Present and Future Temporal Framing and Their Effects on Hope, Hopelessness, and Entrepreneurial Outcomes: A Randomized Field Experiment

The intervention is a training session on women's leadership and climate resilience, delivered as the second module of a two-module entrepreneurship training program. The session covers the same substantive content across conditions. What varies is how that content is framed throughout the session.

Participants assigned to the "present" condition experience the session through the lens of what they control right now in their business. Cases, examples, and discussions are selected and framed to reflect current business situations and present-focused decision making. Participants assigned to the "future" condition experience the session through the lens of where they are taking their business and what they can do in the future to get there. Cases, examples, and discussions are selected and framed to reflect forward business planning and future-focused decision making. In both conditions, the trainer opens and closes the session by explicitly orienting participants to the relevant temporal frame.

Participants assigned to the control condition do not receive the second training module during the data collection period. Instead, they complete a structured review activity based on the first module. They receive the full training after data collection is complete.

2026-06-16

2026-07-31

The primary outcomes of interest are hope and hopelessness, each measured as a change score (post-training minus pre-training):
1. Hope: current feelings of hopefulness about goals and the future, measured using an adapted version of the State Hope Scale (Snyder et al., 1996)
2. Hopelessness: current feelings of hopelessness about goals and the future, measured using an adapted version of the State Hopelessness Scale (Dunn et al., 2014)

Hope and hopelessness are both state-based measures explicitly designed to capture how a person feels at a given moment, making them well-suited to a change score operationalization. Change scores are calculated as the difference between post-training and pre-training survey responses (t2 post minus t1 pre). A positive change score indicates an increase from baseline; a negative change score indicates a decrease. This is the primary operationalization for both variables.

As a secondary analysis, post-training scores will also be examined as outcome variables with pre-training scores included as a covariate. This is of particular interest given that hope and hopelessness may reflect deeply embedded emotional responses to chronic crisis conditions that persist across the study period, and for which the exit score with baseline as covariate may be a more sensitive operationalization.

Hope and hopelessness are scored as means of their constituent items.

The secondary outcomes of interest are three entrepreneurial variables, each measured as a change score (post-training minus pre-training):
1. Entrepreneurial resilience: the capacity to recover and adapt in the face of business adversity, measured using an adapted version of the Connor-Davidson Resilience Scale (Campbell-Sills & Stein, 2007)
2. Entrepreneurial action orientation: the capacity to translate intention into action without rumination or hesitation, measured using an adapted version of the Action-State Orientation Scale (Diefendorff et al., 2000)
3. Entrepreneurial intent: forward commitment to continuing to pursue business goals despite obstacles, measured using a purpose-written scale adapted from the Grit Scale (Duckworth et al., 2007), with intent framing drawn from Sheeran et al. (2005) and obstacle structure from Gollwitzer (1999)

Change scores are calculated as the difference between post-training and pre-training survey responses (t2 post minus t1 pre). A positive change score indicates an increase from baseline; a negative change score indicates a decrease. This is the primary operationalization for all three variables.

As a secondary analysis, post-training scores will also be examined as outcome variables with pre-training scores included as a covariate. This is of particular interest for the entrepreneurial outcome variables given their directional temporal orientations — resilience is backward-looking, action orientation is present-focused, and intent is future-oriented.

All three scales are scored as means of their constituent items after reverse-scoring where applicable.

Participants are women entrepreneurs enrolled in a women's entrepreneurship training program. Target enrolment is 400 participants, allocated approximately 40% to the present condition, 40% to the future condition, and 20% to the waitlist control condition. Participants are randomly assigned at the individual level and training is delivered in cohorts of 25 to 30 participants. Data collection runs June through August 2026.

Each training module is a half-day session. Module 1 and Module 2 are not delivered on the same day. The interval between modules varies by cohort but is consistent at approximately two weeks across all cohorts.

Surveys are administered at three points: before Module 1, and before and after Module 2. This produces a pre-training baseline and a post-training exit measure for all outcome variables, allowing within-person change to be assessed across conditions.

Before each survey, participants complete a brief check to identify whether any significant external events have occurred that might affect their responses. This is particularly important given Haiti's chronic crisis context, where exogenous shocks during the study period could confound results.

At the end of Module 2, participants in the two treatment conditions complete a brief manipulation check confirming whether the temporal framing of the session was perceived as intended.

Participants come from a range of industries and venture stages. Demographic and venture characteristics (e.g., age, education, employee number and type, venture age) are collected at baseline and will be used to confirm that randomization produced comparable groups across conditions. Because participants attend training in groups, intraclass correlation coefficients (ICC) will be used to detect and account for the possibility that people in the same cohort may respond similarly to one another.

Not available

Participants are randomized to conditions using a computer-generated random number generator in Excel. Each participant is assigned a random number and allocated to a condition based on that number. Participants are assigned individually and then grouped into delivery cohorts of 25 to 30 participants until each cohort is full.

Individual. Participants are randomized at the individual level and subsequently grouped into delivery cohorts of 25 to 30 participants. Although randomization is individual, training is delivered in cohorts, and intraclass correlation coefficients (ICC) will be used to account for cohort-level clustering effects in the analysis.

Yes

Approximately 13 to 16 cohorts of 25 to 30 participants each, for a planned total of 400 participants.

400 individual participants (women entrepreneurs enrolled in a women's entrepreneurship training program in Haiti).

160 participants: present temporal framing condition (Treatment A)
160 participants: future temporal framing condition (Treatment B)
80 participants: waitlist control condition

Power calculations were conducted for two planned comparisons using a two-sided t-test with alpha = 0.05 and planned enrollment of 400 participants (160 Treatment A, 160 Treatment B, 80 Control), assuming a standard deviation of 1.00 on a 5-point scale. For the treatment vs. treatment comparison (n=160 vs. n=160), the minimum detectable effect size is a mean difference of 0.35 (Cohen's d = 0.35) at 87.9% power. For the treatment vs. control comparison (n=160 vs. n=80), the minimum detectable effect size is a mean difference of 0.40 (Cohen's d = 0.40) at 83.2% power. The binding minimum detectable effect size is Cohen's d = 0.40, from the treatment vs. control comparison. Note that these calculations do not account for intraclass correlation (ICC) due to cohort-level clustering, and the true minimum detectable effect size may be marginally larger than reported here.