David S. Chester Ph.D.

The Harm Done


A New Way to Measure Self-Harm

Recent research uses voodoo dolls to study self-injury.

Posted Jun 28, 2019

Non-suicidal self-injury (NSSI) is the act of intentionally harming one's own body in a non-lethal manner. Approximately 15-17 percent of young adults and adolescents cut, burn, pick at, hit, or otherwise harm themselves, rendering this an important problem to solve. One of the first steps to reducing self-harm is to identify those who are most likely to engage in such behavior. But how do we measure people's tendency to injure themselves?

Conventional Measurement of NSSI

Typically, self-report questionnaires are used to measure NSSI, such as the Non-Suicidal Self-Injury Assessment Tool. Another measure, the Inventory of Statements About Self-Injury, asks you to indicate the "number of times in your life you have intentionally" performed the following behaviors on yourself: cutting, biting, burning, carving, pinching, pulling hair, severe scratching, banging/hitting, interfering with wound healing, skin rubbing, sticking self w/ needles, and swallowing dangerous substances. However, these self-report measurement approaches to NSSI have 2 main problems:

  1. People may not accurately report their past acts of NSSI. This could be either because they cannot accurately remember, or that they simply do not wish to disclose these past acts due to various concerns (e.g., not wanting to be institutionalized, wanting to be seen as likable by the person administering the measure).
  2. A greater amount of NSSI in the past does not necessarily say anything about how likely a person is to currently hurt themselves—though it is true that current motivations to engage in NSSI are correlated with a greater history of such behavior.

So if we want to bypass these two issues, what can we do? One approach is to seek objective medical records of NSSI, but this is problematic as well given privacy concerns and the fact that not all acts of NSSI lead to hospitalization. Plus, these medical records still do not necessarily tell us about how a person feels in their current state.

A New Approach: The Voodoo Doll Self-Injury Task

To alleviate these issues, my colleagues and I developed and published a new approach that does not rely on self-reports or medical records: the Voodoo Doll Self-Injury Task (VDSIT; researchers can freely access a Qualtrics version of the task here).  

The VDSIT is very simple, quick, and is administered over the internet. After following the online link, individuals are shown an image of a burlap doll that resembles a classic voodoo doll (e.g., with Xs for eyes), with and without 51 pins stuck into it (see below).

David Chester
Voodoo Doll with and without pins
Source: David Chester

Then, they are then told that this doll represents themselves and they will be able to stick anywhere from 0 to 51 sharp pins in the doll. By dragging a virtual slider bar that starts at 0 pins, they can then pick the number of pins to stick into the doll that represents them. That's it!

Our Findings

Across five different studies, we gave this task to over 1,200 undergraduates and Amazon Mechanical Turk workers and found empirical evidence that the VDSIT is a valid measure of NSSI. Supporting the accuracy of the VDSIT, the number of pins participants stuck into the doll corresponded to a greater likelihood of having injured oneself in the past and a corresponded to a greater frequency and severity of such past self-harm. The number of pins stuck in the doll was also linked to psychological variables known to predict self-harm:

We also asked participants how much they wanted to hurt themselves, right then, in the moment. These current motivations towards self-injury were linked to a greater number of pins stuck in the voodoo doll. This effect held for individuals with and without a past history of NSSI, which suggests that our task may be able to predict self-harm intentions/tendencies, even among people who have never harmed themselves in the past. The VDSIT may be a useful way for people to identify those at risk for NSSI, before they even commit the act.

Further, the number of pins stuck into the "self" voodoo doll did not correspond to participants' tendency to harm others, which suggest that the VDSIT captures something specific about hurting oneself and not just hurting people or things in general.

Finally, giving people negative feedback on an essay they wrote or excluding them from a group increased the number of pins participants stuck in the doll that represented them. These experiences are known precursors to self-injury and support the validity of the VDSIT and its ability to capture transient desires to hurt the self.

How Distressing is the VDSIT?

Our primary concern across these studies was to ensure that participants were not harmed by completing the task. As such, we tracked their mood before, during, and after they completed the VDSIT. Across all 3 mood assessments, participants did not, on average, report a significant level of negative emotions or distress, even if they had a past history of harming themselves. This suggests that the VDSIT is not a particularly traumatic or distressing task to complete, even among individuals with past histories of self-harm.

The Future of the VDSIT

The VDSIT should not yet be used as a clinical assessment of self-harm tendencies. This is because the task has yet to be administered to clinical populations of self-harming individuals and much more research on its validity, reliability, and utility remain to be done. Future research should examine this task over time and test whether it causes any adverse effects, predicts future acts of NSSI, and outperforms self-report measures across diverse populations. If the VDSIT meets these criteria, this rather strange-looking approach to NSSI measurement might greatly advance our understanding of the tendency to self-harm.

LinkedIn Image Credit: Fer Gregory/Shutterstock


Chester, D. S., Whitt, Z. T., Davis, T. S., & DeWall, C. N. (2017). The Voodoo Doll Self-Injury Task: A new measure of sub-clinical self-harm tendencies. Journal of Social and Clinical Psychology, 36(7), 554-578.