by Michael D. Anestis, M.S.
Before beginning today's article, I wanted to quickly comment on how much Joye and I have enjoyed the increased participation in the comment sections of our articles (follow this link and this link for strong examples). We greatly enjoy conversing with readers, whether through debates, agreements, or answers to questions. Ideally, participation will evolve over time such that conversations take place primarily between readers, each of whom brings a new perspective and additional data to consider. In the meantime, we welcome your thoughts on this and past articles and look forward to hearing what you have to say.
Recently, I came across an interesting study by Dr. Matthew Nock of Harvard University and one of his graduate students, Irene Belle Janis (Janis & Nock, 2008) that seems quite relevant to several of the articles and discussions we have had recently on Psychotherapy Brown Bag. In this study, the authors sought to examine a pivotal question for clinicians treating self-injurious clients: what is the best way to assess the likelihood of my client engaging in future self-injurious thoughts and/or behaviors? Specifically, the authors sought to determine whether the client's past history of self-injurious thoughts and behaviors or the client's prediction of whether or not he or she will engage in such thoughts and behaviors in the coming months provided a more accurate assessment of actual outcomes.
The rationale for this study was to test the actual utility of a multi-pronged assessment approach that many clinicians use involving both past behaviors and client predictions in determining risk. As an example, the authors wanted to provide an empirical framework within which a clinician can understand risk in a client who has a long history of self-injurious thoughts and behaviors but who indicates in session that he or she will not engage in such thoughts and behaviors in the future. In this situation, is it best to assume the client most accurately understands his or her likely future behavior or to rely upon past behavior as a better harbinger for what is to come next?
Janis and Nock (2008) detailed several prior studies, most of which had been conducted by social psychologists, that have revealed a tendency for individuals to be highly inaccurate in predicting future emotions and behaviors (Diekmann et al., 2003; Epley & Dunning, 2000; Wilson & Gilbert, 2003). For instance, Brickman, Coates, and Janoff-Bulman (1978) asked a sample of individuals how happy their lives would be if they won the lottery and how unhappy their lives would be if they became paraplegic. Most participants indicated they would be extremely unhappy relative to the general population if they were paraplegic and extremely happy relative to the general population if they won the lottery; however, samples of paraplegic individuals have been shown to be no less happy than the general population and samples of lottery winners have been shown to be no more happy than the general population. In other words, our vision of how we would feel in a particular context may not match up with reality. In additional studies run by Gilbert, Pinel, Wilson, Blumberg, and Wheatley (1998), individuals overestimated the duration of their emotional distress in response to a romantic breakup, the failure to receive academic tenure, the defeat of their favored political candidate, rejection by a potential employer, and several other negative life events. With respect to behaviors, people have been shown to be equally inaccurate in their predictions (e.g., Deikmann et al., 2003).
In this particular study, Janis and Nock (2008) surveyed a sample of 64 individuals (51 females) using a structured clinical interview, the Self-Injurious Thoughts and Behaviors Interview (SITBI; Nock, Holmberg, Photos, & Michel, 2007), regarding the degree to which they had engaged in a variety of self-injurious (both suicidal and non-suicidal) thoughts and behaviors over the past 12 months. Additionally, the authors collected data on the participants' predictions regarding whether or not they would engage in self-injurious thoughts and/or behaviors in the coming months. Six months later, follow-up data was obtained for 50 of those individuals, with attrition due to inability to locate participants, refusal to complete the study, and other factors.
The authors reported several interesting findings. First, history of self-injurious thoughts over the past 12 months was significantly correlated with participant prediction of future self-injurious thoughts (r = .71, p < .001). Similarly, history of self-injurious behaviors over the past 12 months was significantly correlated with participant prediction of future self-injurious behaviors (r = .49, p <.001). As such, the authors noted, participants appeared to be basing their predictions, at least in part, on their own history.
The most important analyses in this study, however, examined the relative contributions of past behavior and client predictions in the actual prediction of future thoughts and behaviors. The authors found that, whereas past history of self-injurious thoughts predicted future self-injurious thoughts above and beyond the contributions of client predictions, client predictions did not predict self-injurious thoughts above and beyond the contributions of past history of self-injurious thoughts. Similarly, past history of self-injurious behaviors predicted future self-injurious behaviors above and beyond the contributions of client predictions, but client predictions did not predict future self-injurious behaviors above and beyond the contributions of past self-injurious behaviors. In other words, considering the participant's predictions about future thoughts and behavior did not increase the accuracy of authors' ability to predict future thoughts and behavior and, as such, the utility in considering this particular client variable for this particular set of outcomes appears questionable.
Before I discuss why these findings are important, I want to note a few quick limitations. First, as the authors readily noted, the sample size in this study is very small, thus limiting the degree to which we can generalize from these findings. The authors explained clearly at the end of the manuscript that these findings should be considered preliminary and that independent replication with a larger sample would substantially increase confidence in the validity of the results. Additionally, with only two data points, all we have is a pre and post snap shot rather than a clear examination of fluctuations over time in the frequency and severity of thoughts and behaviors and the relative accuracy of past histories and client predictions. Also, due to the low statistical power of the small sample size, the authors were not able to examine suicidal and non-suicidal thoughts and behaviors separately or to parse thoughts and behaviors into specific types and levels of severity. They noted that future researchers with larger samples would provide important insight by examining such questions with a sample size large enough to detect significant effects.
Now, why are these findings, preliminary though they may be, important for us to consider? Quite simply, they address our own tendencies to be inaccurate, both with respect to predicting our own thoughts and behaviors and in assessing the degree to which a client has insight into what he or she may do in some future context. In our relationships with clients, we can develop biases and beliefs about their readiness to change that could, in certain cases, cause us to underestimate their risk for dangerous, potentially fatal outcomes. This is not to say that, once an individual has a past history of self-injurious thoughts and/or behaviors, that individual should forever be considered at imminent risk; however, it does indicate that a recent history is a cause for concern and should be weighed very heavily in the assessment of risk. As time elapses, particularly for individuals with a minimal history of limited severity, that risk assessment could evolve, but here again, data indicating that the client's prediction is more accurate than his or her past history would be necessary in order to substantiate that claim.
Ultimately, what this study demonstrates is that, although some clients who predict that they will no longer engage in self-injurious thoughts and/or behaviors may be correct, their prediction itself is not a reliable predictor of accuracy. As such, using a client's prediction in the assessment of risk may actually make the clinician's risk assessment less reliable and accurate and place the client at greater risk.
If you would like to learn more about suicide and non-suicidal self-injury, we recommend the following products, all of which are available through our online store:
- Why People Die by Suicide
- Treating Suicidal Behavior: An Effective, Time-Limited Approach (Treatment Manuals For Practitioners)
- The Interpersonal Theory of Suicide: Guidance for Working With Suicidal Clients
- Understanding Nonsuicidal Self-Injury: Origins, Assessment, and Treatment
- Freedom from Selfharm: Overcoming Self-Injury with Skills from DBT and Other Treatments
Mike Anestis is a doctoral candidate in the clinical psychology department at Florida State University
Subscribe in a reader
