Forecasting Models for the Numbers of Indigenous Graduate Students in the Context of Graduate Schools Expansion in Taiwan—Curvilinear Regression of the Ministry of Education's Statistics from 1998 to 2014

Wednesday, 13 July 2016: 14:15
Location: Hörsaal 47 (Main Building)
Oral Presentation
Chien-Lung WANG, National Taitung University, Taiwan
Juhui CHANG, National Taitung University, Taiwan
In Taiwan, 540,000 Indigenous People of 16 Austronesian ethnicities made up 2.28% of 2.3 million (major Chinese) populations. Their educational disadvantages have been improved since Indigenous Education Lawwas promulgated in 1998. In 2014, Indigenous students made up 2.11% of the undergraduates with additional protective enrolling numbers, while only 0.7% of the graduate students are Indigenous. Actually, the yearly numbers of Indigenous graduate students increased sharply from 18 to 1,239, while non-Indigenous numbers expanded 4.3 times from 43,000 to 185,000 from 1998 to 2009 and then have been scaled down 6.5% to 173,000 by the Ministry of Education ever since. This study hypothesized that with non-Indigenous graduate student numbers as forecasting variable and Indigenous numbers as criterion variables, the forecasting models would be “S-Curve” reflecting the scaling down. With methodology of data mining, the Ministry of Education’s statistics was collected and analyzed with SPSS curvilinear regression models including Growth Model and S-Curve to test the hypothesis.

The findings are as followings: Generally, Growth Model is much more fitted (R2=.953) and the forecasting formula is yindigenous graduate student=e (2.177+0.00002585*non-indigenous graduate student). By variables of genders, birth registrations of Indigenous populous municipalities, and larger ethnicities of Amis and Paiwan, Growth Models also indicated the continuous increasing. It reflected Indigenous population growth rate (1.2%) is higher than that of non-indigenous (0.3%), due to numerous youths from mix-marriage families obtain or restore the statuary Indigenous identity with consanguinity. Besides, the coefficient of Growth Model of females’ is higher than that of males’ and females outnumbered males by 50 in recent 2 years, while non-Indigenous males outnumbered females by 20,000. Finally, by variables of smaller Puyuma, Bunun, and Truku ethnicities, S-Curve is much more fitted and the forecasting formula for Puyuma, for example, is ypuyuma graduate student=e(6.893-611625.999/non-indigenous graduate student), showing the numbers are slowing down.