One Size Fits All? A Systematic Study of Heterogeneous Effects of Self-Affirmation on Task Performance
How to cite
Tonke, Sebastian; Schneiders, Sebastian. 2025. "One Size Fits All? A Systematic Study of Heterogeneous Effects of Self-Affirmation on Task Performance." AEA RCT Registry. AEARCTR-0017302. https://doi.org/10.1257/rct.17302-1.1
Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.
* Primary investigator
Abstract
We examine whether the effectiveness of self-affirmation (SA) interventions varies across gender, socioeconomic status (SES), age, and ethnicity. Self-affirmation exercises are widely used in workplaces, (female) leadership programs, and educational settings to support resilience, reduce stress, and improve performance. However, the extent to which SA benefits different demographic groups remains an open question with significant implications for theory and practice. Typically, the benefits of self-affirmation interventions are found for disadvantaged, underrepresented or negatively stereotyped groups. We conduct a high-powered online experiment in which participants are randomly assigned to one of 16 versions of a SA exercise or an active control condition, followed by an incentivized real-effort task to measure productivity. We focus on heterogeneity by gender, SES, age, and ethnicity because these characteristics shape exposure to identity-related pressures, norms, and baseline levels of psychological threat, which are mechanisms directly linked to how SA is theorized to operate. Studying these moderators helps identify when and for whom such interventions are effective, thereby improving the design and scalability of self-affirmation interventions.
Intervention
We estimate heterogeneous treatment effects (gender, SES, age, ethnicity) of self-affirmation interventions.
Our interventions comprise 16 different SA inductions (2x2x2x2 design). We compare the (i) type of self-affirmation (personal experiences vs. values), (ii) whether participants have to write about one value or experience or as many as they like (one vs. many), (iii) whether their anonymously written content is shared with the researchers or not (public vs. private), and (iv) whether the purpose of the self-affirmation exercise is revealed or not (purpose vs. no purpose). Subsequently, participants can work on a real-effort task with substantial piece-rate incentives, in which they have to count the number of zeros in a series of 3x10 matrices filled with randomly generated zeros and ones for a maximum of ten minutes.
Experimental design
Subjects are randomly assigned to a control condition (2 versions) or a treatment condition (16 versions).
Our first set of hypotheses concerns whether there are heterogeneous treatment effects across gender, socioeconomic status (SES), age, and ethnicity for the pooled 16 treatment conditions.
1. Primary Heterogeneity Hypotheses:
H1 (Gender). Self-affirmation will yield larger improvements in task performance for women than for men.
H2 (Socioeconomic Status). Self-affirmation will yield larger improvements in task performance for lower-SES individuals than for higher-SES individuals.
H3 (Age) Self-affirmation will yield larger improvements in task performance for younger individuals than for older individuals.
H4 (Ethnicity). Self-affirmation will yield larger improvements in task performance for individuals from underrepresented or negatively stereotyped ethnic groups than for individuals from majority ethnic groups.
We further hypothesize that any observed heterogeneity in H1-H4 will be consistent with survey evidence from control-condition participants (e.g., women reporting lower confidence or greater evaluative pressure during the task).
2. SA Dimension-Oriented Heterogeneous Treatment Effects
Our second set of hypotheses concerns whether the four design dimensions of the SA interventions systematically interact with participant characteristics. Specifically, we examine gender-, SES-, age-, and ethnicity-based differences in responsiveness to variations in:
a. Affirmation type: personal experiences vs. personal values
b. Content breadth: writing about a single value/experience vs. multiple ones
c. Disclosure: content shared with researchers vs. kept private
d. Transparency: purpose of the exercise revealed vs. not revealed
H5 a-d (Design Interactions). The effectiveness of self-affirmation interventions will vary across demographic groups (gender, SES, ethnicity) depending on the content, breadth, disclosure, and transparency of the SA induction.
We expect that higher psychological ease and higher psychological engagement make SA interventions more effective. Hence, we hypothesize that any observed demographic differences in H5 will be consistent with changes in psychological ease (e.g., comfort, ease) and psychological engagement (e.g., self-relevance, meaningfulness) elicited in each SA intervention.
Design details
Cited by
No citation data available for this registration yet.