sense of Humor Scale ( MSHS ; J. A. Thorson &; F. C. Powell , 1993 ) to 140 undergraduates . Worry was negatively related to sense of humor . Results of a stepwise regression , using the WDQ as the criterion and the MSHS factors as predictors , showed that the humor production factor of the MSHS loaded negatively on Step 1 and the coping with humor factor loaded positively on Step 2 . The results are discussed in relation to W. E. Kelly and M. J. Miller 's ( 1999 ) cognitive model of worry and worry as a coping strategy . <p> Key words : anxiety , coping , humor , worry <p> WORRY , a sequence of negative , sometimes seemingly uncontrollable thoughts ( Borkovec , 1994 ) is a commonly experienced phenomenon . Indeed , Tallis , Davey , and Capuzzo ( 1994 ) found @ @ @ @ @ @ @ @ @ @ once every day . Despite the obvious transitory nature of worry , much of the recent research has approached this phenomenon as a trait , or a propensity to worry . For instance , commonly used measures of worry , such as the Worry Domains Questionnaire ( WDQ ; Tallis , Eysenck , &; Mathews , 1992 ) , have test-retest reliabilities that indicate they measure trait worry ( Tallis , Davey , &; Bond , 1994 ) . <p> Researchers investigating the characteristics of individuals who have a propensity to worry ( trait worriers ) have reported that worriers , when compared with nonworriers , tend to report more physical discomforts , obsessional symptoms , boredom , depression , stress , anxiety , perfectionism , and pessimism ( Borkovec , Robinson , Pruzinsky , &; DuPree , 1983 ; Chang , 2000 ; Jung , 1993 ; Kelly &; Markos , 2001 ; MacLeod , 1994 ; Starcevic , 1995 ; Tallis &; de Silva , 1992 ) . It is noteworthy that although worry is a sympton of the Generalized Anxiety Disorder ( GAD ; American Psychiatric Association , @ @ @ @ @ @ @ @ @ @ For instance , Davey ( 1993 ) reported that worry and anxiety account for separate variances in coping . Worry is more often associated with active cognitive coping , whereas anxiety is associated with avoidant coping , which indicates that worry and anxiety are separate constructs . <p> Kelly and Miller ( 1999 ) proposed a cognitive model of worry that is helpful for conceptualizing individual differences in worry . They suggested that trait worriers ' assumptions about themselves and the world in general are dysfunctional . More specifically , worriers have developed three primary underlying beliefs : ( a ) Everything is potentially dangerous ; ( b ) they should be able to find a perfect solution to every problem and/or avoid potential negative consequences ; and ( c ) they are inadequate or helpless to protect themselves or solve their problems . <p> Hence , worriers attempt to cognitively evaluate situations and their abilities to protect themselves in or from those situations . It follows that the combination of worriers ' beliefs that they should be able to solve their problems and their inadequacy to solve problems elicits a @ @ @ @ @ @ @ @ @ @ underlying beliefs about themselves and the world in general determine individuals ' information processing and characteristic ways of thinking and behaving ( i.e. , personality ) . <p> Perhaps because of the unpleasant traits and experiences associated with worry ( i.e. , anxiety , depression , pessimism ) and the feelings of inadequacy characteristic of many worriers ( Kelly &; Miller , 1999 ) , positive attributes have rarely been examined in relation to worry . One would expect that worry would be conversely related to positive traits , and this is often the case . For instance , Chang ( 2000 ) reported that worry is negatively related to life satisfaction . However , worry is not always negatively related to positive behaviors or traits . Kelly and Ashley ( 2001 ) found that worry was negatively related to excessive alcohol use . Individuals who were less likely to worry were more likely to engage in excessive alcohol consumption . <p> It appears that , despite the negative connotations of worry , it may serve some positive functions . Davey ( 1994 ) found that worry often functions as an @ @ @ @ @ @ @ @ @ @ positive which has not been examined in relation to trait worry is sense of humor . <p> Martin ( 1998 ) defined sense of humor as comprehension , appreciation , and creation of humor . Consistent with this general definition , Thorson and Powell ( 1993 ) suggested that a sense of humor consists of several elements , including recognition of oneself as humorous , recognition of others ' humor , appreciation of humor , a propensity to laugh , a perspective that allows an appreciation of life 's absurdities , and the use of humor to cope . Humor , according to McGhee ( 1979 ) , incorporates a general attitude of playfulness and a coinciding ability to play on ideas . The cognitive ability to manipulate and reframe ideas playfully enables individuals with a sense of humor to view unpleasant events as funny instead of frightening , annoying , or stressful . It is not surprising , therefore , that individuals with a greater sense of humor do not endorse depression , loneliness , negative mood , and social inadequacy ( Cetola , 1988 ; Moran &; Massam , @ @ @ @ @ @ @ @ @ @ Sarmany-Schuller , &; Hampes , 1997 ) and endorse task motivation , cheerfulness , self-esteem , trust , and social dominance ( Hampes , 1999 ; Kuiper , McKenzie , &; Belanger , 1995 ; Overholser , 1992 ; Ruch &; Carrell , 1998 ; Thorson &; Powell , 1996 ) . <p> Several researchers have documented the fact that a sense of humor moderates , or reduces , negative or unpleasant affect . For example , Yovetick , Dale , and Hudak ( 1990 ) found that individuals scoring higher on sense of humor report less anxiety than individuals scoring lower on sense of humor even when placed in an anxious situation . Newman and Stone ( 1996 ) reported that humor leads to decreased self-reported tension and decreased physiological activity associated with stress . Labott and Martin ( 1987 ) indicated that coping by using a sense of humor buffers the effects of negative events on mood states . Brown and Keegan ( 1999 ) found that humor also relieved boredom . Hence , possessing a sense of humor tends to moderate stress and anxiety and allows individuals to @ @ @ @ @ @ @ @ @ @ <p> Although there have been increases in both the humor and worry literatures , both of these topics continue to be relatively neglected by researchers . At the time of this writing , I found no published investigations of the relationship between worry and sense of humor . Therefore , my purpose in this study was to examine this relationship . On the basis of the previous literature establishing the fact that high worry and a low sense of humor have both been related to negative psychological outcomes and that individuals with a greater sense of humor are better able to thwart negative thinking and affect , I hypothesized that worry would have a significant and negative relationship to sense of humor . Method <p> Participants <p> Participants were 140 individuals ( 89 women , 51 men ) enrolled in undergraduate courses at a mid-sized , southwestern university . Totals do not always equal 140 , however , because some data were missing . The average age of the sample was 23.8 ( SD = 8.0 ) , ranging from 18 to 57 . There was no significant difference between the @ @ @ @ @ @ @ @ @ @ .82 . The majority of participants identified themselves as Caucasian ( 63.6% ) . Other ethnicities included African American ( 12.1% ) , Latino ( 7.9% ) , and Asian American ( 12.9% ) . Five ( 3.5% ) respondents did not identify their ethnicity . <p> Instruments <p> I assessed worry with the Worry Domains Questionnaire ( WDQ ; Tallis , Eysenck , &; Matthews , 1992 ) . The original WDQ included 30 items assessing six domains of worry . However , I found one domain ( social concerns ) to be highly related to social desirability . Hence , in this study I used the 25-item WDQ , which does not include the social concerns domain presented by Tallis , Davey , and Bond ( 1994 ) . The 25-item WDQ measures a tendency to worry across five general worry domains , including relationships , lack of confidence , aimless future , work , and financial . <p> The WDQ begins with the statement , " I worry ... " Sample items include " that others will not approve of me , " " that I can not @ @ @ @ @ @ @ @ @ @ I might make myself look stupid . " I presented the WDQ items with a 5-point Likert-type scale on which individuals described their experience as not at all to extremely in agreement with individual items . I summed the responses from the appropriate items to create the worry domains and a total WDQ score . Test-retest reliability ( r = .79 after 4 weeks ) , internal consistency ( a = .92 ) , and validity of the WDQ have been found to be satisfactory ( Tallis , Davey , &; Bond , 1994 ) . <p> I measured sense of humor using the Multidimensional Sense of Humor Scale ( MSHS ; Thorson &; Powell , 1993 ) . The MSHS is a 24-item questionnaire that assesses humor across four factors : Humor Production , Coping With Humor , Humor Appreciation , and Attitudes Toward Humor . Participants rate each item on a 5-point Likert-type scale ranging from strongly disagree to strongly agree . Sample items include " I can say things in such a way as to make people laugh , " " Coping by using humor is an elegant @ @ @ @ @ @ @ @ @ @ who generate humor . " The MSHS was developed through a series of factor analyses from a pool of 124 items . Thorson and Powell ( 1993 ) reported that the four factors remained stable across several large samples . They also reported a coefficient alpha of .92 for the full MSHS . <p> Procedure <p> After the participants had given informed consent , I administered the WDQ and MSHS in random order . The participants also completed a demographics survey that solicited information about age , gender , and ethnicity . They then completed the questionnaires in a group setting . Specific information regarding the nature of the study was not disclosed until participants returned the questionnaires . Results <p> Means , standard deviations , and internal consistencies of the WDQ , MSHS , and their respective subscales are presented in Table 1 . A simple regression was calculated using total WDQ scores as the predictor variable and total MSHS scores as the criterion . WDQ scores accounted for 3% of the variance in MSHS scores . This result was significant , F ( 1,138 ) = 4.37 , p @ @ @ @ @ @ @ @ @ @ that higher worry scores were related to lower MSHS scores ( = -.18 ) . <p> Because of the lack of previous research in this area , I used an exploratory approach to investigate the relationships among the WDQ domains , MSHS factors , and total scores for both scales . Because of its ability to investigate the " best fit " relationships among different variables ( Hair , Anderson , Tatham , &; Black , 1995 ) , I calculated a stepwise multiple regression to identify which types of worry were most related to sense of humor . I used the five WDQ domains as predictor variables and the total MSHS scores as the criterion . The relationships domain was the sole WDQ domain entered on a step accounting for 4% of the variance in total MSHS scores . There was a significant negative relationship between the relationships domain and worry , F ( 1,138 ) = 5.62 , p < .02 , = -.20 . <p> Next , to identify which aspects of sense of humor most related to worry , I calculated another stepwise multiple regression using the @ @ @ @ @ @ @ @ @ @ as the criterion . The Humor Production factor , entered in Step 1 , accounted for 9% of the variance , which was significant , F ( 1,138 ) = 13.59 , p < .0001 , and again negative ( = -.47 ) . The Coping With Humor factor , entered in Step 2 , accounted for an additional 5% of the variance . The increase in variance was significant , F ( 1,137 ) = 7.37 , p < .006 . The beta weights indicated that worry was positively related to coping with humor ( = .28 ) . <p> I calculated separate stepwise multiple regressions using the five WDQ domains as predictor variables and each MSHS factor as the criterion . For the humor production factor , the confidence domain of the WDQ was the only significant predictor , accounting for 10% of the variance in humor production scores , F ( 1,138 ) = 16.06 , p = .0001 , = -.32 . For each of the subsequent regressions , using the remaining MSHS factors as the criterion , I found no WDQ domains that accounted for significant @ @ @ @ @ @ @ @ @ @ results of this study supported the hypothesis . That is , worry has a significantly negative relationship to sense of humor . Thus , it seems that individuals with a sense of humor are less likely to worry . This finding is consistent with previous research in which lower scores on sense of humor were related to negative psychological outcomes ( Nezu , Nezu , &; Blissett , 1988 ; Overholser , 1992 ) and with the finding of Cann , Holt , and Calhoun ( 1999 ) that humor moderates the negative outcomes of stressful events . Russell and Davey ( 1993 ) found that worry appears to occur often as the result of stressful life events . Hence , a sense of humor might moderate the relationship between stress and worry . Researchers should explore this possible moderating effect . <p> The multiple regression results indicated that the Humor Production factor of the MSHS largely contributed to the negative relationship between worry and sense of humor . An understanding of the personality of worriers provides a partial explanation for this result . Because of the fears and inadequacies of @ @ @ @ @ @ @ @ @ @ model , one might assume that worriers would be less likely to participate in risky situations in which they feel unprotected or inadequate . <p> The hesitation of worriers to participate in risky situations appears to also include social situations . A commonly reported characteristic of worriers appears to be a fear of negative social evaluation . Pruzinsky and Borkovec ( 1990 ) found that worriers reported significantly more self-focused attention and social anxiety for publicly displayed aspects of themselves than did nonworriers . Thus , worriers would likely avoid placing themselves in risky social situations that might result in a negative evaluation by others . <p> The production of humor may appear to worriers to be a risky situation that could elicit negative evaluation by others . The relationship found between the confidence worries domain of the WDQ and humor production partially supports this possibility . If worries about confidence hinder individuals from producing humor , it might be expected that worriers would question their ability to produce humor in such a way that others would find favorable . Future research is necessary to determine whether a fear of negative @ @ @ @ @ @ @ @ @ @ A similar explanation might be made for the relationship between worries about relationships and humor . That is , individuals who worry about disrupting , or losing , relationships might be less inclined to use humor for fear that others will not approve of their humor . <p> One interesting finding was that coping with humor was positively related to worry . This result can be partially explained by the functions of worry . For instance , after accounting for anxiety , worry , as measured by the WDQ , has been described as an active cognitive coping mechanism ( Davey , 1994 ) . Hence , if worry and humor both serve as coping mechanisms , it is less surprising that worry and coping with humor are positively related . <p> The results of the present study have implications for the treatment of worry . Previous research indicates that humor is helpful in decreasing negative affective and cognitive states ( Cann et al. , 1999 ; Moran , 1996 ; Moran &; Massam , 1999 ; Newman &; Stone , 1996 ) . Therefore , increasing the production of humor @ @ @ @ @ @ @ @ @ @ reduce worry and other unpleasant affective states , such as depression , which often coexist with worry ( Starcevic , 1995 ) . Furthermore , there is evidence that humor might be more effective in reducing unpleasant psychological outcomes among worriers than among individuals who are not prone to worry . Newton and Dowd ( 1990 ) reported that using therapeutic interventions involving absurdities is more effective with individuals reporting less sense of humor ( such as worriers ) than with individuals who have a greater sense of humor . Using respectful humorous therapeutic interventions , increasing the production of humor , and facilitating a more playful , humorous perspective among clients may prove to be helpful techniques for counselors working with worriers . <p> The relatively low percentage of shared variance between worry and sense of humor should be noted . For instance , although the results were as predicted -- a significant negative relationship was found between worry and sense of humor-the percentage of shared variance between the WDQ and MSHS was somewhat low ( 3% ) . However , the statistical significance of the results indicates that the @ @ @ @ @ @ @ @ @ @ Johnston , 1981 ) . The amount of shared variance ( 10% ) between humor production ( of the MSHS ) and confidence ( of the WDQ ) also suggests that a meaningful and predictable relationship exists between these two variables . Moreover , the high internal consistencies of the scales ( see Table 1 ) suggest that the scales were not greatly affected by error variance ( Nunnally , 1978 ) . Thus , it is likely that the relationship between the two variables in this study is stable and relatively free of error variance . <p> The present study has several limitations that should be considered by future researchers . For instance , the results are based on a college student sample . Therefore , the clinical implications noted in this discussion are tentative . Studies involving clinical samples should be conducted . Also , in the present study I used only one instrument to measure each construct . Future research should include additional measures of worry and sense of humor . <p> In addition , there are some statistical limitations of the present study . For instance , @ @ @ @ @ @ @ @ @ @ , as they sometimes are prone to an inflated risk of capitalizing on chance relationships specific to the sample ( Judd &; McClelland , 1989 ) . Although I believe the use of a stepwise regression was warranted because of the exploratory nature of some analyses used in this study , future studies should include different regression approaches in an attempt to replicate and expand the findings of the present study . <p> Future research should include larger , more diverse samples to attempt to replicate the findings of the present study . A further exploration of the relationship between humor production and worry may provide useful information for understanding reasons worriers report less sense of humor and production of humor . For instance , one question for future research is whether lower humor production among worriers is the result of a fear of negative evaluation or the result of another as yet undetermined factor , such as social inhibition or self-focused attention . More research is also needed to experimentally evaluate the clinical efficacy of using humor as a therapeutic technique for worriers . <p> The author would like to @ @ @ @ @ @ @ @ @ @ helpful comments on an earlier draft of this article . <p> Address correspondence to William E. Kelly , who is now at the Graduate Department of Counseling , George Fox University , 12753 S.W. 68th Avenue , Portland , OR 97223 ; wkelly@georgefox.edu ( e-mail ) . TABLE 1 Means , Standard Deviations , and Internal Consistencies of the WDQ , MSHS , and Subscales ( N = 140 ) PREFORMATTED TABLE <p> 
##4001278 ABSTRACT . The Emotional Stroop ( ES ) task ( I. H. Gotlib &; C. D. McCann , 1984 ) has been proposed as an experimental measure to assess the processing of emotion or the bias in attention of emotion-laden information . However , study results have not been consistent . To further examine its reliability for empirical research , the authors of this study administered the ES task to 33 participants on 2 separate occasions separated by 1 week . Results indicated that retest reliabilities for reaction times ( RTs ) derived from the 3 separate emotion conditions ( manic , neutral , and depressive ) across the 1 week interval were very high . However , consistent with previous research , the reliabilities were very low for the interference indices ( manic and depressive ) . These low reliabilities reflect the very high intercorrelation between the RTs derived from the 3 conditions . The authors concluded that a better indicator of the reliability for this task is the individual RTs from each emotion condition . <p> Key words : emotion , emotional Stroop , reliability , @ @ @ @ @ @ @ @ @ @ &; McCann , 1984 ) is based on the original Stroop task ( Stroop , 1935 ) , which has previously been used to examine attentional processes and the well-known interference or Stroop effect . Results from studies employing the Stroop task have revealed that participants require a longer time to name the color of a stimulus when the word is incongruent than when it appears as a solid color square . That is , they have trouble saying " blue " when blue ink is used in printing the word " red . " <p> Researchers are beginning to focus on the abnormal processing of emotion or attention in emotion-related ( e.g. , affective ) disorders ( Austin et al. , 1999 ; Franke , Maier , Hardt , &; Frieboes , 1993 ; Hill &; Knowles , 1991 ; Kinderman , 1994 ; Lemelin &; Baruch , 1998 ; Tarbuck &; Paykel , 1995 ) . To assess the processing of emotional information in psychiatric disorders , such as depression , researchers may adapt traditional cognitive tests ; the Stroop task has been adapted for research examining the processing @ @ @ @ @ @ @ @ @ @ Williams &; Nulty , 1986 ) <p> The ES task differs from the Stroop task in that emotional- and neutral-content words are presented instead of color-incongruent words and the interference effect is a result of emotional content rather than incongruence of color . Williams , Mathews , and MacLeod ( 1996 ) hypothesized that the ES task measures attentional bias because depressed individuals perform poorly at color naming when the words have a depressed content than when the words have a neutral or manic content . Response latencies from depressed individuals are longer when the stimulus material is negatively valenced than when the material is neutral or positively valenced ( Kindt , Bierman &; Brosschot , 1996 ; Segal , Gemar , Truchon , Guirguis , &; Horowitz , 1995 ; Siegrist , 1997 ; Williams &; Nulty , 1986 ) . <p> It has been assumed that the emotional modification of the cognitive paradigm allows the measurement of inhibition of emotional information in much the same manner as the original cognitive paradigm measures inhibition of cognitive processes ( Kindt et al. , 1996 ) . However , this has not @ @ @ @ @ @ @ @ @ @ processing emotional information is qualitatively different and that such information follows a different neural pathway than emotionally neutral ( cognitive ) information . <p> Results from previous studies using the ES task have been inconsistent ( Williams et al. , 1996 ) and may reflect the fact that it is not a reliable measure . Reliability of the ES task has been investigated in only two previous studies ( Kindt et al. , 1996 ; Siegrist , 1997 ) , with negative results . Kindt et al . Found the reliability of emotional content words to be low ( r = .19 and .25 ) . This finding was replicated by Siegrist using self-relevant words ( r = -.04 ) . <p> A possible threat to the validity of these two studies is the confounding problem in the analysis of reliability of difference scores . This is the paradox for the measurement of change ( Murphy &; Davidshofer , 1994 ; Overall &; Woodward , 1975 ) . The paradox occurs when two variables are highly correlated . The reliabilities of the difference scores for these variables are then always low @ @ @ @ @ @ @ @ @ @ scores are assumed to be made up of true scores and error scores . If two scores are highly correlated , then the true scores must overlap considerably . Therefore , there will be hardly any difference between the true scores , and the difference seen will be almost entirely due to measurement error . <p> Independent of this possible paradox effect , there were a number of minor problems in these previous studies that could be corrected in future research . First , although supposedly healthy controls were used in both studies , the researchers did not take into account the emotional condition of the participants , as there was no assessment of their emotional state . Second , the Kindt et al . ( 1996 ) study retested participants 3 months after the initial testing , although Carmines and Zeller ( 1979 ) recommended that retesting should be conducted no later than 1 month after the initial testing . Another potential problem is in the selection of the emotion-laden and non-laden words . Hill and Knowles ( 1991 ) attributed the lack of consistent findings to the selection of @ @ @ @ @ @ @ @ @ @ by Gotlib and McCann ( 1984 ) and Gotlib and Cane ( 1987 ) were the best types of words to produce an interference effect . Neither of the two previously published studies used the recommended emotional adjectives from the Gotlib studies . <p> Our aim in the present study was to examine the test-retest reliability of the ES task using the Gotlib adjectives . Of particular relevance to this objective was an assessment of the reliability of the interference effects . We examined the paradox within this context . We formed the following hypotheses : If the variables are not highly correlated , the low reliability will indicate that the test should not be used in future research examining emotional processing in either nonclinical situations or for disorders of emotion . If there is low test-retest reliability but the main variables are highly correlated , this will indicate that the test-retest statistic is not appropriate for this task . A high test-retest reliability for the interference effect may indicate that the methodological differences between the present study and the two previously reported studies account for these differences and that the @ @ @ @ @ @ @ @ @ @ processing . Method <p> Participants <p> Participants were 33 members of the staff or student body at Swinburne University or acquaintances of the investigators ( 19 women and 14 men , with a mean age of 27 years SD = 5.54 ) . We screened all the participants for any medical or psychological illnesses by having them complete a standard mental health questionnaire , and all provided informed written consent to participate . The research was approved by the Swinburne University Human Research Ethics Committee . All participants were of White European origin . <p> Materials <p> The task we administered was based on the ES task developed by Gotlib and McCann ( 1984 ) and consisted of color words presented on a computer screen using Arial Black font , size 72 . In total , 120 words were selected from the list supplied by Gotlib and McCann . Of these words , 40 were depressive , 40 were neutral and 40 were manic . We presented the words in one of four colors ; red , green , white , and blue . The object of the task was for @ @ @ @ @ @ @ @ @ @ as possible . We administered the Beck Depression Inventory II ( BDI-II ; Beck , Steer , &; Brown , 1996 ) at both Time 1 and Time 2 to exclude any potential changes in mood over the testing interval . <p> Procedure <p> We tested the participants on two occasions , 1 week apart . We seated them approximately 1 m in front of a computer screen with headphones and a microphone to record response time ( RT ) for the ES task . We instructed the participants to say the color of the presented word as quickly as they could because we were recording their RTs . The words were presented for 1.5 s . Between each word presentation there was a blank screen for 250 ms and a fixation cross for 1 s . To minimize order effects , we gave half the participants the depressive words first , followed by neutral words , and finally the manic words . The remaining participants received the words in the opposite order ( i.e. , manic , followed by neutral and depressive words ) . Results <p> BDI-II scores ranged @ @ @ @ @ @ @ @ @ @ 1 and 2 , respectively . The mean BDI-II score for Session 1 was 2.88 ( SD - 3.27 ) and for Session 2 , 2.19 ( SD- 2.74 ) ; these means reflected very low scores on this scale . A two-tailed t test indicated that the change in the mean score was not significant , t(31) = 1.71 , p > .05 . <p> The mean RTs ( from Session 1 to Session 2 ) were stable over time ( Table 1 ) . Correlations were also computed from RTs in the neutral , depressive , and manic conditions at Time 1 and Time 2 . For neutral words , r = .80 , p < .01 ; for depressive words , r = .80 , p < .01 ; and for manic words , r = .77 , p < .01 . These correlations indicate that the RTs derived from each of the emotional conditions were consistent over time . For each category of words , the correlation was positive . <p> As stated previously , the ES task is an interference task . Hence , test-retest reliability @ @ @ @ @ @ @ @ @ @ the reaction times ( RTs ) . To calculate the reliability of the interference effect , we subtracted the RTs for the emotional words from the neutral words ( e.g. , depressive interference RT depressed - RT neutral , and manic interference RT manic - RT neutral ) for both Time 1 and Time 2 . Test-retest reliability correlations for the interference effect of the ES words were r = .24 , p > .05 , for depressive words and r = -.11 , p > .05 , for manic words . These correlations were nonsignificant and unacceptably low for the purposes of test-retest reliability and indicate the lack of reliability for depressive and manic interference effect . Thus , the interference effects from Session 1 are not comparable to the interference effects seen at Session 2 . <p> We conducted a further analysis on the RTs to examine the paradox for the reliability of difference scores . The correlations of RTs of emotional words with neutral words are contained in Table 2 ; all the correlations were positive , significant , and very strong , and r values were .90 @ @ @ @ @ @ @ @ @ @ the emotional words were strongly correlated with the RTs of the neutral words . Discussion <p> Our major aim in the present study was to examine the test-retest reliabilities of the three emotion conditions as well as the interference indices . The interference indices had been previously hypothesized to reflect the bias in processing emotion-laden words . Our results indicated high test-retest reliabilities for the RTs derived from each emotion condition separately but low test-retest reliabilities for the interference indices . <p> Because the reliability of the interference indices was very low , additional analysis were conducted to confirm the paradox . The low test-retest reliability for the interference indices may have been because of the paradox of using change scores when two measures are highly correlated or , in fact , because of the very low reliability of the interference indices . RTs from all three conditions ( depression , neutral , and manic ) were highly correlated in the present study , indicating that test-retest reliability may not be a valid statistic to use in establishing the reliability of these indices . Perhaps it would be more suitable to @ @ @ @ @ @ @ @ @ @ words in the individual emotion conditions . These findings are consistent with past studies examining the reliability of the ES task ( Kindt et al. , 1996 ; Siegrist , 1997 ) . <p> The modification of the present study using the Gotlib list of words produced test-retest correlations for RTs derived from the different emotion conditions similar to those reported in the two previous studies examining the reliability of the ES . The correlations previously reported were by Kindt et al . ( r = .65 to .84 ) and by Siegrist ( r = .84 to .91 ) . <p> For both these previous studies ( Kindt et al. , 1996 ; Siegrist , 1997 ) as well as the current study , participants were recruited from a university population . The results of the present study indicate that the poor test-retest reliability of the interference effect of the ES is most likely to be due to the " paradox of measurement change . " It is now important for researchers to determine how these interference reliability scores compare with scores in clinical populations , especially given that the @ @ @ @ @ @ @ @ @ @ in depressed and anxious individuals ( Williams et al. , 1996 ) . <p> It is only possible to conclude from the present study that the ES task is reliable when measuring RTs derived from the different emotional conditions . Because of the difficulties in the analysis of difference scores , the reliability of the interference effects remains unknown . Nevertheless , because the RTs derived from the different conditions were highly correlated , it would be sensible to suggest that the interference produced by the ES is also likely to be highly reliable . <p> Address correspondence to P. J. Nathan , Brain Sciences Institute , 400 Burwood Road , Hawthorn , Victoria 3122 , Australia ; pnathan@bsi.swin.edu.au ( e-mail ) . TABLE 1 Means and Standard Deviations for Reaction Time ( ms ) for the Different Stimulus Categories for the Emotional Stroop Task PREFORMATTED TABLE TABLE 2 Correlations of Emotional Words and Neutral Words for Reaction Time at Sessions 1 and 2 PREFORMATTED TABLE <p> 