Treatment design for an experimental test of the satisfaction mirror hypothesis in the public sector

Research output: Conference Papers (RGC: 31A, 31B, 32, 33)33_Other conference paperpeer-review

View graph of relations

Author(s)

Detail(s)

Original languageEnglish
Publication statusPresented - 5 Apr 2023

Conference

TitleIRSPM Conference 2023
PlaceHungary
CityBudapest
Period3 - 5 April 2023

Abstract

This paper reports the design of a randomized controlled trial to test for the effect of job satisfaction on citizen satisfaction during face-to-face encounters between public service employees and citizens, addressing the question: Is high job satisfaction mirrored by high citizen satisfaction? The experiment builds upon the first individual-level observational study of the satisfaction mirror in the public sector (Petrovsky et al. 2022) and aims to move the research agenda on this topic forward by adding causal identification. The satisfaction mirror is theoretically interesting because it bridges the literatures on job satisfaction and citizen satisfaction.

Based on the theoretical argument, which combines the satisfaction mirror and the core idea of the expectancy-disconfirmation model of citizen satisfaction (Van Ryzin 2004), the experiment tests hypotheses about the effects of expressed job satisfaction and service quality on citizen satisfaction. Further, the study also correlationally tests whether expectations (which are elicited at the beginning but not manipulated) moderate the treatment effects.

The study design is centered on a 3*2 between-subjects experiment (i.e. there are six arms). The findings from Petrovsky et al.’s (2022) existing observational study inform a rough estimate of the number of subjects required in each of the six conditions to have statistical power of 0.8 (two-tailed test, alpha = 0.05). The total effect of a one-standard deviation higher job satisfaction would be a 0.6 standard deviation higher citizen satisfaction. Based on this, the estimated minimum number of subjects in each of the six conditions is 45. To guard against too high an effect size estimate, the required number of subjects is increased to 100 subjects per experimental condition. This means there will be 600 experimental subjects.

In the experiment, subjects will be exposed to six different versions of a video showing an interaction between a public service employee and a citizen. The video involves trained research assistants playing out a script. The setting resembles a lost property office of a local public transit agency. The citizen in question inquires about an iPad they lost. The video incorporates two experimental treatments, leading to six different versions of the film that will be shown to randomly assigned sets of experimental subjects. The first treatment condition has the employee act in one of three ways: (i) high job satisfaction, (ii) low job satisfaction, and (iii) neutral acting as the control condition. The second treatment involves objectively poor performance (the official makes a recognizable mistake), with correct performance (no mistake) being the control condition. After watching the encounter, the outcome – subjects’ satisfaction – will be measured. Using feedback on the treatment design and experience from trial runs, the video with the finalized treatments will be produced in the late summer.

Bibliographic Note

Research Unit(s) information for this publication is provided by the author(s) concerned.

Citation Format(s)

Treatment design for an experimental test of the satisfaction mirror hypothesis in the public sector. / Petrovsky, Nick; Xin, Ge; Yu, Jinhai.
2023. IRSPM Conference 2023, Budapest, Hungary.

Research output: Conference Papers (RGC: 31A, 31B, 32, 33)33_Other conference paperpeer-review