Field
Last Published
|
Before
March 31, 2020 03:11 PM
|
After
March 31, 2020 07:27 PM
|
Field
Intervention (Public)
|
Before
In partnership with an online naloxone distributor, we will test the importance of price, information, and stigma in deterring access to naloxone. More specifically, we will implement randomized experiments (via Google Ads) for multiple concurrent interventions related to price, and information. We will monitor ad exposure (or impressions), clicks or traffic driven to the exchange site by different types of advertisements, and subsequent purchases. Google Analytics will also be used to track clicks and purchases made from counties that are randomized to receive no ads.
|
After
In partnership with an online naloxone distributor, we will test the importance of price and information (including stigma) in deterring access to naloxone. More specifically, we will implement randomized experiments (via Google Ads) for multiple concurrent interventions related to price, and information. We will monitor ad exposure (or impressions), clicks or traffic driven to the exchange site by different types of advertisements, and subsequent purchases. Google Analytics will also be used to track clicks and purchases made from counties that are randomized to receive no ads.
|
Field
Primary Outcomes (End Points)
|
Before
ad clicks; site visits; naloxone purchases
|
After
naloxone doses purchases
|
Field
Experimental Design (Public)
|
Before
We will run concurrent experiments through the Google Adwords platform, using their A/B split testing technology, to randomize ads that address information gaps and/or stigma issues that may be barriers to the take-up of naloxone. We exclude counties in which the online naloxone distributor is not selling naloxone (as of March 31, 2020).
Randomization will occur at the county level. We will also randomly allocate counties to a control arm that receives no advertising from us. Across our treatment conditions, we will further randomize the advertised (and thereby offered) price of naloxone. In the end, we have 5 approximately equally-sized groups of counties.
Using 2018 overdose data, we constructed the opioid overdose rate for each county as well as indicators for each quintile of the opioid overdose distribution. To improve balance on these metrics, we re-randomized until all 5 experimental arms were within 0.5 opioid overdoses per 100,000 of each other and the distributions were similar as well (for each quintile, no arm is more than 10 percentage points larger than any other arm).
|
After
We will run concurrent experiments through the Google Adwords platform to randomize ads that address information gaps and/or stigma issues that may be barriers to the take-up of naloxone. We exclude counties in which the online naloxone distributor is not selling naloxone (as of March 31, 2020).
Randomization will occur at the county level. We will also randomly allocate counties to a control arm that receives no advertising from us. Across our treatment conditions, we will further randomize the advertised (and thereby offered) price of naloxone. In the end, we have 5 approximately equally-sized groups of counties.
Using 2018 overdose data, we constructed the opioid overdose rate for each county as well as indicators for each quintile of the opioid overdose distribution. To improve balance on these metrics, we re-randomized until all 5 experimental arms were within 0.5 opioid overdoses per 100,000 of each other and the distributions were similar as well (for each quintile, no arm is more than 10 percentage points larger than any other arm).
|
Field
Secondary Outcomes (End Points)
|
Before
|
After
ad impressions; ad clicks; website visits; revenue
|