Page 60 - Pengurusan Prestasi Nilai Teras POISE UM

A 5-point Likert Scale was used to represent each item in the
instrument. This 5-point Likert scale contained five options: (1) Very
Unsatisfactory, (2) Unsatisfactory, (3) Somewhat Satisfactory, (4)
Satisfactory and (5) Very Satisfactory. Respondents could only choose
one option for each item. It was more concise and easier as there were
varying cognitive levels among the respondents while the time taken to
answer the instrument was shorter (Chyung et al., 2007).

The calculation of scores for each scale is as follows:

Scale 1 2 3 4 5
Score (%) 30 60 75 90 100
The selection criteria for these respondents are as follows: (1)
represented staff from the academic and non-academic pools, (2) the
academic staff were represented by 3 categories, namely the Science,
Non-Science and Clinical Fields, (3) the non academic staff were
represented by 2 categories, namely executive staff and operational
staff, (4) Executive staff consisted of N, W and J Schemes whereas
Operational Staff were from the N, F and J Schemes. Such purposive
60 sampling of respondent selection would represent all available staff
categories at Universiti Malaya.
A pilot study had been conducted for 44 stakeholders to assess 10
selected PYDs. The number of respondents in the range of 25 to 100
staff met the requirements of a pilot study (Cooper and Schindler,
2014). The pilot study was conducted for the purpose of improving the
quality and consistency of the study (In, 2017). It also served to identify
the needs to modify the items or any procedures that did not produce
appropriate feedback (Malmqvist et al., 2019).

Further, Rasch measurement model was employed to evaluable the
reliability of this instrument with the use of WINSTEPS 4.8.1 software.
The Rasch Model would focus on the respondents’ abilities as well as
the difficulty of the items (Bond and Fox, 2015).
There were three comparative measures that had to be done to identify
item misfit in this instrument, namely by analysing item polarity, outfit
mean square (MNSQ) value and outfit z-standard (ZSTD) value.
During the initial stage of analysing the Rasch measurement model, all
36 original items and 44 respondents were analysed. The first review
was done on the respondents. Findings showed that there were four
respondents who had to be removed from the analysis because their
outfit MNSQ value exceeded 2.