|
?:linkContext
|
-
M ost a ttrib u tio n theories share the basic assum ption that individuals are m otivated lo attribute underlying reasons to perceived events . This assum ption is reflected in a ttrib u tio n research, which seems to imply th at the attrib u tio n o f causality is the predom inant form of cognitive activity. T he validity of this assum ption, however, is disputed. M anis , for example, questions the claim that persons are preoccupied with the search for causa! explanations most of their time. If they are not, the conditions under which causal thinking occurs in everyday life need to be circum scribed in more detail. In previous research, the determ inants of causal reasoning have been investigated w ith various non-responsive m ethods, including the content analysis o f written m aterial or o f subjects' free verbalizations in experim ental sim ulations. In other studies, researchers tried to draw conclusions about the occurrence of causal reasoning from the observation of related cognitive processes, such as inform ation search and m em ory (for a review see . The results o f these studies consistently indicate th a t the extent o f causal reasoning is determ ined by two factors (in addition to the less interesting case of explicit 'w hy'-questions), namely the expectedness of the event and its affective valence . Causal reasoning is more likely to be elicited by negative rath er than positive, and by unexpected rath er than expected events. A dditive effects of both factors (where the highest degree of causal thinking is obtained after negative and unexpected events, and the lowest degree of causal reasoning after positive and expected events) as well as occasional interaction effects have been observed. , fo r exam ple, found in an experim ental sim ulation that negative valence increased the extent of causal reasoning especially for events of high probability, but less so for events of low probability. Several processes may contribute to these findings. In the fram ew ork of schema theory (c f for reviews), one might argue that events are understood lo the extent that they are consistent with higher-order know ledgc-stm ctures (schem ata). In such cases, a search for causal explanations seems to be unnecessary. R ath er, a person should only engage in causal thinking if an event that he o r she experiences is inconsistent with the available schem ata. In this fram ew ork, unexpectedness o f an event can be conceptualized as a special case of schem a-inconsistency. assumes th at unexpected events trigger som e kind of cognitive orienting response. T he deviation from a norm (e.g. an expectancy) needs to be explained, w hereas events that are consistent with an expectancy have been frequently analysed in the past fo r their possible causes, so that the expense of attrib u tio n al activities can be avoided. In addition, negative events and their accom panying negative affective states may instigate causal reasoning beca'use individuals are m otivated to avoid unpleasant states . To do so, they need adequate knowledge a b o u t the poten tial causes of negative events. M oreover, searching explanations for negative events m ay serve the purpose o f reducing the aversive affect by finding external and self-irrelevant causes for the initially unpleasant event . Sim ilarly, action identification theory (W egner and predicts that persons who experience negative affect during or after completion o f an action may consider the action in m ore detail at a low er level of abstraction and may think about possible reasons for the action's outcom e. In ad d itio n , the process of causal reasoning itself may reduce the intensity of an em otion . Explaining why an event occurred prom otes a dissociation of affect th a t has been experim entally dem onstrated (e.g. as well as applied in various therapeutic approaches . Il must be taken into account, how ever, that an event's affective valence and subjective probability are usually confounded; that is, unpleasant events are less likely than pleasant events. Let us consider, for exam ple, the outcom e o f an exam ination: Failing an exam is m ore unpleasant, but also occurs less often than passing an exam. This n atu ral confounding of ihe tw o factors 'subjective probability' and 'affective valence' renders it difficult to isolate th eir relative im pact on causal reasoning. Indirect evidence for th e natural confounding o f expected ness and valence is provided by a m ela-analysis o i m ood studies th at revealed that m ost respondents report being in a happy m ood most of the lime . This result suggests that events which elicit positive moods may appear to be the rule, whereas events which elicit negative feelings represent exceptions from the rule . This leads to an am biguity in the interp retation of research findings that suggest m ood-dependent increases or decreases of causal and analytic thinking (e.g. : Do persons in a negative mood think m ore analytically, only because the event that made them feel sad was unexpected to them ? C an bad m oods and negative events, therefore, be conceptualized as subcaiegories o f unexpected events, as H astie proposed? O r is there a genuine impact of m oods on causal thinking, independent of their expeciedness? This issue cannot be settled on the basis o f the available em pirical evidence, because in field studies, the natural confounding o f the two variables has to be put up with, and in experim ental sim ulations or role-plays, the m ood-relevant events are produced by the subjects them selves. In order to disentangle the effects of the affective valence and subjective probability of an event on subsequent attributions, the tw o variables were independently m anipulated in a 2 * 2-factorial laboratory experim ent. Subjects were exposed to an event that was either pleasant or unpleasant, and eith er expected or unexpected. Affective valence was operationalized as success versus failure in an ability test, and high versus low expectancies were induced by presenting corres ponding d istributions of success and failure in a com parable population before the test was adm inistered. T he central dependent variable was the extent o f causal reasoning after receiving the test result. If subjective probability is the central determ inant o f causal attrib u tio n , unexpected events should always elicit explanations, and causal reasoning should increase after unexpected success as well as after unexpected failure. If, on the other hand, there is a genuine influence o f an event's affective valence on causal attrib u tio n , failure should elicit m ore causal reasoning than success* over and above the effects of expectedness. Subjects and procedure F orty-three students of the University o f M annheim were recruited as paid subjects for a study on the im provem ent of professional skills tests. The experim ent was conducted in single sessions. Subjects were told that they would be adm inistered 10 items of a professional skills test, and that after w orking on this task they were to answer various questions ab o u t the testing situation. The personal relevance o f the task was enhanced by telling subjects th at ihe test from which the items were taken was a valid predictor of professional success. The alleged 'professional skills test' items were taken from the 'S tan d ard 1 and 'A dvanced Progressive M atrices'(R aven, 1958, I962)1. To assure th at the task was well understood, the experim enter provided detailed explanations o f two exam ples. Subsequently, subjects were inform ed that they could win a chocolate bar if they perform ed successfully. Success was defined as having solved at least seven of th e 10 items. On a table in one corner of the laboratory, the subjects could see several chocolate bars, which had proved to be effective incentives in previous studies (M iinkel, S irack and . This additional reward was introduced to provide a salient m ark o f success, which would facilitate an expectancyindependent subjective definition of success. To m anipulate the expectancy of success, subjects were either inform ed that 77 per cent (high expectancy conditions) or 23 per cent (low expectancy conditions) of the previous student subjects had reached the criterion. Before subjects were adm inistered the test items, the effectiveness o f the expectancy m anipulation was checked. The question read: 'D o you think that you will succeed in solving seven or more o f the 10 items correctly?' The answ er had to be given on a scale ranging from 1 ('no, certainly nol*) to 11 ('yes, quite certainly'). A fter com pletion of this m anipulation check, the experim enter presented the 10 lest items and m ade sure th a t subjects kept the tim e limit of 60 seconds per item. Subjects were forced to choose a response alternative within this time limit and were told th at one m ay 'often guess the correct answer w ithout being able to fully explain the solution principle. So please d o m ake a choice on each item even if you are not certain about the correct solution'. This instruction was necessary to ascertain the credibility of the feedback. W hen subjects had m ade their choices for all 10 items, the experim enter took the response sheet and went to a different table, ostensibly to check the test result. Up to this point, the experim enter was blind with respect to the success versus failure condition. W hether the subject would receive positive or negative feedback was determ ined by a card that the experim enter look from a pile invisible to the subjects. In the success condition, the subject was told that he or she had eight items correct, and the experim enter let him or her choose one o f the chocolate bars. In the failure condition, th e subject was inform ed th at he or she had only fo u r items correct and therefore had unfo rtu n ately failed to win a chocolate b ar. The difficulty of the items and the tim e limit had been chosen so th a t the success as well as the failure feedback were highly credible. A n inform al interview at the end o f the experim ent revealed th at only one subject w as suspicious about the validity o f the feedback. The d ata of this subject were excluded from the analysis. Before the dependent variables were assessed, the influence of ihe success versus failure feedback on subjects' affective slaie was assessed with the question: 'H ow do you feel ai this very m om ent?' The response scale ranged from I ('very b a d 1 ) to 11 ('very good1 ). Subsequently, subjects were asked to answer a series of questions that were ostensibly designed to assess how he or she experienced the testing situ atio n . All questions pertained to the extent and intensity o f causal reasoning. T he first tw o questions were phrased in an open-answer form at. The first question pertained to the testing situation in general, the second one specifically to the test result. Finally, the intensity o f causal reasoning was directly assessed in a closed response form at. T hus, care was taken to present first open ended, non-responsive questions and to successively narrow the focus on the lest result and finally on causal reasoning. Subjects were first instructed to th in k about the testing situation and to write down everything that cam e to mind. They were provided a legal sized sheet with 18 lines for this task. There was no lime limit. A fter this thought listing, subjects were instructed to think about their test result and to w rite dow n whatever came to mind about this aspect. In all other respects, the procedure was identical to the first thought listing. Thus, the focus was narrow ed on resuli-related cognitions, but causal thinking was not yei explicitly m entioned. After com pletion of the thought listings, subjects were instructed to read through each listing and to specify how m any causal explanations of ihe test result it contained. The interp retatio n of subjects' own counting o f causal explanations is som ew hat problem atic, because subjects might not have produced any 'spontaneous' causal explan atio n s b u t m ight rather have generated reasons after being questioned to check their thought listings, which m ay have caused them to re-interpret the listings in terms of causal explanations. Therefore, both thought listings were also content-analysed by two independent judges for the to ial num ber o f words an d sentences they contained, ihe num ber of causal explanations and the num ber o f sentences m entioning positive or negative feelings2. The mean in ter-rater correspondence for the num ber of causal explanations and the num ber o f feelings m entioned was r = 0.65 and r = 0.73, respectively. F or those cases ab o u t which the judges disagreed, the arithm etic mean of their ratings was com puted for further analyses. A com parison o f the total num bers of sentences and words that the judges had counted revealed perfect inter-judge correspondence. G. Finally, subjects were directly asked to rale the intensity of causal reasoning: 'How intensely did you think about the reasons for your test result?'T he answers were to be reported on a scale ranging from 1 ('not intensely al all') to 11 ('very intensely'). As expecied, the presentation of population norm s affected the subjective probability to reach the criterion set by the experim enter. Subjects in the 'high ex p ectancy'(77 per cent success) condition reported a higher expectancy o f success (M = 7.3) th an subjects in the 'low expectancy' (23 per cent success) condition (A /= 5.9), r(40) = 1.91,/? < 0.04, one-tailed. To test the effectiveness of the m anipulation of success and failure, the answers to the question, 'H ow do you feel at this very m om ent?', were subjected to a 2 (success versus failure) * 2 (high versus low expectancy) factorial analysis of variance. This analysis revealed only a main effect of the 'success/failure' m anipulation: As expecied, subjects in the success conditions fell significantly betler (M = 8.5) than subjects in the failure conditions (A/ = 7.1), /T[l, 38) = 8.15,/? < 0 .0 1 . The expected ness o f Jhe result did not affect subjects' affective stale, F< 1 for both the expectancy main effect and the two-way interaction. Thus, the m anipulation checks revealed that the realization of different levels of subjective probability and affective valence was successful. It might be argued, however, that subjects in the failure conditions still scored above the scale m idpoint on the question of affective state and thus felt reasonably well. On the other hand, there is evidence that even depressive subjects do not score far below the scale m idpoint on questions o f present m ood , which implies that there may be a genera! tendency to present oneself favourably on questions of this kind. Even if one is in doubt whether the failure condition induced categorically negative m ood, there is a significant m ood difference between success and failure conditions, so th a t subsequent Tesults can be interpreted on the basis of a relative m ood difference. The percentages o f causal explanations in relation to the total num ber of statem ents in the thought listings about the testing situation and the test result are shown in the upper part of . The 2 (success versus failure) * 2 (high versus low expectancy) factorial analyses of variance revealed th at subjects' reported thoughts about the testing situation as well as about the test result contained m ore causal explanations when the subjects had experienced a negative event (failure; ATs = 35 per cent and 21 per cent, respectively) than when they had experienced a positive event (success: A/'s = 12 p e rce n t and 5 per cent, respectively). The corresponding main effects of 'affective valence' are statistically significant for both thought listings, F s (1, 38) = 7.38 and 6.56, p 's < 0 .0 2 . In addition to this m ain effect, an unexpected statistical trend em erged in the First thought listing, pertaining to the testing situation in general: Subjects tended to generate more causal explanations after an expected (M -32 per cent) than after an unexpected event (M = 16 per cent), F (l, 38) = 3.11, p < 0.09. This surprising result may be due to the fact that, in the high expectancy conditions, the experim enter had already presented an 'explanation' by telling subjects the population norm s. This explanation may have been easily accessible in these conditions and may have increased the num ber of causal explanations. The results of the ratings of two independent judges, again subjected to 2 (success versus failure) * 2 (high versus low expectancy) factorial analyses o f variance, parallel the subjects' own ratings. The percentages of causal explanations are displayed in the lower part of . In the jud g es' as well as the subjects' judgm ent, the negative event elicited m ore causal explanations (M = 30 per cent for thoughts on the testing situation, and A/ =18 per cent for thoughts on the test result) th an the positive event {M= 14 per cent and 7 p e rc e n t, respectively), fl[1 ,37) = 5.003,/? < 0.04, and F (l, 38) = 5.88, p < 0.03. N o m ain effect fp r expecledness and no two-way interaction was obtained, all p 's > 0.25. In the first thou g h t listing, pertaining to the testing situation, the percentages of references to positive or negative feelings were unaffected by experim ental conditions, Finally, the num ber of sentences and words in both thought listings were unaffected by experimental conditions, all p 's > 0.10. T he reported results provide first experim ental evidence for the hypothesis that the affective valence o f an event and the resulting m ood state can influence causal thinking, independently o f the event's expectedness. Negative events lead to intensified causal reasoning in com parison lo positive events of equal probability. The results of the m anipulation checks d e a rly dem onstrate that a confounding o f subjective probability and affective valence could be avoided in the present study. Therefore, affective valence seems to be a predictor o f causal attributions in its own right. Thus, the need to explain negative events is not merely a subcategory of the need to explain unexpected events. O n the other h an d , no evidence for an increase in causal reasoning after unexpected events was obtained in the present study. One possible explanation for this unexpected finding is that the expectedness of the outcom e was not as im portant to subjects as its affective valence in this particular situation. In different types of situations, the relative im pact o f expectancy and valence on causal reasoning may be quite variable. Nevertheless, the m anipulation check revealed a significant difference in subjects' expectations between the high and low expectancy conditions, indicating the effectiveness o f the present m anipulation. Thus, it is safe to conclude that the observed effects o f affective valence were independent of subjective probability, despite a successful variation o f the latter variable. N ow th a t the influence of affective valence on causal reasoning has been dem onstrated and disentangled from expectancy effects, the question arises, why persons in a negative m ood are likely to generate m ore causal attributions than persons in a positive m ood. In this regard, it is im portant to note parallel findings in other research areas th a t suggest pronounced influences of individuals' affective slate on inform ation processing. It has been dem onstrated, for example, that positive m ood, as com pared to neutral m ood, elicits simplified processing strategies and less system atic problem solving on com plex judgm ental tasks . Similarly, persons in a negative m ood were found to elaborate the argum ents of a persuasive com m unication system atically, whereas persons in a positive m ood did not engage in system atic processing. Accordingly, subjects in a depressed m ood were more persuaded by strong than by weak argum ents, whereas subjects in an elated mood were equally affected by both types of argum ents in their cognitive responses as well as in their attitu d e change . There are, at least, three possible explanations for the increase in causal and analytic reasoning under the influence of negative events and the resulting negative feeling slates. First, an intensified search for causal explanations makes it m ore likely that external and self-irrelevant attributions for the event may be identified . T hereby the evaluation of the event and its consequences may change and the negative feelings associated with it may be reduced. Second, the increase of causal reasoning after a negative event m ay serve the purpose of preparing actions that are likely to elim inate the aversive state o r to avoid sim ilar events in the future. In order to actively elim inate or avoid certain situations, a person must have reasonably accurate know ledge about their causation, i.e., he or she needs to know what conditions are to be changed or avoided. A third explanation does not focus on the results o f causal attributions, but on the process o f causal reasoning itself. Causal thinking itself may reduce the intensity o f affective states (e.g. , This hypothesis is supported by the finding th at thinking about pleasant or unpleasant past life events leads to pronounced changes in affective states if it is vivid and concrete. If, on the other hand, thinking is focused on causal explanations for the event, changes in affect are n o t obtained Experim ent 3). T hus, causal thinking m ay serve for the control o f em otion. Obviously, these process assum ptions are not m utually exclusive. W hich one is most applicable, and how far the different m echanism s may be intertw ined, cannot be decided o n the basis o f the av ailab lcd ata and aw aits clarification by further research. T he purpose o f the present study was to disentangle the natural confounding of affective valence and subjective probability in a controlled experim ent. The results suggest th a t affective valence by itself should be tak en into account as a determ inant of causal attrib u tio n in its own right. April). 'Erwartung und Valenz als Auslösebedingungen von Nachdenkprozessen'. 3rd Tagung der Arbeitsgruppe Sozialpsychologie Ist schlechte Stimmung die Ausnahme? Eine Metaanafyse von S tim m ungsuntersuchungen'. 26th Tagung experimentell arbeitender Psychologen Therapeutic com ponents shared by all psychotherapies sychotherapy Research and Behavior Change. The Master te n u r e Series Causes and effects o f causal attribution' Journal o f Personality and Social Psychology The Psychology o f Interpersonal Relations Some factors influencing decision making strategy and risk taking A ffect and Cognition The process of causal attribution American Psychologist Decentration as a means for reducing aggression after exposure to violent stimuli European Journal o f Social Psychology Cognitive social psychology Personality and Social P sychologr Bulletin Der Einfluß der experimentellen Honorierung auf Stimmung und Wohlbefinden: Macht Schokolade glücklich?'. 29th Tagung experimentell arbeitender Psychologen A dvan ced Progressive Mairices. Set //. 1962 R evision T heorien konzepigesteuerter Informationsverarbeitung Theorien der Sozialpsychologie S tim m ung als Information M ood, misattribution, and judgments o f subjective well being: Informative and directive functions o f affective states' Journal o f Personality and Social Psychology Reported emotions and conventions and emotionality among college students Journal o f Personality an d Social Psychology Experimental reduction o f stress based on ego-defense theory* Social Psychology Happiness and reminiscing: The role of time perspective, affect, and mode o f thinking' Journal o f Personality and Social Psychology Schematic bases o f social information processing S ocial Cognition: The Ontario Sym posium What do people think they're doing? Action identification and human behavior Psychological Review Action identification Cognition: Foundations o f Social Behavior Spontaneous' causal thinking Psychological Bulletin Cognitive mediation o f positive affect in persuasion Social C ognition S o cia l Cognition W as ist sozial erw ünscht? Der Grad sozialer Erwünschthcit von Einstellungsitem s Michael Braun Neuregelung zum Versand nachfolgender Arbeitsberichte betr.: Zuraa Arbeitsberichte Nr 87/06 87/12 Gruss Angelika Gerd Bohner Herbert Bless Norbert Schwarz Fritz Strack Unversilal Mannheim FRG Zentrum lur Umfragen, Methoden und Analysen. ZUMAeV. Mannheim. FRG Universität Mannheim. FRG Neuregelung zum Versand nachfolgender Arbeitsberichte betr.: Zuraa Arbeitsberichte Nr 87/06 87/12 Received 11 November 1987 Accepted 13 M ay 1988 GROBID - A machine learning software for extracting information from scholarly documents Die oben genannten Arb.-Berichte sind zwischenzeitlich in leicht geänderter Fassung in verschiedenen Zeitschriften veröffentlicht. Deshalb soll zukünftig auf Anfragen nicht mehr die alte Arb.-Berichtsversion verschickt werden, sondern eine Kopie des Artikels. Die Zeitschriftenartikel dürfen nicht in Deckel mit ZUMA-Aufdruck gebunden werden. Eine Kasterkopie des jeweiligen Artikels befindet sich in der Mappe für Masterkopie des zu ersetzenden Arbeitsberichts. Eine aktuelle Liste vorhandener ZUMA-Arbeitsberichte soll beigelegt werden. Vorläufig sollen die bereits kopierten Arb.-Berichte noch verschickt werden, bis die bereits erstellten Kopien aufgebraucht sind. Zukünftig werden nur noch Kopien von den Zeitschriftenartikeln hergestellt. IN T R O D U C T IO N M ost a ttrib u tio n theories share the basic assum ption that individuals are m otivated lo attribute underlying reasons to perceived events (e.g. H eider, 1958; Kelley, 1973) . This assum ption is reflected in a ttrib u tio n research, which seems to imply th at the attrib u tio n o f causality is the predom inant form of cognitive activity. T he validity of this assum ption, however, is disputed. M anis (1977) , for example, questions the claim that persons are preoccupied with the search for causa! explanations most of their time. If they are not, the conditions under which causal thinking occurs in everyday life need to be circum scribed in more detail. In previous research, the determ inants of causal reasoning have been investigated w ith various non-responsive m ethods, including the content analysis o f written m aterial or o f subjects' free verbalizations in experim ental sim ulations. In other studies, researchers tried to draw conclusions about the occurrence of causal reasoning from the observation of related cognitive processes, such as inform ation search and m em ory (for a review see W einer, 1985) . The results o f these studies consistently indicate th a t the extent o f causal reasoning is determ ined by two factors (in addition to the less interesting case of explicit 'w hy'-questions), namely the expectedness of the event and its affective valence (H astie, 1984; Weiner, 1985) . Causal reasoning is more likely to be elicited by negative rath er than positive, and by unexpected rath er than expected events. A dditive effects of both factors (where the highest degree of causal thinking is obtained after negative and unexpected events, and the lowest degree of causal reasoning after positive and expected events) as well as occasional interaction effects have been observed. Abele (1986) , fo r exam ple, found in an experim ental sim ulation that negative valence increased the extent of causal reasoning especially for events of high probability, but less so for events of low probability. Several processes may contribute to these findings. In the fram ew ork of schema theory (c f Taylor and C rocker, 1981; Schwarz, 1985 for reviews), one might argue that events are understood lo the extent that they are consistent with higher-order know ledgc-stm ctures (schem ata). In such cases, a search for causal explanations seems to be unnecessary. R ath er, a person should only engage in causal thinking if an event that he o r she experiences is inconsistent with the available schem ata. In this fram ew ork, unexpectedness o f an event can be conceptualized as a special case of schem a-inconsistency. Sim ilarly, W einer (1985) assumes th at unexpected events trigger som e kind of cognitive orienting response. T he deviation from a norm (e.g. an expectancy) needs to be explained, w hereas events that are consistent with an expectancy have been frequently analysed in the past fo r their possible causes, so that the expense of attrib u tio n al activities can be avoided. In addition, negative events and their accom panying negative affective states may instigate causal reasoning beca'use individuals are m otivated to avoid unpleasant states (W einer, 1985; Schw arz, 1987) . To do so, they need adequate knowledge a b o u t the poten tial causes of negative events. M oreover, searching explanations for negative events m ay serve the purpose o f reducing the aversive affect by finding external and self-irrelevant causes for the initially unpleasant event (W yer and C arlslon, 1979) . Sim ilarly, action identification theory (W egner and Vallacher, 1986) predicts that persons who experience negative affect during or after completion o f an action may consider the action in m ore detail at a low er level of abstraction (V allacher and W egner, 1987) and may think about possible reasons for the action's outcom e. In ad d itio n , the process of causal reasoning itself may reduce the intensity of an em otion (Schw arz, 1987) . Explaining why an event occurred prom otes a dissociation of affect th a t has been experim entally dem onstrated (e.g. Speism an, Lazarus, M ordkoff and D avison, 1964; Leyens, Cisneros and Hossay, 1976; S track, Schwarz and G schneidinger, 1985) as well as applied in various therapeutic approaches (Frank, 1982) . Il must be taken into account, how ever, that an event's affective valence and subjective probability are usually confounded; that is, unpleasant events are less likely than pleasant events. Let us consider, for exam ple, the outcom e o f an exam ination: Failing an exam is m ore unpleasant, but also occurs less often than passing an exam. This n atu ral confounding of ihe tw o factors 'subjective probability' and 'affective valence' renders it difficult to isolate th eir relative im pact on causal reasoning. Indirect evidence for th e natural confounding o f expected ness and valence is provided by a m ela-analysis o i m ood studies th at revealed that m ost respondents report being in a happy m ood most of the lime (Bless and Schwarz, 1984) . This result suggests that events which elicit positive moods may appear to be the rule, whereas events which elicit negative feelings represent exceptions from the rule (Som m ers, 1984; Schwarz and Clore, 1983; Schwarz, 1987) . This leads to an am biguity in the interp retation of research findings that suggest m ood-dependent increases or decreases of causal and analytic thinking (e.g. Isen, M eans, P atrick and Nowicki, 1982; Abeie, 1986) : Do persons in a negative mood think m ore analytically, only because the event that made them feel sad was unexpected to them ? C an bad m oods and negative events, therefore, be conceptualized as subcaiegories o f unexpected events, as H astie (1984) proposed? O r is there a genuine impact of m oods on causal thinking, independent of their expeciedness? This issue cannot be settled on the basis o f the available em pirical evidence, because in field studies, the natural confounding o f the two variables has to be put up with, and in experim ental sim ulations or role-plays, the m ood-relevant events are produced by the subjects them selves. In order to disentangle the effects of the affective valence and subjective probability of an event on subsequent attributions, the tw o variables were independently m anipulated in a 2 * 2-factorial laboratory experim ent. Subjects were exposed to an event that was either pleasant or unpleasant, and eith er expected or unexpected. Affective valence was operationalized as success versus failure in an ability test, and high versus low expectancies were induced by presenting corres ponding d istributions of success and failure in a com parable population before the test was adm inistered. T he central dependent variable was the extent o f causal reasoning after receiving the test result. If subjective probability is the central determ inant o f causal attrib u tio n , unexpected events should always elicit explanations, and causal reasoning should increase after unexpected success as well as after unexpected failure. If, on the other hand, there is a genuine influence o f an event's affective valence on causal attrib u tio n , failure should elicit m ore causal reasoning than success* over and above the effects of expectedness. M E T H O D Subjects and procedure F orty-three students of the University o f M annheim were recruited as paid subjects for a study on the im provem ent of professional skills tests. The experim ent was conducted in single sessions. Subjects were told that they would be adm inistered 10 items of a professional skills test, and that after w orking on this task they were to answer various questions ab o u t the testing situation. The personal relevance o f the task was enhanced by telling subjects th at ihe test from which the items were taken was a valid predictor of professional success. The alleged 'professional skills test' items were taken from the 'S tan d ard 1 and 'A dvanced Progressive M atrices'(R aven, 1958, I962)1. Independent variables and m anipulation check To assure th at the task was well understood, the experim enter provided detailed explanations o f two exam ples. Subsequently, subjects were inform ed that they could win a chocolate bar if they perform ed successfully. Success was defined as having solved at least seven of th e 10 items. On a table in one corner of the laboratory, the subjects could see several chocolate bars, which had proved to be effective incentives in previous studies (M iinkel, S irack and Schwarz, 1987) . This additional reward was introduced to provide a salient m ark o f success, which would facilitate an expectancyindependent subjective definition of success. To m anipulate the expectancy of success, subjects were either inform ed that 77 per cent (high expectancy conditions) or 23 per cent (low expectancy conditions) of the previous student subjects had reached the criterion. Before subjects were adm inistered the test items, the effectiveness o f the expectancy m anipulation was checked. The question read: 'D o you think that you will succeed in solving seven or more o f the 10 items correctly?' The answ er had to be given on a scale ranging from 1 ('no, certainly nol*) to 11 ('yes, quite certainly'). A fter com pletion of this m anipulation check, the experim enter presented the 10 lest items and m ade sure th a t subjects kept the tim e limit of 60 seconds per item. Subjects were forced to choose a response alternative within this time limit and were told th at one m ay 'often guess the correct answer w ithout being able to fully explain the solution principle. So please d o m ake a choice on each item even if you are not certain about the correct solution'. This instruction was necessary to ascertain the credibility of the feedback. W hen subjects had m ade their choices for all 10 items, the experim enter took the response sheet and went to a different table, ostensibly to check the test result. Up to this point, the experim enter was blind with respect to the success versus failure condition. W hether the subject would receive positive or negative feedback was determ ined by a card that the experim enter look from a pile invisible to the subjects. In the success condition, the subject was told that he or she had eight items correct, and the experim enter let him or her choose one o f the chocolate bars. In the failure condition, th e subject was inform ed th at he or she had only fo u r items correct and therefore had unfo rtu n ately failed to win a chocolate b ar. The difficulty of the items and the tim e limit had been chosen so th a t the success as well as the failure feedback were highly credible. A n inform al interview at the end o f the experim ent revealed th at only one subject w as suspicious about the validity o f the feedback. The d ata of this subject were excluded from the analysis. Before the dependent variables were assessed, the influence of ihe success versus failure feedback on subjects' affective slaie was assessed with the question: 'H ow do you feel ai this very m om ent?' The response scale ranged from I ('very b a d 1 ) to 11 ('very good1 ). D ependent variables Subsequently, subjects were asked to answer a series of questions that were ostensibly designed to assess how he or she experienced the testing situ atio n . All questions pertained to the extent and intensity o f causal reasoning. T he first tw o questions were phrased in an open-answer form at. The first question pertained to the testing situation in general, the second one specifically to the test result. Finally, the intensity o f causal reasoning was directly assessed in a closed response form at. T hus, care was taken to present first open ended, non-responsive questions and to successively narrow the focus on the lest result and finally on causal reasoning. Thought listings Subjects were first instructed to th in k about the testing situation and to write down everything that cam e to mind. They were provided a legal sized sheet with 18 lines for this task. There was no lime limit. A fter this thought listing, subjects were instructed to think about their test result and to w rite dow n whatever came to mind about this aspect. In all other respects, the procedure was identical to the first thought listing. Thus, the focus was narrow ed on resuli-related cognitions, but causal thinking was not yei explicitly m entioned. After com pletion of the thought listings, subjects were instructed to read through each listing and to specify how m any causal explanations of ihe test result it contained. Content analysis The interp retatio n of subjects' own counting o f causal explanations is som ew hat problem atic, because subjects might not have produced any 'spontaneous' causal explan atio n s b u t m ight rather have generated reasons after being questioned to check their thought listings, which m ay have caused them to re-interpret the listings in terms of causal explanations. Therefore, both thought listings were also content-analysed by two independent judges for the to ial num ber o f words an d sentences they contained, ihe num ber of causal explanations and the num ber o f sentences m entioning positive or negative feelings2. The mean in ter-rater correspondence for the num ber of causal explanations and the num ber o f feelings m entioned was r = 0.65 and r = 0.73, respectively. F or those cases ab o u t which the judges disagreed, the arithm etic mean of their ratings was com puted for further analyses. A com parison o f the total num bers of sentences and words that the judges had counted revealed perfect inter-judge correspondence. G. Bohner. H Bless, N. Schw ar: and F. Strack Intensityjudgments Finally, subjects were directly asked to rale the intensity of causal reasoning: 'How intensely did you think about the reasons for your test result?'T he answers were to be reported on a scale ranging from 1 ('not intensely al all') to 11 ('very intensely'). RESULTS M anipulation checks As expecied, the presentation of population norm s affected the subjective probability to reach the criterion set by the experim enter. Subjects in the 'high ex p ectancy'(77 per cent success) condition reported a higher expectancy o f success (M = 7.3) th an subjects in the 'low expectancy' (23 per cent success) condition (A /= 5.9), r(40) = 1.91,/? < 0.04, one-tailed. To test the effectiveness of the m anipulation of success and failure, the answers to the question, 'H ow do you feel at this very m om ent?', were subjected to a 2 (success versus failure) * 2 (high versus low expectancy) factorial analysis of variance. This analysis revealed only a main effect of the 'success/failure' m anipulation: As expecied, subjects in the success conditions fell significantly betler (M = 8.5) than subjects in the failure conditions (A/ = 7.1), /T[l, 38) = 8.15,/? < 0 .0 1 . The expected ness o f Jhe result did not affect subjects' affective stale, F< 1 for both the expectancy main effect and the two-way interaction. Figure I. M ea n p erccn ta g es o f causal attrib u tion s in su b je c ts' c o g n itio n s a b o u t the testin g situ a tio n (left) a nd the test result (right) as a fu n ctio n o f valencc and subjective probab ility: Su bjects' self-ratings (ab ove); results o f con ten t an alysis (b elo w ). S -success, F = failure Thus, the m anipulation checks revealed that the realization of different levels of subjective probability and affective valence was successful. It might be argued, however, that subjects in the failure conditions still scored above the scale m idpoint on the question of affective state and thus felt reasonably well. On the other hand, there is evidence that even depressive subjects do not score far below the scale m idpoint on questions o f present m ood (Bless, Bohner, Schw arz and Strack, 1988) , which implies that there may be a genera! tendency to present oneself favourably on questions of this kind. Even if one is in doubt whether the failure condition induced categorically negative m ood, there is a significant m ood difference between success and failure conditions, so th a t subsequent Tesults can be interpreted on the basis of a relative m ood difference. Though! listings The percentages o f causal explanations in relation to the total num ber of statem ents in the thought listings about the testing situation and the test result are shown in the upper part of Figure 1 . The 2 (success versus failure) * 2 (high versus low expectancy) factorial analyses of variance revealed th at subjects' reported thoughts about the testing situation as well as about the test result contained m ore causal explanations when the subjects had experienced a negative event (failure; ATs = 35 per cent and 21 per cent, respectively) than when they had experienced a positive event (success: A/'s = 12 p e rce n t and 5 per cent, respectively). The corresponding main effects of 'affective valence' are statistically significant for both thought listings, F s (1, 38) = 7.38 and 6.56, p 's < 0 .0 2 . In addition to this m ain effect, an unexpected statistical trend em erged in the First thought listing, pertaining to the testing situation in general: Subjects tended to generate more causal explanations after an expected (M -32 per cent) than after an unexpected event (M = 16 per cent), F (l, 38) = 3.11, p < 0.09. This surprising result may be due to the fact that, in the high expectancy conditions, the experim enter had already presented an 'explanation' by telling subjects the population norm s. This explanation may have been easily accessible in these conditions and may have increased the num ber of causal explanations. Content analysis The results of the ratings of two independent judges, again subjected to 2 (success versus failure) * 2 (high versus low expectancy) factorial analyses o f variance, parallel the subjects' own ratings. The percentages of causal explanations are displayed in the lower part of Figure 1 . In the jud g es' as well as the subjects' judgm ent, the negative event elicited m ore causal explanations (M = 30 per cent for thoughts on the testing situation, and A/ =18 per cent for thoughts on the test result) th an the positive event {M= 14 per cent and 7 p e rc e n t, respectively), fl[1 ,37) = 5.003,/? < 0.04, and F (l, 38) = 5.88, p < 0.03. N o m ain effect fp r expecledness and no two-way interaction was obtained, all p 's > 0.25. In the first thou g h t listing, pertaining to the testing situation, the percentages of references to positive or negative feelings were unaffected by experim ental conditions, Finally, the num ber of sentences and words in both thought listings were unaffected by experimental conditions, all p 's > 0.10. D ISC U SSIO N T he reported results provide first experim ental evidence for the hypothesis that the affective valence o f an event and the resulting m ood state can influence causal thinking, independently o f the event's expectedness. Negative events lead to intensified causal reasoning in com parison lo positive events of equal probability. The results of the m anipulation checks d e a rly dem onstrate that a confounding o f subjective probability and affective valence could be avoided in the present study. Therefore, affective valence seems to be a predictor o f causal attributions in its own right. Thus, the need to explain negative events is not merely a subcategory of the need to explain unexpected events. O n the other h an d , no evidence for an increase in causal reasoning after unexpected events was obtained in the present study. One possible explanation for this unexpected finding is that the expectedness of the outcom e was not as im portant to subjects as its affective valence in this particular situation. In different types of situations, the relative im pact o f expectancy and valence on causal reasoning may be quite variable. Nevertheless, the m anipulation check revealed a significant difference in subjects' expectations between the high and low expectancy conditions, indicating the effectiveness o f the present m anipulation. Thus, it is safe to conclude that the observed effects o f affective valence were independent of subjective probability, despite a successful variation o f the latter variable. N ow th a t the influence of affective valence on causal reasoning has been dem onstrated and disentangled from expectancy effects, the question arises, why persons in a negative m ood are likely to generate m ore causal attributions than persons in a positive m ood. In this regard, it is im portant to note parallel findings in other research areas th a t suggest pronounced influences of individuals' affective slate on inform ation processing. It has been dem onstrated, for example, that positive m ood, as com pared to neutral m ood, elicits simplified processing strategies and less system atic problem solving on com plex judgm ental tasks (Isen el al., 1982) . Similarly, persons in a negative m ood were found to elaborate the argum ents of a persuasive com m unication system atically, whereas persons in a positive m ood did not engage in system atic processing. Accordingly, subjects in a depressed m ood were more persuaded by strong than by weak argum ents, whereas subjects in an elated mood were equally affected by both types of argum ents in their cognitive responses as well as in their attitu d e change (Bless et al., 1988; W orth and M ackie, 1987) . There are, at least, three possible explanations for the increase in causal and analytic reasoning under the influence of negative events and the resulting negative feeling slates. First, an intensified search for causal explanations makes it m ore likely that external and self-irrelevant attributions for the event may be identified (W yer and C arlston, 1979) . T hereby the evaluation of the event and its consequences may change and the negative feelings associated with it may be reduced. Second, the increase of causal reasoning after a negative event m ay serve the purpose of preparing actions that are likely to elim inate the aversive state o r to avoid sim ilar events in the future. In order to actively elim inate or avoid certain situations, a person must have reasonably accurate know ledge about their causation, i.e., he or she needs to know what conditions are to be changed or avoided. A third explanation does not focus on the results o f causal attributions, but on the process o f causal reasoning itself. Causal thinking itself may reduce the intensity o f affective states (e.g. Speism an et al., 1964; Schw arz, 1987) , This hypothesis is supported by the finding th at thinking about pleasant or unpleasant past life events leads to pronounced changes in affective states if it is vivid and concrete. If, on the other hand, thinking is focused on causal explanations for the event, changes in affect are n o t obtained (Strack et al., 1985, Experim ent 3). T hus, causal thinking m ay serve for the control o f em otion. Obviously, these process assum ptions are not m utually exclusive. W hich one is most applicable, and how far the different m echanism s may be intertw ined, cannot be decided o n the basis o f the av ailab lcd ata and aw aits clarification by further research. T he purpose o f the present study was to disentangle the natural confounding of affective valence and subjective probability in a controlled experim ent. The results suggest th a t affective valence by itself should be tak en into account as a determ inant of causal attrib u tio n in its own right. R ÉSU M É ' it e m n u m bers (in th e order o f a d m in istr a tio n ) in th e failure c o n d itio n s were: A d van ced P rogressive M atrices (A P M ) S e l II: 2 4 ,2 1 , 17; S ta n d a rd P rogressive M atrices (S P M ) S et E: 12; A PM S et [I: 28. 13, 16, 23, 2 6 , 19. In ih e s u c c e ss c o n d itio n s: A P M S e l II: 13, 14, 24; S P M S e l E: 9; A P M S etII; 17, 23, 26, 20 , 16, 15. fo llo w in g in stru ctio n w as given <0 th e ju d g es: 'J. C o u n t the num ber o f sen te n c es and th e nu m ber o f w o r d s on ea ch th o u g h t listing. 2. Read the listin g carefu lly and check the num ber o f c au sal ex p la n a tio n s for th e le sl result it c o n ta in s 3. R ead the listin g carefu lly and ch eck the num ber o f statem en ts th at express p o sitiv e feelin gs (.like jo y , pleasure, relief, h a p p in ess, p o sitiv e s u rp ris e etc.). 4. R ead ihe listin g carefu lly and c h eck th e num ber o r statem en ts that ex p ress n eg a tive feelin gs (like sad n ess, anger, d e p r e ssio n , u n h a p p in e s s , negative surprise etc.)'. B o th judges analysed the w h ole corp us o f response sheets. G . Bohner, H. Bless, A'. Schwarz and F Sirack all p 's > 0 .1 5 . In the second thought listing, pertaining to the test result, main effects of affective valence were obtained. After success, subjects' thought listings contained m ore references to positive (M = 0.66) and fewer references to negative (M -0.11) feelings than after failure (M = 0.05 and M -0.40, respectively), fl[l, 38) = 20.58, p < 0.0005. and fl[l, 38) = 4.72, p < 0.04. No other main or interaction effects emerged, all p 's > 0.15. These results provide further evidence for a change of affective state in reaction to success and failure. bjects' m ean ratings o f intensity o f causal reasonin g as a fu nction or valence and subjective prob ab ility (elev en -p o in i scale ranging from 1. 'not inten sely at all*, lo 11, 'very intensely"). S = success. F = failure Intensity of causal reasoning Subjects' self-ratings o f the intensity of their causal reasoning are consistent with the results reported so far. As can be seen in Figure 2, subjects in the failure conditions reported a greater intensity of causal reasoning {M -7.3) than subjects in the success conditions (M -5.4. F[ 1, 38) = 4.39, p < 0.05. A gain, n o significant effect o f subjective probability o f the outcom e emerged, all other F< 1. Qu'est ce qui élicite l'attribution causalc: l'im pact de la valence et des probabilités subjectives. Plusieures études sur le terrain et des sim ulations expérim entales ont démontré que le raisonnement causal augmente après des événements inattendus aussi bien qu'après des événements désagréables. Cependant, des événements désagréables apparaissent moins probables dans la vie quotidienne que des événements agréables. De fait, la probabilité d'un événement et sa valeur hédonique étaient confondues de manière naturelle dans ces études. Pour isoler la contribution des deux facteurs, la probabilité subjective et la valence d'un événement étaient manipulées indépendamment dans une expérience de laboratoire. Les sujets ont répondu à un soi-disant 'test de savoir-faire professionel' et étaient informés de leur réussite ou de leur échec selon un critère établi par l'expérimentateur. La probabilité subjective de réussite a été manipulée en informant les sujets de la distribution des réussites et des échecs dans une population com parable (soit 23 per cent soit 77 per cent réussissaient le test). Les résultats indiquent un effet marqué de la valence: lorsque le sujet croyait avoir échoué, l'intensité du raisonnement causal était plus forte et les raisons avancées pour expliquer l'échec plus nombreuses que lorsque le sujet croyait avoir réussi, et cela indépendamment de la p robabilité a priori de la réussite. Aucun effet du charactère attendu ou inattendu de l'événem ent sur l'explication causale a été trouvé. Plusieurs processus médiateurs sont discutés. Z U SA M M E N FA SSU N G ln einer Reihe von Feldstudien und Sim ulaiionscxpenm cnten wurde gezeigt, daß sow ohl erwartungsdiskrepante als auch negative Ereignisse das Ausmaß kausalen Nachdenkens erhöhen. Da im Alltag allerdings unangenehme Ereignisse für unwahrscheinlicher gehalten werden als angenehme, war somit die subjektive Wahrscheinlichkeit eines Ereignisses mit der durch die Valenz des Ereignisses ausgelösten Stimmung konfundiert. Um diese Konfundierung aufzulösen, wurden in einem Laborexperiment subjektive Erwartung und Valenz unabhängig voneinander variiert. Versuchspersonen bearbeiteten einen angeblichen Berufseignungstest und erhielten anschließend eine Erfolgs-oder M ißerfolgsrückm eldung, bezogen auf ein vom Versuchsleiter vorgegebenes Kriterium. Die subjektive Erfolgswahrscheinlichkeit wurde variiert, indem den Versuchspersonen die Verteilung von Erfolg und Mißerfolg in der Population m itgeteilt wurde (*23 Prozent vs. 77 Prozent der Teilnehmer erreichen das Kriterium'). Die Ergebnisse zeigen einen ausgeprägten Effekt der Valenz: Sowohl die Intensität des kausalen Nachdenkens als auch die Anzahl der Ursachen für das Ergebnis, über die die Versuchspersonen nachdenken, ist nach negativer Rückmeldung größer als nach positiver. Die vorliegende Untersuchung bietet keine Evidenz für eine Zunahme kausalen Nachdenkens nach unerwarteten Ereignissen. Verschiedene vermittelnde Prozesse werden diskutiert.W illibald Nagl, Karl F. Schuessler M ethoden der Analyse zeitbezogener Daten. V ortragsskripten der ZU M A Arbeitstagung vom 25.9. bis 5.10,79 Erika Brückner, Hans-Peter Kirschner, R olf Porst, Peter Prüfer, Peter Schm idt M ethodenbericht zum "ALLBUS 1980" M anfred Küchler, Thom as P. Wilson, Don H. Zim m erm an Integration von qualitativen und quantitativen Forschungsansätzen Gerhard Arminger, H orst Busse, M anfred Küchler Verallgemeinerte Lipeare M odelle in der em pirischen Sozialforschung Glenn R. Carroll Dynamic analysis of discrete dependent variables: A didactic essay M anfred Küchler Z ur M essung der Stabilität von W ählerpotentialen M anfred Küchler Z ur Konstanz der Recallfrage R olf Porst " ALLBUS 1982" -Systematische Variablenübersicht und erste A nsätze zu einer Kritik des Fragenprogramms Peter Ph. M öhler SA R -Sim ple AND Retrieval mit dem Siem ens-EDT-Textm anipulationsprogram m C ornelia Krauth Vergleichsstudien zum "ALLBUS 1980" W erner Hagstotz, Hans-Peter Kirschner, R olf Porst, Peter Prüfer M ethodenbericht zum "ALLBUS 1982" B em d W egener Tw o approaches to the analysis of judgm ents of prestige: Interindividual differences and the general scale R o lf Porst Synopse der ALLBUS-Variablen. D ie Systematik des A LLBU S-Fragenprogram ms und ihre inhaltliche Ausgestaltung im A LLBUS 1980 und ALLBUS 1982 M anfred Küchler, Peter Ph. Möhler Qualshop (ZUM A-Arbeitstagung zum "D atenm anagem ent bei qualitativen Erhebungsverfahren") -Sammlung von Arbeitspapieren und -berichten, Teil I + 0 B em d W egener G ibt es Sozialprestige? Konstruktion und V alidität der M agnitude-Prestige-Skala Peter Prüfer, M argrit Rexroth Erfahrungen mit einer Technik zur Bewertung von Interview erverhalten ' O ne th o u g h t listin g a b o u t th e testin g situ a tio n w as u n read ab le, so there w ere o n ly 41 valid coses fo r this part o f the analysis. G Bohner H Bless N April). 'Erwartung und Valenz als Auslösebedingungen von Nachdenkprozessen'. 3rd Tagung der Arbeitsgruppe Sozialpsychologie Abele Erlangen, FRG Ist schlechte Stimmung die Ausnahme? Eine Metaanafyse von S tim m ungsuntersuchungen'. 26th Tagung experimentell arbeitender Psychologen H Bless G Bohner N Schwarz F Sirack Bless. H. and Schwarz, N. Nürnberg. F R G Happy and mindless? M oods and the processing o f persuasive com m unications'. Manuscript, subm itted for publication Therapeutic com ponents shared by all psychotherapies J D Frank sychotherapy Research and Behavior Change. The Master te n u r e Series Harvey, J. H. and Parks, M-M Washington APA I Causes and effects o f causal attribution' R Hastie Journal o f Personality and Social Psychology 46 The Psychology o f Interpersonal Relations F Heider Wiley New York Some factors influencing decision making strategy and risk taking A M Jsen B Means P Patrick G Nowicki A ffect and Cognition Clark, M. S. and Fiske. S. T. The process of causal attribution H H Kelley American Psychologist 28 Decentration as a means for reducing aggression after exposure to violent stimuli J Levens T Cisneros J Hossay European Journal o f Social Psychology 6 Cognitive social psychology M M Anis Personality and Social P sychologr Bulletin 3 Der Einfluß der experimentellen Honorierung auf Stimmung und Wohlbefinden: Macht Schokolade glücklich?'. 29th Tagung experimentell arbeitender Psychologen T Strack F Schwarz N Aachen, FRG A dvan ced Progressive Mairices. Set //. 1962 R evision J C Raven Sets A. B. C, D. a n d E. Lewis, London. Raven, J. C. Lewis. London S tandard Progressive Matrices T heorien konzepigesteuerter Informationsverarbeitung N Schwarz Theorien der Sozialpsychologie Frey, D. and Irle, M. 3 N Schwarz S tim m ung als Information Heidelberg Springer M ood, misattribution, and judgments o f subjective well being: Informative and directive functions o f affective states' N Schwarz G Clorc Journal o f Personality and Social Psychology 45 Reported emotions and conventions and emotionality among college students S Somm Ers Journal o f Personality an d Social Psychology 46 Experimental reduction o f stress based on ego-defense theory* J C Speism An R S Lazarus A Ordkoff L Davison Social Psychology 68 Happiness and reminiscing: The role of time perspective, affect, and mode o f thinking' F Strack N Schwarz E Gschneidinger Journal o f Personality and Social Psychology 49 Schematic bases o f social information processing S E Taylor J Crocker S ocial Cognition: The Ontario Sym posium Higgins, E. T., Herman, P. and Zanna, M. Erlbaum, Hillsdale, NJ What do people think they're doing? Action identification and human behavior R R Vallacher D M Wegner Psychological Review 94 Action identification D M Wegner R R Vallacher Cognition: Foundations o f Social Behavior Sorrentino, R. M. and Higgins, E. T Spontaneous' causal thinking B Weiner Psychological Bulletin 97 Cognitive mediation o f positive affect in persuasion L T Worth Mackic Social C ognition 5 S o cia l Cognition R S W Yer D E Carlston Erlbaum, H illsdale, N .J. Inference a n d A ttribu tion W as ist sozial erw ünscht? Der Grad sozialer Erwünschthcit von Einstellungsitem s Michael Braun R einer Trom eter
(xsd:string)
|
![[This page as RDF]](https://data.gesis.org/gesiskg/static/rdf-icon.gif){kind=icon}