Should the messenger be shot if something wrong happens? Judging responsibility of a speaker giving uncertain statements.

co-author: Sarah R. Beck

In three experiments we tested how a speaker predicting uncertain events using verbal probabilities or percentages is held responsible according to which uncertainty format s/he used.
In experiment 1, the speaker used percentages, positive verbal probabilities or negative verbal probabilities in eight scenarios. Speakers were held more responsible if they used negative verbal probabilities rather than percentages or positive verbal probabilities. Speakers were also held more responsible when the outcome was negative than when it was positive. In experiment 2, speakers used round percentages (e.g., 70%) or precise percentages (e.g., 71%) within the same scenarios. Only speakers who used round percentages were held more responsible after a negative outcome. Across experiments 1 and 2 participants’ judgements ignored the predicted likelihood of the uncertain event.
Experiment 3 replicated experiment 2 with an additional measure of the intention to recommend to another. Speakers were held more responsible for positive than for negative outcomes and when the event’s likelihood was high rather than low. Intention to recommend was rated higher after positive outcomes than after negative outcomes; this difference was bigger when the event’s likelihood was high. Level of precision had no effect on the responsibility judgements or on the intention to recommend.

Overcoming the framing effect when making decisions based on verbal probabilities: Having more time is helpful but not enough

co-author: Sarah R. Beck

Uncertain outcomes can be described by raw probabilities (e.g., There is 40% chance), but also by verbal probabilities (e.g., There is a chance, It is not absolutely certain). Beyond their probabilistic meaning verbal probabilities also have a directionality (Teigen & Brun, 1995), i.e. can be positive or negative. The directionality gives verbal probabilities a framing effect on decisions: when presented with the likelihood that a drug for a headache will work, people typically recommend other people taking this drug more often if the likelihood was given using a positive verbal probability (e.g., some possibility) than if it was given using a negative verbal probability (e.g., quite uncertain), despite both verbal probabilities being judged by other participants as having the same probabilistic meaning (Teigen & Brun, 1999). In this study we made the first investigation into the potential differences in processing directionality and probabilistic meaning that could explain the framing effect of the directionality. We will also examine the conditions potentially favourable to this effect.

In experiment 1, nineteen participants chose between two outcomes described by verbal probabilities. In one third of the trials the probabilistic meaning was controlled and the directionality varied. In another third the directionality was controlled and the probabilistic meaning varied. In the last third, both dimensions were different, reinforcing each other (congruent trials; e.g., a positive verbal probability carrying a high probabilistic meaning) or contradicting each other (incongruent trials; e.g. a negative verbal probability carrying a high probabilistic meaning).
When both dimensions differed, participants chose more quickly between congruent verbal probabilities than between incongruent verbal probabilities. When only one dimension (directionality or probabilistic meaning) varied, participants chose more quickly between positive verbal probabilities and between pairs of high probabilistic meaning than between negative verbal probabilities and between pairs of low probabilistic meaning. Participants were also more accurate (i.e. chose the verbal probability carrying the highest probabilistic meaning most often) when choosing between congruent verbal probabilities and between positive verbal probabilities than when choosing between incongruent verbal probabilities and between negative verbal probabilities. Finally when the probabilistic meaning was held constant, participants tended to choose the outcome with the positive verbal probability more often than chance.

In experiment 2, twenty participants completed the same task under two time conditions (in counterbalanced order): in the limited time condition, they had to answer within five seconds; in the unlimited time condition, they could take as much time as they needed. We aimed here to replicate the results of experiment 1 and to investigate if the framing effect could be overcome when time pressure was removed. In both time conditions the same pattern as in experiment 1 was observed regarding accuracy and response time. However when the probabilistic meaning was held constant, we observed that the preference for the positive verbal probability was no longer observed given unlimited time.

In both experiments and both time conditions we observed a longer response time in the incongruent and low conditions, i.e. if the probabilistic meaning was contradicted by the directionality and if it was inconsistent with the task goal (find some treasure). This signals that people do not consider independently the probabilistic meaning or the directionality and suggests that the framing effect of directionality cannot be explained by people considering only the directionality. Also, even when there was no time pressure, the framing effect was seen for incongruent verbal probabilities. However if they had the same probabilistic meaning the framing effect was not observed in the unlimited time condition. Therefore having more time to choose between two verbal probabilities helps to overcome the framing effect of directionality, but this is the case only in some conditions. We conclude that the framing effect of directionality does not result only from performance costs and we will present suggestions to explore the pragmatics factors which are favourable to this framing effect.

Overcoming the framing effect when making decisions based on verbal probabilities: having more time is necessary but not enough.

co-author: Sarah R. Beck

Uncertain outcomes can be described by raw probabilities (e.g., There is 40% chance), but also by verbal probabilities (e.g., There is a chance, It is not absolutely certain). Beyond their probabilistic meaning verbal probabilities also have a directionality (Teigen & Brun, 1995), i.e. can be positive or negative. This directionality gives verbal probabilities a framing effect on decisions: when presented with two drugs for a headache, people typically prefer the one that is described by a positive verbal probability (e.g., some possibility) rather than the one introduced by a negative verbal probability (e.g., quite uncertain), even if both verbal probabilities are judged by other participants as having the same probabilistic meaning (Teigen & Brun, 1999). In this study we made the first investigation into the potential differences in processing directionality and probabilistic meaning that could explain the framing effect of the directionality.
In experiment 1, twenty participants chose between two outcomes described by verbal probabilities. In one third of the trials the probabilistic meaning was controlled and the directionality varied. In another third the directionality was controlled and the probabilistic meaning varied. In the last third, both dimensions were different, reinforcing each other (congruent trials; e.g., a positive verbal probability carrying a high probabilistic meaning) or contradicting each other (incongruent trials; e.g. a negative verbal probability carrying a high probabilistic meaning).
When both dimensions differed, participants chose more quickly between congruent verbal probabilities than between incongruent verbal probabilities. When only one dimension (directionality or probabilistic meaning) varied, participants chose more quickly between positive verbal probabilities and between ones of high probabilistic meaning than between negative verbal probabilities and between ones of low probabilistic meaning. Participants were also more accurate (i.e. chose the verbal probability carrying the highest probabilistic meaning most often) when choosing between congruent verbal probabilities and between positive verbal probabilities than when choosing between incongruent verbal probabilities and between negative verbal probabilities. Finally when the probabilistic meaning was held constant, participants tended to choose the outcome with the positive verbal probability more often than chance.
In experiment 2, twenty participants did the same task under two time conditions: in the limited time condition, they had to answer within five seconds; in the unlimited time condition, they could take all the time they needed. The order of the time conditions was counterbalanced. Besides replicating the results of experiment 1, our aim was to investigate if the framing effect can be overcome under some conditions, here when relieving the time pressure.
The same pattern as in experiment 1 was observed regarding accuracy and response time. However when the probabilistic meaning was held constant, actual answers indicated that the preference for the positive verbal probability was cancelled out when participants had all the time they needed.
Even if relieving the time pressure, the framing effect occurred when the verbal probabilities to compare were incongruent. However if they had the same probabilistic meaning the framing effect was cancelled out. We discuss how these results can be integrated.

Should the messenger be shot if s/he was wrong? Judging responsibility of a speaker giving uncertain statements.

co-author: Sarah R. Beck

People prefer to be told about chances by numerical probabilities (i.e., percentages) but to express them with verbal probabilities (e.g., It is likely; see e.g., Erev & Cohen, 1990). Here we tested the possibility that the vagueness of verbal probabilities makes people feel less responsible for their predictions, and therefore prefer them.
Seventy-two students judged a speaker responsibility in eight scenarios. For a third of participants the speaker used percentages while for the two other thirds s/he used verbal probabilities (of positive directionality for one third and of negative for the other).
The communication mode had an effect on responsibility judgements: speakers were judged more responsible when they used negative verbal probabilities. We also found that speakers were held more responsible when the predicted event did not occur, whatever was its chance to happen.
Whether being wrong in your prediction or bad consequences are underlying this last effect will be discussed.

Children's­ judgements­ and decisions under verbal uncertaint­y

poster, co-author: Sarah R. Beck

Little is known about how children understand verbal probabilities and make decisions based on them. Gourdon and Villejoubert (2009) found that 8-year-old children were able to make decisions based on verbal probabilities in the same way as adults (i.e. relying on both the likelihood and the directionality), though they judged the likelihood of events described by these verbal probabilities based only on the directionality. Aiming to replicate these findings we asked 23 7-year-olds and 24 8-year-olds to evaluate verbal probabilities expressed in daily situations. Results indicated that 7-year-olds were influenced by directionality when judging expected value, whereas 8-year-olds were affected when judging likelihood. When making decisions, the directionality had no effect. Surprisingly, the likelihood had no effect on responses. We discuss potential differences between the Gourdon and Villejoubert’s task and ours, and future research.

Costs of thinking about verbal probabilities: Is processing directionality easier than processing probabilistic meaning?

poster, co-author: Sarah R. Beck

Uncertain outcomes can be described by raw probabilities (e.g., "There is 40% chance"), and also by verbal probabilities (e.g., "There is a chance", "It is not absolutely certain"). Beyond their probabilistic meaning verbal probabilities also have a directionality (Teigen & Brun, 1995), i.e. can be positive or negative. In this study we made the first investigation into the potential differences in processing directionality and probabilistic meaning. Twenty participants chose between two outcomes described by verbal probabilities. In one third of the trials the probabilistic meaning was controlled and the directionality was varied. In another third the directionality was controlled and the probabilistic meaning was varied. In the last third both dimensions were different, reinforcing each other (congruent trials; e.g., a positive one carrying a high probabilistic meaning) or contradicting each other (incongruent trials; e.g. a negative one carrying a high probabilistic meaning). Accuracy, directionality of the actual answer and response time were recorded. When both dimensions differed, accuracy and response time indicated that participants found it easier to choose between congruent than between incongruent verbal probabilities. When one dimension or
the other was held constant and the other varied, we found evidence that participants were influenced by both probabilistic meaning and directionality. Participants found easier to make decisions between positive verbal probabilities and between ones of high probabilistic meaning than
between negative verbal probabilities and between ones of low probabilistic meaning. Finally when the probabilistic meaning was held constant, participants tended to choose the outcome with the positive verbal probability more often than chance. We discuss the possibility of a positivity bias
influencing the process of verbal probabilities.

A comparison of children and adults’ judgements and decisions based on verbal uncertainty statements

co-author: Gaëlle Villejoubert

Children distinguish less well than adolescents the numerical meaning conveyed by verbal probabilities (e.g., Mullet & Rivet, 1991). Little is known, however, about children’s ability to grasp the directionality of verbal probabilities (Teigen & Brun, 1995). We expected children to only be influenced by directionality and congruence of statement framing with their goal. Thirty children and 29 adults made probability judgements and decisions in a treasure hunt context. Results revealed that children are sensitive to the numerical meaning of verbal probabilities in decisions, and also in probability judgements related to goal-incongruent statement framings. The different demands implied by judging probabilities and decision-making will be discussed, as well as the independence of directionality and numerical value in adults’ interpretation of verbal probabilities.

Should the messenger be shot if something wrong happens? Judging responsibility of a speaker giving uncertain statements.

co-author: Sarah Beck

The messenger should be shot only if he was wrong; Judging responsibility of a speaker giving uncertain statements

co-author: Sarah R. Beck

It is not uncertain that taking your time reduces framing effects in verbal probabilities

co-author: Sarah R. Beck

Are French children more rational than English children? It depends on how you look at the data

Much of what we communicate to children about uncertainty is through verbal descriptions. Yet, to our knowledge, only two studies have examined children’s comprehension of verbal probabilities (Mullet & Rivet, 1991; Watson & Moritz, 2003). Furthermore these studies did not take into account the directionality of verbal probabilities (Teigen & Brun, 1995).
In a first study (Gourdon & Villejoubert, in press), we manipulated the directionality and the likelihood of verbal probabilities, and the framing of the event subject of the uncertainty. We found that 8-year-olds took into account the directionality and the value of the uncertain event when judging verbal probabilities, but not the likelihood. However we also found that the three dimensions were taken into account by children when making decisions.
Aiming to replicate these results with an English speaking sample we used only positive framing of the event and a more realistic context. We found that the English sample was not influenced by the likelihood when decision making, unlike the French children in the first study. Furthermore, the effect of the directionality on probability judgement, expected value judgement and decision making was inconsistent.
However we also found in both studies that decision-making could be predicted by judgements of expected value, i.e. French and English children make consistent decisions, even if English children do not efficiently rely on the information communicated to them.
The two methodologies we used were somewhat different. The first paradigm involved repeated trials with the same context and verbal probabilities were highlighted. In the second study, we used different contexts and integrated the verbal probabilities into the story. This may have increased memory demands. The use of different scenarios gave rise to an unexpected effect in the decision-making task where the consequences of the risky choice could have overridden the effect of the directionality and the numerical value. Future research will investigate the impact of these methodological differences by comparing the two tasks in the same sample.

Are French children more rational than English children?

co-author: Sarah R. Beck