Psychological Safety in virtual student groups: A validation study

Abstract

Virtuelt samarbeid har blitt mer vanlig de siste årene, bade på arbeidsplasser og i høyere utdanning. Formålet med denne oppgaven var å undersøke konstrukt validiteten til Edmundsons Team Psychological Saftey skala i et utvalg med virtuelle studentgrupper. Utvalget besto av 344 studenter som tok det prosjekt-baserte master emnet Eksperter i Team ved NTNU. Datamaterialet ble udnersøkt med en single-level CFA og en multilevel CFA. Reliabiliteten ble undersøkt ved hjelp av Intraclass correlation coefficients (ICCs), Cronbach’s alpha and McDonald’s omega. Gjennomsnittlig ICC(1) verdi for spørsmålene som ble brukt for å måle psykologisk trygghet var .107. ICC(1) verdien for individenes gjennomsnittlige score på psykologisk trygghet skalaen var .182. ICC(2) verdien ble estimert til .397. De estimerte goodness-of-fit verdiene for CFA (single-level) modellen var χ2 (14) = 29.824, p = .008, CFI = .938, TLI = .907, RMSEA = .057 og SRMR = .039. Den gjennomsnittlige standardiserte faktor ladningen var .544. Cronbach’s α = .726 og McDonalds’s ω = .724. De estimerte goodness-of-fit verdiene for CFA (multilevel) modellen var χ2 (28) = 53.322, p = .002, CFI = .930, TLI =.895, RMSEA = .051, SRMRwithin = .044, og SRMRbetween = .147. Den gjennomsnittlige standardiserte faktor ladningen var.516 for within-nivået, og .768 for between-nivået. McDonald’s ω var .695 for within-nivået, og .936 for between-nivået. Reliabilitets og goodness-of-fit estimatene fra de bekreftende faktoranalysene viste alt i alt at en-faktor strukturen i Edmundsons Team Psychological Safety skala hadde tilstrekkelig nivå av model fit i utvalget med virtuelle studentgrupper. Analysene konkluderte dermed med at konstruktvaliditetsbevisene for Team Psychological Safety i virtuelle studentgrupper var tilfredsstillende.

Virtual teamwork has become more common in recent years, both for the workplace and in higher education. The aim of this study was to investigate the construct validity of Edmundson’s Team Psychological Safety scale in a context of virtual student groups. The sample used in this study consisted of 344 students taking part in a project-based master’s course at a Norwegian university. The data was analyzed with a single-level CFA and a multilevel CFA. Reliability was investigated with Intraclass correlation coefficients, Cronbach’s alpha and McDonald’s omega. The average ICC(1) value for the items of the psychological safety construct was .107 (. The ICC(1) value for the mean psychological safety score of the participants was .182. The ICC(2) value was estimated to .397. Estimates for the single-level analysis were χ2 (14) = 29.824, at p = .008, CFI = .938, TLI = .907, RMSEA = .057 and SRMR = .039. The average standardized factor loading was .544. Cronbach’s α = .726 and McDonalds’s ω = .724. Estimates for the multilevel CFA were was χ2 (28) = 53.322, at p = .002, CFI = .930, TLI =.895, RMSEA = .051, SRMRwithin = .044, and SRMRbetween = .147. The average standardized factor loading was .516 for the within-level, and .768 for the between level. McDonald’s ω was .695 for the within-group level, and .936 for the between-group level. Overall, reliability estimates and goodness-of-fit estimates from the confirmatory factor analyses demonstrated an adequately good level of fit for a one factor structure of psychological safety in the context of virtual student groups. The Team Psychological Safety scale was shown to have acceptable levels of construct validity for the virtual student groups included in this analysis.