Academic Commons

Theses Doctoral

How Unreliable are adult-reported suicide attempts? An examination of correlates and underlying causal mechanisms of discordant reporting over time

Palmetto, Niki P.

The challenge of capturing suicide attempts in the population, plague its examination. The reliability of adult-reported lifetime suicide attempts had not been rigorously explored prior to this work, and therefore estimates have remained largely unchallenged. This dissertation explicitly sought to fill this research gap by utilizing a longitudinal study, comprising two waves of data collection, in which information on suicide attempts was obtained at both time points. Chapter 1 presents a systematic review of the literature depicting the state of the literature with regards to the reliability of suicidality measures (e.g. ideation, plans, and attempts). Few studies assessed correlates of discordant reporting, and no studies examined the reliability of adult-reported suicide attempts. Drawing upon the correlates reported within studies, as well sources of heterogeneity across studies, I posited four plausible causal mechanisms underlying discordant suicidality reporting; recall failure, reinterpretation, conscious denial, and lack of construct comprehension. Extending these findings, I proposed that the likelihood of each mechanism is influenced by factors such as the severity of the suicidality, amount of time passed since the suicidal event, social desirability effects, mood context, and suicide construct validity. In Chapter 2, I assessed the reliability of adult-reported lifetime attempts as reported in a large, population-based longitudinal study, and found reports to be moderately reliable, with a Kappa coefficient of 0.51. I hypothesized that discordant reporters would be more similar to individuals who reported a past attempt at both waves (Concordant yes responders), compared with individuals who reported no attempt at both waves (Concordant no responders). I found that indeed, discordant reporters were more similar to the former, potentially signifying that discordant reporters are true attempters who underreported their attempt at one time point. Further, I hypothesized that discordant reporters would be less likely to have a history of depressive disorders compared with Concordant yes responders; positing that this history would serve as a marker for attempt severity, and that discordants would have less severe attempts, which would therefore be more easily forgotten or reinterpreted. Contrary to this hypothesis however, discordants were as likely as Concordant yes individuals to have a history of depressive disorders, and unexpectedly, discordants were much less likely to have a history of suicidal ideation. It is therefore plausible that a history of suicidal ideation serves as a marker for attempt severity, and/or that discordant reporters are characterized by more impulsive attempts. In Chapter 3, I examined how a respondent's current depressed mood may influence the recall, and hence reporting of attempts. Based on established mood-recall theories, I tested three competing hypotheses to determine if a current depressed mood would enhance (mood-congruent recall), inhibit (mood memory deficit effect), or have no effect on the recall (mood-independent recall) and reporting of attempts. I hypothesized that discordant reporters would demonstrate a mood-congruent pattern of reporting, such that a depressed mood at the time of the interview would increase the likelihood that a respondent would report an attempt at that wave. There were in fact, distinct mood-congruent reporting effects among Recanters, yet mood-independent effects detected among New endorsers. This may indicate that New endorsers are a unique group of discordant responders, which warrant further examination. Still, because respondents in our sample were over 20 times more likely to recant than newly endorse, and comparatively, there was limited power within our New endorser group, I believe these results may be generalized to assert that overall, discordant responders report in a mood-congruent fashion.


  • thumnail for Palmetto_columbia_0054D_10403.pdf Palmetto_columbia_0054D_10403.pdf application/pdf 1.45 MB Download File

More About This Work

Academic Units
Thesis Advisors
Link, Bruce George
Ph.D., Columbia University
Published Here
November 9, 2011