This study applied the documentary - analytical method. The sample size consisted of all centers and branches of PNU in 31 provinces and about 500 centers. In this study, the Ordinary Least Square (OLS) regression model was used for data analysis. To obtain the variables, at first, a list of factors affecting demand in higher education was obtained using the Delphi method, literature review, and interview with scholars. In the Delphi method, the indices and variables were first extracted by examining the documents, followed by distribution to 15 professors of higher education, economic sciences and social sciences, and finally, the indicators were approved. After the required investigations and another interview with scholars, the data of effective variables were finally collected. In this study, the dependent variable was the number of enrollers (The required data for this section are obtained from PNU) during the BA course in PNU from 2001 to 2014, annually. The use of unsteady time series in common methods of econometrics, may result in the creation of false regression. Also, it is required to ensure the steady nature of the applied time series, to estimate the parameters of the studied model. Among the most common methods for this purpose is the augmented Dickey Fuller Test. In this test, relevant statistics of Dickey Fuller test is compared with the critical value of the Mc Kinnon table. If the estimated absolute value is higher than the absolute value of the Mc Kinnon test, the null hypothesis based on existence of unit root is rejected, which refers to the steady nature of the time series. Otherwise, the time series is unsteady and the reliability of time series should be tested through differentiation. In the data analysis section, MICROFIT and MIPLE software were used.
In this research, the coefficient of determination and the adjusted coefficient of determination were used to select independent variables. Variable variations were dependent on two components that are divided according to the described and unexplained changes. The amount of variation described above affects and increases the coefficient of determination. Therefore, using coefficient changes, determining the type of variables is based on the variables that are redundant and effective. In addition, attempts have been made to reduce the variation and do not explain the number of independent variables. These variables are divided to economic, social, and cultural variables and are interpreted after the estimated impact on the number of students. Wald statistics are used as one of the indicators for selecting independent variables and then models are estimated using the OLS method. The reason is that the researchers were dealing with time series that are influenced by independent variables (number of students). Therefore, the use of Autoregressive (AR) or Autoregressive - Moving Average (ARMA) models was irrelevant. This is because independent variables are not interdependent. These data are derived from the time series data of the sources of the Iranian Statistics Center, Ministry of Science, Research and Technology, Ministry of Co-operation, Labor and Social Welfare and PNU.
Furthermore, PNU focuses on income and expenses. This means that the statistical community is comprised of the entire PNU students. The tuition fee for all students in centers and units is transferred to the Center’s central account in Tehran, hence, the number of students can be considered as a national variable and the information of the independent national variables is used.
Regarding the type of effect of independent variables on the dependent variable, the choice of regression lines of the problem and the variables to be considered together, in addition to the theoretical ones, will also depend on the significance of the regression lines. This means that when only the effect of a set of variables on the dependent variable is considered, other effective variables are inevitably avoided and this may lead to deviations. Hence, attention has to be paid to the choice of variables, because only factors affecting the number of students are not cultural or just economic factors that explain the changes in the dependent variable. Therefore, inevitably, and economic and labor market variables may be seen in a model together. This increases the R-squared and increases the explanatory power of the regression lines. For the purpose of data analysis, four regression models are used as follows:
DLNS1 = 2.604 - 0.016 + 0.006 - 0.766 + 0.146 + 0.375 + 0.144 + 0.399 + 0.294
R-Squared = 0.95; DW=1.956.
Where; DLNS1 = number of enrollers in Bachelor course of PNU; C = intercept; LBT = total state budget; LYK = average income of manufacturing laborers; LU = unemployment rate of youth of age 15-24; DLNT = total state population; DLCPI = total consumer price index; DLYF = income per master degree; DLN2 = workforce graduated of high school; DLBA = higher education budget.
DLNS2 = 12.325 + 8.458 + 0.087 + 0.583 - 22.107 + 21.310 + 2.824 + 0.744 - 0.083
R-Squared = 0.95; DW = 1.376.
Where; DLSE = total employed portion in industry; LSEE = ratio of educated employees to total employees; DLCT = Student scholarship and loan; LGDP= gross domestic production; LGNP = gross national production; DLNEE = total workforce; LYA = average annual household income;
DLYD= income per diploma degree.
DLNS3 = -0.418 + 0.412 + 0.111 - 2.918 - 0.598 - 0.078 + 0.045
R-squared = 0.41; DW = 2.
Where; LB1 = urban household dimension; LS3 = population of primary student; LCA = average household cost; LYL = Bachelor degree income; LNE = number of employed people in industrial workshops; DLN3 = academic graduated workforce.
DLNS4 = -30.869 + 5.211 - 4.241 + 0.853 + 0.725 - 6.363 + 0.822 + 4.406 + 0.862 - 0.955
R-Squared = 0.65; DW = 2.07.
Where; LB2 = rural household dimension; DLS2 = number of secondary school students; LS1 = number of high school students; LSK = agriculture department tuition; DLSA = basic science tuition; DLSAE = human science tuition; DLN1 = educating or graduated population in labor market; LYE = national income; LSF = technical and engineering department tuition.