Water Scarcity and Specificity of Saving Tips

This interventions aims to increase water conservation efforts in Namibia. Three text mesage treatments via mobile phone ask customers to use less water. The treatments however vary the degree of their specificity. The treated customers will receive one treatment in English and one in Oshiwambo (local language).

1. Untreated comparison group

2. Specific tips treatment group:
Customers receive three specific savings tips as for example “Shorten the time you spend in the shower by a minute or two”.

3. Semi-specific ("Betty Crocker’s Eggs") treatment group:
Customers receive three semi-specific saving tips as for example “How can you re-use water in your household (for example from the kitchen)?”.

4. Unspecific treatment group:
Customers are asked to come up with three own ideas on how to reduce water consumption.

2017-08-05

2017-12-31

Water consumption behavior and payment behavior.

I use administartive data on water consumption and payment behavior from the Namibian Water Corporation. The main outcome variables (water consumption and payments) show very high variance, which reduces power. Data transformations (logs and winsorizing) to deal with outliers are necessary to reduce variance as well as looking at subsamples with lower variance.

Customers will be randomly assigned to one of three treatment conditions or the baseline. They will receive a short text messages in both English and Oshiwambo (Local language) to encourage water conservation efforts as described in the intervention. Behavioral predictions are contradicting.

Hypothesis 1: Effectiveness increases with specificity of tips
Complexity and limited cognitive resources reduce the ability and motivation to act on pro-environmentally behavior (Steg, 2016). The specification of saving tips reduces the cognitive effort to find ways to act on pro-environmental behavior. Further, people are more likely to engage in a certain behavior if the task is perceived as easy (Ajzen, 1991). Literature on implementation intentions shows that specifying a concrete action plan increases the likelihood of acting on that behavior (Gollwitzer & Sheeran, 2006). Once a plan is established, people execute their plans without having to invest cognitive effort, making their behavior swift and efficient (Gollwitzer, 1999). From a practical perspective, reducing complexity and simplifying difficult choices has been shown to be among the most effective “nudges” (Thaler & Sunstein, 2008). Hence, this body of research suggests that a specific saving should be most effective.

Hypothesis 2: Effectiveness decreases with specificity of tips
Previous research has shown that people have a tendency to overvalue their own ideas over the ideas of others. Hooshangi and Loewenstein (2016) for example show that subjects in the laboratory are investing more in their own entrepreneurial ideas than in others holding the quality of the entrepreneurial ideas constant. Norton, Mocho and Ariely (2012) show that subjects increase their valuation of self-made products, the so called “Ikea-effect”. While unspecific saving tips as for example “Please find a way to safe water” could benefit from the active involvement in the idea-generating process yet they require recipients to invest own cognitive efforts.
Additionally, less specific tips avoid that recipients receive tips that already in use (for example if a car is never washed with water anyways) or if the savings tip not applicable (for example if the recipient does not own a car). Lastly, involving participants to express their opinions is an important factor to improve the acceptability of pro-environmental strategies (Dietz & Stern, 2008; Steg, 2016).

Randomization done in office by a computer to balance pre-treatment payment and consumption behavior. Treatments are stratified geographically by pipeline location.

Randomization is done at the individual level (customer)

No

Total amount of customers that are assigned to receive a text message: 15231.
Due to outdated phone numbers and technological restrictions (e.g., network erros), not all customers will receive a text message.

Total amount of customers that are assigned to receive a text message: 15231. Due to outdated phone numbers and technological restrictions (e.g., network erros), not all customers will receive a text message.

1. Untreated comparison group: 3788
2. Specific tips treatment: 3811
3. Semi-specific tips treatment: 3803
4. Unspecific tips treatment: 3829