Behavioural Change in the Unemployed

Professor Felicia Huppert introduced the session, noting that Coppin is unusual among PhD students in being already established in the applied world, working with politicians and large unemployed populations rather than purely within academic settings. She framed his confirmation as exciting precisely because it would bring rigorous methodology to substantial real-world experience.

Coppin set out the structure of the 40-minute presentation: introduction, key research questions, literature review and theoretical framework, research design including hypotheses and method, ethics considerations, analysis, and a closing reflection.

Coppin opened by establishing that work is an essential component of a thriving life, providing engagement, relationships, meaning, purpose and accomplishment beyond financial benefits. The literature consistently demonstrates that well-being declines on entering unemployment and rises on returning to work. The overall hypothesis of the PhD: well-being interventions for unemployed jobseekers increase the likelihood of finding work and achieving sustainable employment outcomes.

A crucial reality framed the entire research design: government funding limitations mean that well-being interventions cannot be provided to every jobseeker. The thesis therefore had to develop and validate an approach for identifying which jobseekers would benefit most from stage-matched interventions, drawing on the Transtheoretical Model of Change. Outcomes would be measured primarily in hard employment data (whether participants secured and retained jobs), with secondary measures including changes in self-efficacy, well-being, and grit.

Study 1, the validation of a measure of jobseeking commitment to change, asks: can we reliably measure discrete and consistent stages of change in the jobseeking context, and does such a measure predict behavioural outcomes (specifically, securing employment)? An informal poll of 42 Australian employment advisors found that 72% of standard employment service interventions are action-based (job search skills, interview skills, CV writing), suggesting a potentially critical mismatch between intervention type and jobseeker readiness.

Study 2 focuses on developing and testing well-being interventions contextualised for unemployed young people in Brisbane, comparing employment outcomes with a control group not receiving the workshops.

Study 3 asks whether stage-matched interventions, deployed across general unemployed populations Australia-wide, deliver superior employment outcomes compared with the standard government control. Approximately 8,000 participants in the experimental group and 12,000 in the control group across the same employment service offices.

Unemployment is universally bad for the economy, communities, and individuals. The Australian Government spent approximately A$18.4 billion in 2013-14 on direct unemployment support. Government services typically measure extrinsic barriers to employment (literacy, qualifications, driver's licence, work experience) but do not measure the intrinsic barriers identified by Coppin's work: motivation, attitude, confidence, well-being.

A senior member of the Australian Department of Employment told Coppin: "without addressing those intrinsic barriers, the extrinsic ones will fail. You won't get a driving licence, you won't try to improve your qualifications, unless you have the motivation, attitude and well-being to try to do so." The UK Department of Work and Pensions has called the ability for an advisor to predict, before allocating resources, which jobseekers will return to work without help and which will require intensive support, the "holy grail of government employment services."

Coppin cited Kroft, Lange and Notowidigdo's 2013 audit study, which sent fake resumes to real job postings in 100 US cities, varying only the candidate's name and length of unemployment. Short-term unemployed (less than six months) with industry experience: 17% interview rate. Short-term unemployed without industry experience: 9% interview rate. Above six months without experience: callback rate falls dramatically, with only ~3% receiving an interview. Approximately half a percent of long-term unemployed without experience reach interview stage. The data validated the long-held but academically unproven claim that six months of unemployment is a critical tipping point beyond which the probability of return to work drops sharply.

Despite enormous government policy investment, no academic literature directly addresses whether jobseekers are genuinely jobseeking. The closest existing research applies the Transtheoretical Model of Change to addictive behaviours (smoking, drinking) and to one specific category of unemployed: those off work due to workplace injury (the URICA-VC study, Levesque et al, and the LASER study, Lam et al). Neither directly addresses general unemployment in the Australian context.

Coppin walked through the five stages of change developed by Prochaska and DiClemente: precontemplation (no intention to change), contemplation (considering change in the next six months), preparation (intending to change soon, lacking some resources), action (actively making the change), and maintenance (sustaining the change over six months). The most important implication of their research: efficient behavioural change depends on doing the right thing at the right time. Stage-mismatched interventions are not just ineffective; they can be actively counterproductive, alienating recipients and reducing the likelihood of future engagement.

Coppin acknowledged the academic literature on stage-of-change is genuinely mixed. For every study supporting its efficacy, another raises doubts. Whitelaw and colleagues observed a relative paucity of strong supportive evidence and questioned whether the theory was simplifying complex behavioural reality usefully or merely providing clinicians with a comforting categorisation framework. Coppin quoted Einstein: "Everything should be made as simple as possible, but not simpler."

URICA-VC (Levesque et al., 2002), the University of Rhode Island Change Assessment for Vocational Counsellors, refined a 96-item original survey down to 12 items. Studied 150 participants who had been off work due to workplace injury for 13 months, tracked over 30 months. Found that action-stage participants were four times more likely to return to work than precontemplators.

LASER (Lam et al., 2010), the Lam Assessment of Stages of Readiness, a 14-item paper-based assessment with 150 participants, 98.2% female and 83% African American, drawn from US welfare. Tracked at six months, found similar results to URICA-VC.

The limitations of both studies for Coppin's purposes: small samples, narrow demographic populations, US welfare context (substantially different from Australian and UK systems), and (in the LASER study) failure even to reference the prior URICA-VC work. The Australian thesis would need substantially larger and more demographically diverse samples to make the methodology defensible to government partners managing 700,000 Australian welfare recipients (and 3 million UK equivalents).

Coppin drew on Marie Jahoda's seminal 1982 work Employment and Unemployment: A Social-Psychological Analysis, which identified five latent functions of employment beyond income: time structure, regular activity, social engagement, meaning and purpose through shared collective effort, and accomplishment. Coppin highlighted that Jahoda's framework, developed in 1982, almost exactly anticipates Seligman's PERMA model of well-being (positive emotion, engagement, relationships, meaning, accomplishment), suggesting well-being interventions should be a particularly rich source of effective intervention design for the unemployed.

Coppin reviewed the four key studies establishing the mental health consequences of unemployment: Lucas, Clark, Diener and Georgellis on the set point of life satisfaction (the 5.1-year recovery finding referenced in Coppin's earlier 2015 talk); Paul and Moser's meta-analysis confirming significant negative effects on mental health, well-being, and depression; studies showing increased stress, anxiety, depression, and decreased work satisfaction following unemployment; Jefferies and Darby establishing unemployment as a direct cause of depression.

Coppin recounted asking Martin Seligman directly: what should we do with the unemployed to improve their chances of returning to work? Seligman pointed him to The Resilience Factor by Karen Reivich and Andrew Shatté, the book that underpins both the Penn Resiliency Program and Comprehensive Soldier Fitness. The book's seven pillars of resilience (emotional regulation, impulse control, causal analysis, self-efficacy, realistic optimism, empathy, reaching out) provided the structural foundation for Coppin's intervention design.

Coppin layered Bandura's three self-efficacy-building mechanisms (emotional arousal, performance experiences, verbal persuasion) into the workshop design.

The single most important methodological insight in the talk. Coppin cited Grove and colleagues' meta-analysis of 136 studies comparing mechanical (structured questionnaire) prediction with clinical (expert judgment) prediction across general medicine, mental health, business, education, and training settings. The finding: mechanical prediction systematically outperforms clinical judgment, with no exceptions in the entire 136-study set. Most counterintuitively, clinicians' accuracy decreases the more time they spend face-to-face with the subject. Implication for Coppin's thesis: a 12-item structured questionnaire administered online would predict employment outcomes more accurately than a 20-year experienced employment advisor talking to the same person, validating the use of mechanical assessment at scale.

Coppin's UK pilot of the paper-based 12-item assessment with 105 jobseekers in Mooloolaba and 84 in Essex revealed several practical and ethical issues: scoring took 26 minutes per participant (clinically untenable in 30-minute employment service appointments); four participants required additional language support not anticipated in the protocol; the ethics requirement that participants read the consent form was not consistently observed; a bias emerged in Job Centres where when scored by benefits-administering staff, 92% of participants self-reported as action stage compared with realistic distributions in independent settings; and a 32% misinterpretation rate when scored manually (the chart-matching method introduced too much subjectivity).

The University of Rhode Island collaboration converted the assessment to an online format with one-second scoring and zero misinterpretation, making the methodology scalable.

The second study tested a five-half-day workshop intervention, contextualised for 75 unemployed young people in Brisbane, against the most successful prior intervention in Australian youth unemployment history (the Brotherhood of Saint Laurence's programme). Five workshop themes (Strengths, Positivity, Mental Toughness, Mindset, Character & Goals) covering 31 contextualised positive psychology interventions. Workshop branding was changed from "well-being workshops" (which alienated some young men) to "resilience workshops", which produced 100% attendance.

The intervention drew on direct collaboration with several major figures in the field: Barbara Frederickson on positivity, Carol Dweck on mindset, Andrew Shatté on resilience (and notably, Shatté grew up in Brisbane and his first job was with Cromwell Employment Services), Brian Niemiec at the VIA Institute on character strengths.

The third study would deploy stage-matched interventions across approximately 8,000 jobseekers nationally, with 12,000 in the same employment offices receiving the standard government services as a control. After initial assessment, advisors would print a stage-specific intervention guide showing what to do for each jobseeker. Two hundred employment advisors would be trained to deliver the workshops. Job outcomes would be tracked at six months. The design would compare stage-matched intervention outcomes with controls, with secondary regression analyses on changes in self-efficacy, well-being, grit, and other psychological capital constructs.

The Q&A section produced several pointed methodological challenges. On three factors but five stages: a questioner asked why the structure measured three factors (action, contemplation, precontemplation) but identified five discrete stages of change. Coppin explained that one stage (maintenance) was inappropriate for the unemployment context (it presupposes already having and keeping a job), and that "preparation" emerged from intermediate scoring patterns rather than being a discrete factor.

On positive versus negative behaviour change: a questioner observed that most stage-of-change literature concerns getting people to stop doing something (smoking, drinking) rather than start doing something (jobseeking). Coppin acknowledged this as a significant theoretical issue the thesis would need to address.

On the active control problem: the most pointed challenge, to validate stage-specific intervention effects, the methodology needs to randomly assign people to interventions across different stages. Coppin acknowledged the design did not include this and that government partners would not permit randomisation across stages on ethical grounds (it would mean deliberately giving inappropriate interventions to some participants).

On the comparison with theory of planned behaviour: a questioner raised whether Ajzen's Theory of Planned Behaviour offered a more parsimonious framework. Coppin acknowledged the conceptual overlap, particularly around intention strength.

On integrating the Australian Job Seeker Classification Instrument: a questioner asked how Coppin's intrinsic measures would be combined with the existing extrinsic Australian government Job Seeker Classification. Coppin's response: the data systems track this and the extrinsic-intrinsic combination produced a striking finding, that 36% of jobseekers classified as "Stream 1" (closest to employment) were actually in precontemplation or contemplation, while 34% of "Stream 4" jobseekers (furthest from employment) were genuinely in action stage. The conventional government model was systematically misallocating intervention resources.

On training quality control across 200 advisors: Coppin acknowledged this as a real methodological constraint, mitigated through detailed scripts, recorded videos with end-of-training tests, observation guides for managers, and reliance on employment service providers to track delivery.

On reassessment over time: a questioner noted that participants' stage-of-change might shift over six months. Coppin acknowledged this as a study limitation imposed by the agreement with government partners, who would only fund a single assessment.