Social initiatives and stakeholder perceptions

Participants are randomized individually into four conditions. In all conditions, they are presented with a hypothetical company in the retail sector. The treatments are then administered through two successive vignettes where the first varies the introduction of a diversity and inclusion (DEI) programme in the company across conditions, and the second its discontinuation. Specifically, the conditions are as follows:

• Never DEI condition (T0 – control): The first vignette states that the company considered introducing a DEI programme, but decided not to. The second vignette states that the company reviewed its employee policies and decided to continue without a DEI programme.
• Start and maintain DEI condition (T1): The first vignette states that the company considered introducing a DEI programme, and decided to do so. The second vignette states that the company reviewed its employee policies and decided to continue its DEI programme.
• Non-political DEI reversal condition (T2): The first vignette is identical to condition T1. The second vignette states that the company reviewed its employee policies and decided to discontinue its DEI programme for economic (non-political) reasons.
• Political DEI reversal condition (T3): The first vignette is identical to condition T1. The second vignette states that the company reviewed its employee policies and decided to discontinue its DEI programme due to political pressure.

Please see pre-analysis plan for additional details.

2026-03-23

2026-03-27

Stakeholder perceptions of a company

See pre-analysis plan for details.

We will perform an online randomized controlled trial studying how social initiatives affect stakeholder perceptions of companies. Information on company social initiatives will be varied across one control condition and three treatment conditions, using vignettes.

We plan on recruiting about 400 participants via Prolific in each treatment arm for this experiment. In total, this amounts to 1600 participants after implementing the below prespecified exclusion criteria. The study will be programmed in Qualtrics and distributed via a Qualtrics link. We will use a representative United States sample from Prolific based on the factors political affiliation, sex and age.

Through vignettes, we present participants with a fictional company and provide general background information about the firm. We employ a 1×4 experimental design to randomize information regarding a corporate programme aimed at ensuring diversity and inclusion in the companies hiring, promotions, and work culture. Specifically, we manipulate the status of this programme across four distinct conditions: one where the company has never implemented such a programme (the control), another where they started and continued it despite political controversy, a third where they started but discontinued it for economic/non-political reasons, and a final condition where the programme was discontinued due to political pressure .

After the vignette treatments, we compare willingness to work for the company across the four conditions.

Please see attached pre-analysis plan for additional details,

Randomization by computer.

Individual

No

1600 individuals

1600 individuals

400 individuals in the control group, and 400 individuals in each of the three treatment groups.

Please see attached pre-analysis plan.

Pre-analysis plan Social initiatives and stakeholder perceptions

After a number of years in which corporations have increased their use of social initiatives and internal diversity policies, corporations are now facing pressure to discontinue such policies, in particular in the United States. How this affects corporate incentives to introduce, maintain or reverse such policies depends on stakeholder reactions. We present results from a well-powered, pre-registered, online vignette experiment where we test the effect of reversing diversity, equity and inclusion (DEI) policies on respondent willingness to work for a company. A politically representative sample of 1600 US respondents were randomized into four treatment arms. In the first treatment they were presented with a hypothetical company that introduces and maintains a DEI programme, in the second treatment a company that introduces but then discontinues a DEI programme for non-political reasons, in the third treatment the DEI programme is introduced but discontinued due to political pressure, and in the fourth control condition the company never introduces a DEI programme. The results show that respondents exhibit a stronger willingness to work for a company that introduces and maintains a DEI programme compared to the control. However, reversing a DEI programme for non-political reasons leads to worse employee perceptions than never having a programme in the first place. Moreover, this negative effect is significantly greater if the programme is discontinued due to political pressure. While reversing a DEI programme is adversely related to perceptions of professionalism of a company, we find that the main mechanism explaining the drop in stakeholder perceptions is a perceived lack of authenticity. Results are also highly heterogeneous by political preference, liberal respondents favour DEI and disfavour DEI reversal, while conservative respondents oppose DEI programmes and are not mollified by their reversal.

Berge, Joel Wenæs and Berge, Lars Ivar Oppedal and Kolstad, Ivar, The DEI Reversal Penalty (August 25, 2026). Available at SSRN: https://ssrn.com/abstract=7348300 or http://dx.doi.org/10.2139/ssrn.7348300

https://papers.ssrn.com/sol3/papers.cfm?abstract_id=7348300