Primjena mrežne analize u psihologijskim istraživanjima / Anamarija Lonza.
Sažetak

Mrežni pristup predstavlja novu paradigmu u proučavanju odnosa između psiholoških konstrukata i manifestnih varijabli. Prema tome pristupu varijable tvore autonomni dinamički sustav, a konstrukt se ne promatra kao njihov uzrok, već kao rezultat njihove kompleksne međusobne interakcije. Analitički se taj pristup zasniva na metodi mrežne analize koja modelira varijable kao čvorove povezane skupom bridova. Ovaj rad predstavlja svojevrsni pregled skupa postupaka u mrežnoj analizi, odnosno nudi njihovo pojašnjenje te njihovu praktičnu primjenu na dvama različitim setovima podataka. Prvi set podataka predstavlja rezultate na upitniku DASS-21 (N = 1016) i u okviru njega demonstrirani su procjena parametara mreže, izračun mjera centralnosti čvorova, identifikacija klastera u mreži te provjera stabilnosti parametara. Rezultati pokazuju da najveću centralnost imaju čestica depresivnosti Osjetio/la sam kao da se nemam čemu radovati, čestica anksioznosti Osjetio/la sam da sam blizu panici i čestica stresa Osjećao/la sam se jako nervozno. Čvorovi mreže očekivano su se grupirali u tri klastera koji sadržajno reprezentiraju Depresivnost, Anksioznost i Stres. Analize stabilnosti pokazale su ograničenu stabilnost bridova, dok je stabilnost centralnosti čvorova ovisila o korištenoj mjeri. U drugome istraživanju, koje sadrži podatke o stavovima adolescenata prema izgledu vlastitoga tijela, prikazan je Test usporedbe mreža usporedbom mreže adolescenata (n = 524) i adolescentica (n = 763). Rezultati pokazuju da se dvije mreže komponenata stava o izgledu vlastitoga tijela ne razlikuju supstancijalno.; The network approach represents a novel paradigm for exploring relations between psychological constructs and observable variables. According to this approach, variables form an autonomous dynamical system; the psychological construct is therefore not viewed as their common cause, but a result of their complex interactions. From an analytical point of view, this approach is based on network analysis — a set of procedures which models variables as nodes connected by a set of edges. This paper presents an overview of network analytical procedures. In other words, it offers a brief explanation of the methods, as well as their practical application in two separate datasets. The first dataset represents data on DASS-21 (N = 1016) and it serves to demonstrate network estimation, centrality measures calculation, community detection and network stability analyses. According to the results, the highest centrality was obtained for the depression item I felt that I had nothing to look forward to, anxiety item I felt I was close to panic, and stress item I felt that 1 was using a lot of nervous energy. As expected, nodes were grouped into three clusters, namely Depression, Anxiety and Stress. Stability analyses demonstrated limited stability of edge strength, while the stability of node centrality depended on the measure used. In the second dataset, which represents data on adolescents' attitudes towards one's body appearance, the Network Comparison Test was demonstrated by comparing male (n = 524) and female (n = 763) networks. Results showed that the two networks do not differ substantially.