Glass Walls: Experimental Evidence on Access Constraints Faced by Women

Last registered on May 04, 2020


Trial Information

General Information

Glass Walls: Experimental Evidence on Access Constraints Faced by Women
Initial registration date
April 03, 2019
Last updated
May 04, 2020, 7:25 PM EDT



Primary Investigator

Harvard Kennedy School

Other Primary Investigator(s)

PI Affiliation
Princeton University
PI Affiliation
Lahore University of Management Sciences
PI Affiliation
Lahore University of Management Sciences

Additional Trial Information

Start date
End date
Secondary IDs
Growth is enabled when individuals can access the opportunities offered to them. Yet individuals, especially women, often face significant barriers in doing so. This paper provides evidence on such access constraints in the context of skill acquisition. Using experimental evidence from rural Pakistan, we show that physical distance poses a significant hurdle. Women whose villages are randomly selected to receive a training center are more than three times as likely to enroll and complete a skills development course than women who have to travel an average distance of just a few kilometers. Over half of this penalty is paid simply upon crossing the village boundary and therefore cannot be readily reconciled with time or economic costs associated with travel. Instead, this “boundary effect” is likely due to non- economic/social costs women face when leaving the perceived safety of their villages. This constraint is costly to compensate. Using exogenous variation in stipend offered, we estimate that an amount equivalent to half of monthly household expenditure would need to be paid to allow women to cross this boundary. In examining factors that may ameliorate this barrier, we find that the boundary effect is lower for women who come from more ethnically diverse communities. While informational and social interventions have little impact, providing reliable group transportation goes a long way in addressing this access constraint. This suggests that while non-economic obstacles faced by women are indeed substantial, policy interventions attuned to the local context can offer feasible ways to ameliorate them.
External Link(s)

Registration Citation

Cheema, Ali et al. 2020. "Glass Walls: Experimental Evidence on Access Constraints Faced by Women." AEA RCT Registry. May 04.
Former Citation
Cheema, Ali et al. 2020. "Glass Walls: Experimental Evidence on Access Constraints Faced by Women." AEA RCT Registry. May 04.
Experimental Details


Intervention Start Date
Intervention End Date

Primary Outcomes

Primary Outcomes (end points)
Voucher acceptance, voucher submission, course enrollment, course completion
Primary Outcomes (explanation)

Secondary Outcomes

Secondary Outcomes (end points)
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
Our study is part of our collaboration with the Punjab Skills Development Fund (PSDF). In 2011 and 2012, we conducted a baseline survey of over 11,000 households and discovered a huge latent demand for skill acquisition. Subsequent pilot training programs and follow-up surveys revealed five potential access constraints underlying a wide gap between expressed demand and actual take-up: distance, information, social norms, reliable transportation and money.

This pre-analysis plan focuses on additional heterogeneity analysis for a program entitled Skills for Market phase B (SFM-B), which rolled out in 2013-14. We randomly selected 243 villages from our original baseline survey to take part in the program and receive training in sewing and tailoring—the most demanded skills among rural women. In addition, with the five access constraints in mind, we coupled the training with the following treatments to encourage program take-up:

1. Distance: We selected 108 villages to house a training center. We refer to these villages as Village Based Training (VBT) villages and the rest as non-Village Based Training (nVBT) villages.

2. Information: Because people might not fully understand the benefits of a skills training program, we randomly selected 66 villages in which we conducted hour-long all-female trainee engagement (TE) sessions to disseminate program information.

3. Social norms: Because restrictive social norms might discourage women from accessing training programs, we conducted community engagement (CE) sessions separately for males and females in 81 villages. Building on basic program information that TE conveyed, CE sessions discussed not only the social challenges women face to accessing the training but also how community members could help women overcome such challenges.

4. Reliable transportation: Because a lack of safe and reliable transportation compounds the physical distance constraint, we arranged group transportation (GT) for 54 nVBT villages to the corresponding training centers.

5. Money: The program sought to lessen the monetary constraint by offering a monthly stipend to all trainees for attending the training. We offered every household a base stipend of Rs. 1,500 and randomly allocated a stipend top-up. These additional stipends were as high as 4,500 PKR. To minimize fairness concerns, we introduced the stipend amount at both the village and the household level such that stipends did not vary by more than 1,000 PKR across households within a village.
Experimental Design Details
Randomization Method
Randomization done by a computer.
Randomization Unit
We randomized at the village level.
Was the treatment clustered?

Experiment Characteristics

Sample size: planned number of clusters
Our analysis currently includes 243 villages.
Sample size: planned number of observations
About 6,390 women from rural Punjab, Pakistan.
Sample size (or number of clusters) by treatment arms
108 villages with village-based training centers: 27 with training centers and community engagement treatment; 39 with training centers and trainee engagement treatment; and 42 with training centers only.

135 villages without village-based training centers: 27 with trainee engagement treatment, 27 with community engagement treatment, 27 with group transport treatment, 27 with both group transport and community engagement treatments, and 27 without any additional treatment.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)

Institutional Review Boards (IRBs)

IRB Name
Harvard University - Area
IRB Approval Date
IRB Approval Number
Analysis Plan

Analysis Plan Documents

042019_Glass_Walls_PreAnalysisPlan_v6 _Final_.pdf

MD5: 5b34fdb8954af8f014996ee6a5446c47

SHA1: 3e5aac50b0b74303163fdc5577ae15dcc2db09a8

Uploaded At: April 03, 2019


Post Trial Information

Study Withdrawal

There are documents in this trial unavailable to the public. Use the button below to request access to this information.

Request Information


Is the intervention completed?
Data Collection Complete
Data Publication

Data Publication

Is public data available?

Program Files

Program Files
Reports, Papers & Other Materials

Relevant Paper(s)

Reports & Other Materials