Trust and Authority in Civil-Military Relations: Experimental Evidence from Germany

The study is a two-part online experiment. In the first part, participants complete a Trust Game in which roles are randomly assigned and anonymous pairings are created ex post. Officer cadets and civilian students are matched within and across groups, and the design allows the measurement of trust (sender transfers) and trustworthiness (receiver returns) under in-group and out-group conditions. In the second part, participants (military and civilian) take part in an authority-related allocation task in which they are randomly assigned to either an authority treatment or a control condition. This part of the study aims to investigate whether an authority successfully influences allocation decisions and whether the effect differs between military and civilian participants.

The study combines two experimental parts that are presented in randomized order: a Trust Game and an authority-related allocation task.
In the Trust Game module, participants are randomly assigned to the role of sender or receiver. The sender receives an endowment of 10 points and chooses a transfer between 0 and 10 points. The transferred amount is multiplied for the receiver, and second-mover responses are elicited via the strategy method. Matching is anonymous and ex post. The design includes three matching conditions at the university level: University of Hamburg–University of Hamburg, University of Hamburg–Helmut Schmidt University, and Helmut Schmidt University–Helmut Schmidt University. This allows the comparison of trust and trustworthiness across officer cadets and civilian students in in-group and out-group settings.

In the authority game, participants in the role of the decision-maker have to allocate 10 experimental points between themselves and a randomly assigned participant from the study. They have to choose between two options: Option 1, the fair option, which allocates 5 points to each of the two participants; and Option 2, the unfair but payoff-maximizing option, which allocates 9 points to the decision-maker and 1 point to the other participant. Participants are randomly assigned to an authority or control condition. In the authority treatment, the experimenter in the role of an external authority explicitly recommends participants to choose the payoff-maximizing but unfair allocation. In the control condition, the recommendation is not provided. Furthermore, sociodemographic variables are collected to allow differentiation between the decisions of military and civilian participants. In sum, the second part of the study employs a 2x2 between-subjects design.

2026-04-27

2026-06-22

Part 1: Trust Game primary outcomes:
The primary outcome variables are (i) the sender’s transfer 𝑥 ∈ {0, … ,10} from an initial endowment of 10 points and (ii) the receiver’s conditional return 𝑦(𝑥), elicited via the strategy method for each possible transfer level. Following Johannemann et al. (2016), the transferred amount is multiplied by a constant factor (e.g., tripled) before reaching the receiver, and the receiver’s return determines the final payoffs of both players. Expected payoffs are thus a direct function of the sender’s transfer and the receiver’s return decision.
The main endpoints are the average transfer 𝐸[𝑥] and the average return ratio, where 𝑚 denotes the multiplication factor. These outcomes are compared across group constellations (OA–OA, OA–Ziv, Ziv–Ziv) to test for (i) a general group effect in sender trust, (ii) in-group effects, and (iii) out-group transfer effects. Correspondingly, receiver behavior is analyzed via average conditional returns across these constellations to test for differences in trustworthiness.

In line with the hypotheses of the original study, higher transfers and higher return ratios are expected for officer cadets, particularly in in-group interactions, with a weaker but positive extension to out-group interactions. Expected earnings (ex-ante and realized) serve as an additional aggregated endpoint reflecting joint cooperative outcomes.

As a supplementary outcome, a lottery-based risk elicitation task is included. Individual risk preferences, measured via a discrete lottery choice (e.g., safe vs. risky payoff), are used to control for heterogeneity in sender behavior, as transfer decisions may partly reflect risk attitudes rather than pure trust. The lottery choice is not a treatment but enters the analysis as a control variable affecting observed trust behavior.

Part 2: Authority Game Primary outcomes: Allocation decision (Option 1 vs. Option 2).
The primary outcome is the participant’s binary allocation decision between a fair and an unfair payoff distribution.

The key comparisons test whether (i) authority increases the likelihood of choosing the payoff-maximizing but unfair option, and (ii) whether this effect is stronger among officer cadets than among civilian students.

The second part of the experiment tests three hypotheses:
Hypothesis 1: An authority recommending the unfair allocation increases the share of unfair decisions (i.e., the unfair option) compared to the control condition.
Hypothesis 2: In the absence of an authority, military and civilian participants do not differ significantly in their decision to choose the unfair option.
Hypothesis 3: Under an authority, military participants are more likely than civilian participants to choose the unfair option.

Part 2 collects additional data through additional questions following the allocation decision and the post-experimental questionnaire. These include participants’ first and second order belief, perceptions of the decision environment, and assessments of the authority situation. In addition, sociodemographic characteristics such as age and gender are elicited. The questionnaire further captures participants’ experiences with and perceptions of individuals with a military background.

The study consists of a Trust Game and an Authority Game. Participants are officer cadets or civilian students. Additional survey items capture perceived security developments, "Zeitenwende" perceptions, and sociodemographic characteristics. In the first part of the experiment, the design follows a conceptual replication of Johannemann et al. (2016) and extends it with contextual and subgroup analyses. The second part of the experiment examines whether military and civilian individuals differ in their responsiveness to an authority in a morally relevant allocation decision.

The study combines two experimental parts, both implemented online and presented in randomized order across participants.

In the Trust Game, participants are randomly assigned to sender or receiver roles. Senders receive an endowment of 10 points and choose a transfer between 0 and 10 points; the transferred amount is multiplied before reaching the receiver. Receiver responses are elicited via the strategy method for all possible incoming amounts. Matching is anonymous and conducted ex post. Participants are assigned to interaction partners such that approximately 50% of decisions occur in in-group constellations (same university) and 50% in out-group constellations (different university). This ensures balanced variation across the three relevant dyad types (University of Hamburg–University of Hamburg, University of Hamburg–Helmut Schmidt University, Helmut Schmidt University–Helmut Schmidt University).

The second part of the experiment follows a 2×2 between-subjects design combining authority instruction (authority vs. no authority) with participant socialization (military vs. civilian). Participants make a single allocation decision between a fair distribution (5 points for the decision-maker and 5 points for another participant) and an unfair, payoff-maximizing distribution (9 points for the decision-maker and 1 point for another participant).
The second part of the experiment tests three hypotheses:
Hypothesis 1: An authority recommending the unfair allocation increases the share of unfair decisions (i.e., the unfair option) compared to the control condition.
Hypothesis 2: In the absence of an authority, military and civilian participants do not differ significantly in their decision to choose the unfair option.
Hypothesis 3: Under an authority, military participants are more likely than civilian participants to choose the unfair option.

Randomization by computer (treatment, group, role).

Individual level.

No

none

300 participants (approximately 150 from the University of Hamburg and 150 from the Helmut Schmidt University of the Bundeswehr).

Trust Game: approximately equal allocation to in-group and out-group interactions (about 50% each), with participants drawn from roughly 150 officer cadets and 150 civilian students.
Authority game:
150 participants control (75 civil and 75 military).
150 participants authority treatment (75 civil and 75 military).

With approximately 300 participants and balanced assignment across groups and treatments, the study is designed to detect moderate effect sizes in the main outcomes; a conservative benchmark for the primary comparisons is approximately Cohen’s 𝑑 ≈ 0.4–0.5.