Does race salience affect support for policy attention to the opioid crisis?

We will randomize whether an introductory discussion of the opioid crisis includes a statistic about race. We will either not include any information by race, highlight that most overdose deaths are among White individuals, or emphasize the disproportionately high overdose deaths rates among Black and American Indian/Alaska Native individuals.

2024-05-23

2024-05-31

Share of respondents who report that the amount of public attention and spending on policies to decrease overdose deaths is “too much.”

We will calculate the share for each arm and compare each treatment arm to the control arm. Our full sample includes a “nationally-representative sample” and an “opioid-dependent sample.” We will use the full sample for the main analysis.

Respondents will be asked:
In your opinion, public attention to the opioid crisis and spending on policies to decrease overdose deaths, such as more opioid use treatment access and more naloxone access, is
 Too much (a)
 Just right (b)
 Not enough (c)

We will also study the following outcomes:
1) Share of respondents who report that the amount of public attention and spending on policies to decrease overdose deaths is “too much” or “just right.”
2) Share of respondents who select that “opioid use is a choice and people should just not use opioids.” (This answer is only an option for respondents who previously answered that the amount of public attending and spending dedicated to decrease overdose deaths is “too much” or “just right.” Those not given this question will be assigned to not answering that “opioid use is a choice and people should just not use opioids.”)
3) Maximum willingness-to-pay for naloxone policy that would reduce overdose deaths by 1 / 100,000.
4) Share who agree or strongly agree that opioid addiction is a choice. (We will also study each option separately: strongly agree, agree, neither agree or disagree, disagree, strongly disagree)
5) Share who answer that those found misusing opioids should serve jail time.
6) Share who believe that prisons should provide MOUD to prisoners.
7) Share who believe that prisons should provide naloxone to prisoners upon release.


Our sample includes a “nationally-representative sample” and an “opioid-dependent sample.” We will conduct analyses of our primary and secondary outcomes for each sample separately as secondary analyses. When making this stratification, we will rely on our own questions about opioid dependence.

In addition, we will make the same comparisons for stratified samples: (1) people who know someone impacted by opioids in any of the following ways: (a) struggled with an opioid addiction, (b) needed medical attention because of an opioid addiction, (c) sought treatment for an opioid addiction, (d) Faced police contact, arrest, or incarceration due to an opioid addiction, or (e) are likely or very likely to overdose from opioid use; (2) people who are personally impacted by opioid use (same reasons as above) or report being likely or very likely to overdose from opioid use. In each stratification, we will show the comparable results for the sample not meeting the conditions.

(3) We will also stratify by race, gender, and political affiliation.

(4) Finally, we will stratify based on the 2022 overdose death rate of the state.

We will randomize respondents into one of the three arms with equal probability.

Not available

The code for the survey will perform the randomization in real-time.

Individual

No

2000 individuals

2000 individuals

We expect about 667 in each arm.

Because we have two treatment arms, we (conservatively) use alpha=0.025 in the power analysis. Our full sample consist of 2000 respondents. We expect each comparison to involve approximately 1333 observations. If the baseline share responding that there is “too much” attention and spending on the opioid crisis is 0.3, then with alpha=0.025, we will be able to detect a difference of 0.0800 with power=0.8.