Partial Least Squares Model Validation
In 2003, 2,226 banks were registered for conducting business in UK, comprising 261 private banks, 504 savings banks, 1,395 cooperative banks and 66 other banks (specialized credit institutions, state owned banks, etc. ). For this research ? 200 largest banks in UK were chosen, based on their total assets as reported in ? balance sheet of ? year 2003 (latest available figures at ? time of preparing ? survey). ? cumulated balance sheets of ? 200 largest banks account for more than 90 per cent of ? cumulated balance sheet of ? whole UK banking market (estimation based on (Bundesbank 2004) and (Karsch 2004).
To assess ? hazard perceptions of ? managers in charge of business processes, four banking processes were selected which are generally not regarded as areas of core competence for banks (Lamberti and Pohler 2004): back office/settlement processes for transactions in securities, consumer credits, domestic payments and foreign exchange/money market. All 200 top banks were contacted by phone to personally identify ? managers responsible for ? business processes mentioned above. Some banks do not offer all four products to their clients, so only 593 questionnaires were sent out.
? time period for sending back ? questionnaires was six weeks, from Might 15 to June 30, 2005. Managers who had not returned ? questionnaire by June 1 received ? phone call asking if they needed assistance. This action resulted in an increased quota of responses. Overall, 218 usable questionnaires from 26 banks were returned out of ? total model of 593 managers in UK’s 200 largest banks. This equals ? response rate of 36. 8% amongst managers and 63% of ? banks approached. ? results of ? survey and further model information are presented in ? next section. Descriptive Data Analysis
Need essay sample on "Partial Least Squares Model Validation"? We will write a custom essay sample specifically for you for only $13.90/page
? cumulated assets of ? responses accounted for more than 90% of ? total cumulated UK banking balance sheet. This is only ? rough estimate, as ? questionnaire asked for ? sum of assets on an interval scale to ensure anonymity. ? response rate amongst large banks (assets > EUR 20bn) was exceptionally high (79. 6%). ? distribution of responses amongst ? banking groups (private banks, savings banks, cooperative banks, other banks) matches ? distribution in ? model, as does ? size of ? banks. ? present state of BPO adoption within each bank was captured by asking ? respondents to choose one of ? six options.
? respondents’ statistical characteristics show ? mean of 8. 75 years of experience in ? present position and an average of 71. 3 employees managed. 85. 9% of ? respondents have ? hierarchical position at ? 2nd or 3rd level of ? corporation. Control Variables In order to control for ? influence of distinctive characteristics of ? individual banks, certain factors were analyzed to see whether they have ? systematic influence: type of procedure, type of bank (? vast majority of UK banks belong to one of three institutionalized groups: private banks, savings banks or cooperative banks (Hackethal 2004)) and outsourcing adoption status.
Multi-Group Analysis (as suggested by (Chin 2000)) was conducted and no statistically significant influence of ? tested variables on ? structural model was detected. Additionally ? effect of organization size was measured through ? one-indicator-construct (only organization size) loading on ? intention to increase ? level of BPO as suggested by Dibbern and Chin (2005).
? result shows weak loading (path coefficient 0. 045) and is not significant (t-value 1.08), implying that in our model organization size has no impact on ? overall intention to increase ? level of BPO. This section presents ? results of ? model validation. This includes ? test of ? measurement model and ? structural model. ? measurement model consists of both reflective and formative measured indicators. ? fundamental difference between these concepts is that formative indicators form or constitute ? latent construct, while reflectively modelled constructs, in contrast, form their indicators (Chin 1998b).