NEW UPDATE: Completed trials may now upload and register supplementary documents (e.g. null results reports, populated pre-analysis plans, or post-trial results reports) in the Post Trial section under Reports, Papers, & Other Materials.
Unintended Consequences of Health Awareness Campaigns: A Field Experiment in Cairo
Initial registration date
August 27, 2019
August 29, 2019 8:43 AM EDT
University of Goettingen
Other Primary Investigator(s)
The Center for Evaluation and Development
Additional Trial Information
We study the effect of health awareness campaigns on cognitive load and behavioral preferences implementing a ﬁeld experiment with a representative sample of adult individuals residing in Cairo, Egypt. Participants in a door-to-door survey were assigned to watch videos about a commonly spread and highly contagious disease aﬀecting children or a placebo video. Consistent with dual process theories, we ﬁnd that the health awareness video impaired cognition of the targeted population in terms of lower scores on cognitive ability and ﬂuid intelligence tests compared with the control group. The results show that participants who experienced ﬁnancial and health stress in their daily life were mostly aﬀected by the treatment. The ﬁndings emphasize the need to account for cognitive taxation when designing interventions and determining target beneﬁciaries.
Ibanez, Marcela and Ghida Karbala. 2019. "Unintended Consequences of Health Awareness Campaigns: A Field Experiment in Cairo." AEA RCT Registry. August 29.
We investigate whether exposure to information has unintended negative consequences on the targeted population. Particularly we ask whether access to health- awareness information can generate cognitive tax among the targeted beneﬁciaries hence inﬂuencing their decision making process. To generate awareness on the causes, consequences, and recommended precautions related to Accute Respiratory Infection, we launched an independent door to door campaign. Adults with and without children were exogenously selected to be exposed to one of two health awareness videos or to a control placebo video. The videos replicated commonly used formats in awareness campaigns and lasted on average 12 minutes.
Intervention Start Date
Intervention End Date
Primary Outcomes (end points)
We consider the impacts on mental clarity and ﬂuid intelligence. In addition we study the impacts on risk preferences in gains and losses, competition preferences, time preferences in the short and long run and inconsistency in time preferences.
Primary Outcomes (explanation)
To test for cognitive functioning we randomly choose a set of three Raven matrices to be solved by each participant. Raven’s test is a common component in IQ tests and is used to measure the capacity to think logically and solve problems independently of previously acquired knowledge and education. Based on the performance of participants, we build an indicator of cognitive load that is equal to the proportion of correctly solved matrices.
To measure aversion to risks in the gain domain, we follow the method suggested by Cardenas and Carpenter (2013). Participants were asked to select their preferred lottery out of six binary lotteries presented in a form of a ring. Each lottery had two possible outcomes with equal probabilities of payment. These payoﬀs were chosen such that their mean and variance increase as one moves clockwise. We calculated the coeﬃcient of risk aversion r based on a constant relative risk aversion utility function.
Time preferences were measured using a simple binary choice-based method. Participants were presented with two choice-lists corresponding to two diﬀerent time horizons. In the ﬁrst exercise participants were presented with a list of options, A and B, where A allows the participant to receive $X in one month while B oﬀers $Y in two months with all the choices showing $Y>$X. The one month delay in payment oﬀered the participant an interest i allowing him to gain Y=X+i in a month. If the subject prefers the 100 EGP today then we can infer that the discount rate is higher than i % per month; otherwise, we can infer that it is x% or less. In the second exercise the participants repeated the same task, however, this time they were presented with two binary choices A and B where A oﬀered $X in one month while B oﬀered $Y in seven months. To test for inconsistency in time preferences across time horizons, we compare the indif-
ference point in each of the above binary choices. We also construct a measure of hyperbolic discounting deﬁned as a preference for lower interest rate in seven months compared with two months.
To measure competition preferences we follow the method suggested by Niederle and Vesterlund (2007). We use an indicator variable on whether the individual self-selected to be paid according to competitive payments instead of a piece rate payment.
Secondary Outcomes (end points)
Secondary Outcomes (explanation)
We used a between-subject experimental design in which participants were exposed to one of three diﬀerent videos.
The health awareness video discussed the causes, symptoms and consequences of acute respiratory Infections (ARIs). This video was presented in one of two formats. The ﬁrst version of the video focused in the long term consequences of ARIs on children’s health. In this version, the information presented emphasized the contagious characteristic of the disease, the problems of late diagnosis and the health complications that could arise on the long run. The second version of the video focused on the ﬁnancial burden associated with
treating the direct and indirect health consequences of ARIs. In this version, we oﬀered precise ﬁgures on the estimated costs of medication and treatment. Both versions comprehensively explained the required precautions in order to reduce vulnerability to the infection. The placebo treatment discussed the consequences of wildlife poaching and ways to facilitate conservation. To ensure comparability, the videos were designed to be homogeneous in terms of duration, narration, language, and complexity of footage and information presented. Throughout this paper, we will refer to the above manipulations as the health, ﬁnance and control (placebo) treatments.
The videos were composed of six segments each lasting approximately two minutes. Following each of the six segments, we interrupted the video to obtain measures on cognitive functioning and individual preferences (e.g. risk, time, competition).
Experimental Design Details
The sample used was drawn in a two step-procedure using household lists generated in 2014. The selection and treatment allocation was done using Schneider and Schlather (2017) approach for multiple treatment assignment. The main idea of this procedure is to minimize
the diﬀerences in a covariance matrix resembling the observable characteristics of participants allocated to the diﬀerent treatments. The minimization is performed by replacing selected individuals from the comparison groups with new individuals from the sample until the process converges to a low diﬀerence in the statistic. We applied this procedure separately for married couples and for a comparable sample of never married men and women.
Individual. We consider single men and women and compared them with married participants.
Was the treatment clustered?
Sample size: planned number of clusters
Sample size: planned number of observations
Sample size (or number of clusters) by treatment arms
215 individuals in control, 205 health treatment and 220 in finance treatment.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
INSTITUTIONAL REVIEW BOARDS (IRBs)
Post Trial Information
Is the intervention completed?
Is data collection complete?