Sustainability ratings help consumers understand the environmental impact of their purchases. Such ratings have increased the consumers’ sustainable choices in the electrodomestics and housing markets. In the particular case of energy labels, sustainable products are also associated with private benefits due to future cost reductions in energy expenditure. These results question the potential effectiveness of sustainability ratings for other products, such as food, where the link between environmental and private benefits is less clear. In two incentivized experiments (N=749), we study how consumers use sustainability ratings when these ratings are dissociated from private benefit, i.e. product quality. Participants chose between two products based on their quality and sustainability, which were presented in separate rating scales, alongside the products’ prices. Furthermore, we study how consumers integrate the usage of ratings with other information provided from other sources. Halfway through the experiment, we provide information regarding the underlying value behind the ratings. Using a between-subject design, we modify the information provided and analyze the impact of such information on the participants’ subsequent choices. Our findings indicate that even when sustainability ratings are not connected to the products’ quality, participants make use of them to decide which products to buy. We also find that participants underreact to new information, and make inefficient choices based on their decisions from before. Moreover, to track the participants’ attention and analyze potential heterogeneous usage of the information we use process-tracing methods. We find that participants show highly heterogeneous attention patterns, which are linked to differential weighting of the product’s attributes (price, quality, and sustainability) during the decision. While our information treatment has little effect on attention allocation to individual attributes, participants correctly recall the information at the end of the experiment. These results suggest that participants partially neglect new information, and anchor to their initial decision rules formed before the information treatments.

Keywords: Attention, Sustainability ratings, conjoint analysis, information treatments, MouselabWeb

We examine the role of intermediaries in the hiring process, such as head hunters and recruitment agencies, and how these intermediaries may contribute to discrimination based on (biased) beliefs about the employers. For most large corporations, there are typically two approaches to hiring: standardized processes managed by the HR department or the outsourcing of this responsibility to recruitment agencies or headhunters. These intermediaries play a crucial role by providing the employer with a curated list, recommendations, or ratings of job applicants.
However, intermediaries might not only consider which applicant is a good fit for the job, but also whether the applicant matches the firm’s “taste”. Intermediaries are driven to find candidates who are more likely to be hired, and as a result, they might anticipate and perpetuate discrimination in the recruitment and selection processes if they expect managers to engage in discriminatory practices.
In our research, we conduct three pre-registered experiments to explore how intermediaries predict the performance of job candidates depending on who evaluates them. Using a database from Kausel et al. (2012), we developed an incentivized paradigm where participants had to evaluate real candidates in a job application process based on information obtained during the application process. Specifically, participants observe standardized scores of candidates’ General Mental Ability, Conscientiousness (from Big 5), the score on an unstructured interview and the gender of the candidate. Additionally, we add a random score that is completely unrelated to the actual selection process and participants are aware of this. The candidates in our sample were all hired by the company and they had a performance evaluation three months after being hired.
In our first experiment, we compared how participants predicted the performance of the different candidates and how they expected others to make the same predictions (first-order beliefs). Participants observed all the candidate’s characteristics and then predicted their performance ratings. We recorded their visual attention using a Gazepoint 3 HD eye-tracker. Participants rate 32 candidates with different attribute compositions. After participants have rated all candidates, we ask them to predict how other people, that performed this experiment before, rated the same candidates. We incentivize their beliefs using the method from Krupka & Weber (2013). Our pre-registered hypotheses were that (1) individuals predict differently compared to what they expect others to do (2) any changes in the relative weighting of the different attributes will be accompanied by a shift in attention to said attributes.
Our results show that participants do not discriminate when predicting performance, but do expect that others will discriminate against women. Additionally, we find that differences in the relative weighting of the different attributes is correlated with their level of attention.
In our second experiment, we explore further whether the expected bias coming from others depends on the others’ individual characteristics. In this experiment, we use a modification of the previous paradigm. We present participants with the information of one candidate and ask them for their predictions of how six previous participants (evaluators) scored that given candidate. The six evaluators differ themselves on their gender (Man/Woman) and their level of their math skills, measured by the Berlin numeracy task (divided into high, mid or low math skills). Our pre-registered hypotheses is that participants will predict differently (i.e. different relative weights to the attributes) for men and women; and for the different math skills.
The results show an expected bias towards women for all six evaluators, replicating our results from the first experiment. Moreover, we find that (1) men are expected to show a greater bias against women and (2) people with high math skills (compared to mid and low) put a higher weight on general mental ability (capacity to process information such as graphs and tables). These results are consistent with in-group out-group behaviour. Post-hoc analyses show that participants expect that others with their same gender will have less of an in-group bias than the opposite gender.
We are currently developing our third experiment, which is based on the results of our second experiment. In this experiment, we use a similar paradigm to the one in our first experiment. In this experiment, participants will predict the evaluations of multiple candidates for only one evaluator at the time (second-order beliefs). After participants have answered their second-order beliefs for one candidate, we subsequently ask them about the same candidates for a different evaluator. The aim of this experiment is twofold. First, we want to do a confirmatory analysis for the results found in study two. Namely, are men and women expected to predict differently for other men and women? Do people expect higher math-skill evaluators to overweight the general mental ability when compared to evaluators with lower math skills? We will also further explore our exploratory results: do women (compared to men) and high-skilled math participants (compared to low-skilled) predict less in-group biases?
Secondly, by using eye-trackers, we aim to trace the participants behaviour and answer in this case whether (1) more attention to an attribute correlates with a higher weight of said attribute on the decision; and (2) is there a shift in attentional focus when the participants predict for different evaluators?
Our current and future results inform about the role of intermediaries in the hiring markets; and how intermediaries can enhance the levels of discrimination in said market. We also provide insightful information of whether shifts in attention are able to predict shifts in behaviour.

Keywords: Recruitment and Selection, Beliefs, Gender discrimination

Does attention have a causal impact on risky decisions? We address this question in a preregistered experiment in which participants accept or reject a series of mixed gambles while exogenously varying how information can be sampled. Specifically, in each trial participants observe the outcomes of a mixed-gamble with gains and losses presented sequentially. To isolate the causal role of attention on the decision process, we manipulate for how long each outcome is presented before showing the next one. Our results partially confirm our preregistered hypotheses that longer exposure to an outcome increases its weight on the decision. We find specific effects in the domain of losses, but not gains. A longer presentation duration of losses leads to increased sensitivity for losses, such that lotteries with higher losses are rejected more often when losses are presented for longer. To our surprise, when gains are presented for longer, the participants show increased sensitivity to both gain and loss values in their decision. Further analyses show that specifically participants with higher impulsiveness become more sensitive to outcome values when gains are presented for longer. Attentional impulsiveness is the strongest driver of this effect. Jointly, these results support the notion that attention has a causal impact on risky choice. Moreover, our results underline the moderating role of impulsiveness on the relationship between attention and choice.

Keywords: Attention, Impulsiveness, Loss Aversion, Random Utility Models.
JEL Classifications: D81, D83, D87, D91

Gambling disorder (GD) is a behavioural addiction characterized by impairments in decision-making, favouring risk- and reward-prone choices. One explanatory factor for this behaviour is a deviation in attentional processes, as increasing evidence indicates that GD patients show an attentional bias toward gambling stimuli. However, previous attentional studies have not directly investigated attention during risky decision-making. 25 patients with GD and 27 healthy matched controls (HC) completed a mixed gambles task combined with eye-tracking to investigate attentional biases for potential gains versus losses during decision-making under risk. Results indicate that compared to HC, GD patients gambled more and were less loss averse. GD patients did not show a direct attentional bias towards gains (or relative to losses). Using a recent (neuro)economics model that considers average attention and trial-wise deviations in average attention, we conducted fine-grained exploratory analyses of the attentional data. Results indicate that the average attention in GD patients moderated the effect of gain value on gambling choices, whereas this was not the case for HC. GD patients with high average attention for gains started gambling at less high gain values. A similar trend-level effect was found for losses, where GD patients with high average attention for losses stopped gambling with lower loss values. This study gives more insight into how attentional processes in GD play a role in gambling behaviour, which could have implications for the development of future treatments focusing on attentional training or for the development of interventions that increase the salience of losses.

Keywords: gambling disorder; loss-aversion, attentional bias; risk; reward; eye-tracking.

We investigate the role of visual attention in risky choice in a rich experimental dataset that includes eye-tracking data. We first show that attention is not reducible to individual and contextual variables, which explain only 20\% of attentional variation. We then decompose attentional variation into individual average attention and trial-wise deviations of attention to capture different cognitive processes. Individual average attention varies by individual, and can proxy for individual preferences or goals (as in models of “rational inattention” or goal-directed attention). Trial-wise deviations of attention vary within subjects and depend on contextual factors (as in models of “salience” or stimulus-driven attention). We find that both types of attention predict behavior: average individual attention patterns are correlated with individual levels of loss aversion and capture part of this individual heterogeneity. Adding trial-wise deviations of attention further improves model fit. Our results show that a decomposition of attention into individual average attention and trial-wise deviations of attention can capture separable cognitive components of decision making and provides a useful tool for economists interested in choice.

Keywords: Attention, Random Parameter Models, Eye-tracking, Loss Aversion
JEL Classifications: D81, D83, D87, D91

Behavioral strategy research aims to understand and improve strategic decision- making processes by analyzing managers’ psychological traits and cognitive activities. Our work establishes a bridge between behavioral strategy and cognitive neuroscience by performing an experimental design that uses eye-tracking analysis to understand the effect of visual attention on heuristics about managerial decisions. Our experimental approach allows researchers to show and quantify the effect of visual attention on repeated managerial decisions. Our analysis identify some attentional sources of decision-making efficiency, which are primarily driven by selectivity, as decision-makers learn to focus on fewer pieces of information to make a managerial decision. Moreover, each of these pieces of information is accessed fewer times as participants gain experience with the task. In addition, we show that incorporating attentional information into a heuristic improves the explanatory power of managerial decisions.

A signiffant gender gap in the labor market has been observed in almost ev- ery country (World Economic Forum, 2021). Existing studies have shown that gender differences in academic performance, career choices, and preferences can partly explain the gender gap (Blau and Kahn, 2017). Women and men per- form differently in several experimental tasks involving math, logic, and verbal skills (e.g., Gneezy et al., 2003; Niederle and Vesterlund, 2010, 2011; Shurchkov, 2012). Additionally, several studies document the gender differences in compet- itiveness, risk preferences, and social preferences (Croson and Gneezy, 2009). These results suggest that payment schemes affect men and women differently. Men are more responsive to competitive payment schemes (e.g., Gneezy et al., 2003) whereas women are more responsive to pro-social incentives (e.g., Tonin and Vlassopoulos, 2015). The goal of this project is to study whether payment schemes can be used to tackle the gender gap in the labor market and lead to more equality. We achieve this goal by answering three questions. Do payment schemes affect the performance of men and women differently in a given task? Do these payment schemes affect the beliefs about the genders’ performance as well? Finally, do these changes vary with the gender stereotype associated with the task? 1