CC BY
4
https ://doi.org/10.29013/EJEMS-20-2-119-124
Chen Gege,
Sandy Spring Friends School, MD E-mail: [email protected]
FINANCIAL WORRIES OVER KIDS' COLLEGE EDUCATION AMONG ADULTS IN 2017
Abstract
Aim: This study aims to 1) examine the predictors of parients' financial worries over kids' college education in 2017 2) build a predictive model for parients' financial worries over kids' college education using artificial neural network and compare its performance to logistic regression model.
Method: The National Health Interview Survey (NHIS) in 2017 was used. All the participants who were eligible were randomly assigned into 2 groups: training sample and testing sample. Two models were built using training sample: artificial neural network and logistic regression. Receiver operating characteristic (ROC) were calculated and compared for these two models for their discrimination capability.
Results: About 53% of 9129 Adults had Financial worries over kids' college education, about 56.7% among the female and 48.5% among the male.
According to the logistic regression, the female was more likely than the male to have financial worries over kids' college education. The non-Hispanic adults were less likely to have financial worries over kids' college education than Hispanic adults. The older adults were less likely to have financial worries over kids' college education. Compared to Northeast region, Midwest and South were less likely to have financial worries over kids' college education. Compared to people who were not employed but looking, people who were not employed and not looking were less worried.
According to this neural network, the most important predictors were age, Hispanic or not, working or not and gender.
For training sample, the ROC was 0.59 for the Logistic regression and 0.67 for the artificial neural network. In testing sample, the ROC was 0.60 for the Logistic regression and 0.62 for the artificial neural network.
Conclusions: In this study, we identified several important predictors for parients' financial worries over kids' college education in 2017 e.g., gender, age, region and working status. This provided important information for social works to design and implement measures for depression prevention. We built a predictive model using artificial neural network as well as logistic regression to provide a tool for early detection. As to performance of these two models, logistic and artificial neural network regression had a similar discriminating capability.
Keywords: Financial worries, college education, social work, financial depression prevention.
1. Introduction: The cost of tuition and room- continued to rise in the 2017-2018 school year, ac-and-board for both public colleges and private ones cording to the College Board. The average tab at a four-
year in-state public college rose 3.1 percent to $20,770, and the cost at private institutions jumped 3.5 percent to $46,950. Roughly 60 percent ofundergraduates between ages 18 and 24 enrolled in a four-year bachelor's degree program that have taken out student loans say they are responsible for covering more than half of the total cost of their education, the survey found [1].
Seven out of 10 college students feel stressed about their personal finances, according to a new national survey. Nearly 60 percent said they worry about having enough money to pay for school, while halfare concerned about paying their monthly expenses [2].
More U. S. parents worry about having enough money to pay for their children's college education than other Americans worry about any common financial concerns. The 73% of parents of children younger than 18 who worry about funding college tops the 70% of lower-income Americans who worry about having enough money to pay for medical costs in the event of a serious illness or accident [3].
This study aims to 1) examine the predictors of parients' financial worries over kids' college education in 2017 2) build a predictive model for parients' financial worries over kids' college education using artificial neural network and compare its performance to logistic regression model.
2. Data and Methods:
Data:
The National Health Interview Survey (NHIS) is the principal source of information on the health of the civilian noninstitutionalized population of the United States and is one of the major data collection programs of the National Center for Health Statistics (NCHS) which is part of the Centers for Disease Control and Prevention (CDC).
The National Health Interview Survey (NHIS) Data 2017 was used in this study.
URL: https://www.cdc.gov/nchs/nhis/about_ nhis.htm
Models:
We used logistic regression models to calculate the predicted risk. Logistic regression is a part of a
category of statistical models called generalized linear models, and it allows one to predict a discrete outcome from a set of variables that may be continuous, discrete, dichotomous, or a combination of these. Typically, the dependent variable is dichoto-mous and the independent variables are either categorical or continuous.
The logistic regression model can be expressed with the formula: ln(P/P-1) = £0 + p1*X1 + (32*X2 + ... .+ (3n*Xn
A package called "neuralnet" in R was used to conduct neural network analysis. The package neuralnet focuses on multi-layer perceptrons (MLP, Bishop, 1995), which are well applicable when modeling functional relationships.
Variables:
The outcome variable is percentage of How worried are you right now about not having enough money to pay for your children's college? (ASIC-COLL)
Table 1.- Variables used in this study
SEX 1: male 2: female
ORIGIN_I Hispanic Ethnicity: 1: yes; 2: no
RACRECI3 1: White 2: Black 3: Asian 4: All other race groups*
AGE_P Age <18 years old 0-17
Region 1 Northeast 2 Midwest 3 South 4 West
3. Results
About 53% of 9129 Adults had Financial worries over kids' college education, about 56.7% among the female and 48.5% among the male.
Basically, a corrgram is a graphical representation of the cells of a matrix of correlations. The idea
is to display the pattern of correlations in terms of the correlation value. The positive correlations are
their signs and magnitudes using visual thinning and shown in blue, while the negative correlations are
correlation-based variable ordering. Moreover, the shown in red; the darker the hue, the greater the
cells of the matrix can be shaded or colored to show magnitude of the correlation.
Figure 1. matrix of correlations between variables
According to the logistic regression, the female over kids' college education. Compared to Northeast
was more likely than the male to have financial wor- region, Midwest and South were less likely to have
ries over kids' college education. The non-Hispanic financial worries over kids' college education. Com-
adults were less likely to have financial worries over pared to people who were not employed but looking,
kids' college education than Hispanic adults. The people who were not employed and not looking were
older adults were less likely to have financial worries less worried.
Table 2.- Logistic Regression for Having Financial worries over kids' college education
Estimate Std. Error z value Pr(> z )
(Intercept) 1.577 0.151 10.415 <2e-16 ***
AGE P -0.017 0.002 -9.866 <2e-16 ***
Male -0.325 0.044 -7.353 0.000 ***
HISPAN NO -0.478 0.059 -8.082 0.000 ***
MARRIED -0.059 0.045 -1.316 0.188
White 0.053 0.077 0.692 0.489
Black 0.033 0.098 0.334 0.738
Midwest -0.159 0.071 -2.235 0.025 *
South -0.167 0.065 -2.558 0.011 *
West -0.131 0.071 -1.854 0.064
Working -0.005 0.087 -0.060 0.952
NWorNLor -0.276 0.096 -2.877 0.004 **
Figure 2. Artificial Neural Network in training sample
In above plot, line thickness represents weight magnitude and line color weight sign (black = positive, grey = negative). The net is essentially a black box so we cannot say that much about the fitting, the weights and the model. Suffice to say that the
training algorithm has converged and therefore the model is ready to be used.
According to this neural network, the most important predictors were age, Hispanic or not, working or not and gender. ents and Midwest residents.
Figure 3. Variable Importance in Artificial Neural Network
For training sample, the ROC was 0.59 for the Logistic regression and 0.62 for the artificial neural Logistic regression and 0.67 for the artificial neural network. network. In testing sample, the ROC was 0.60 for the
False positive rate
Figure 4. ROC in training sample for Logistic Regression (Red) vs Neural Network (Blue)
0.0 0.2 0.4 0.6 0.8 1.0
False positive rate
Figure 5. ROC in testing sample for Logistic Regression (Red) vs Neural Network (Blue)
4. Discussion artificial neural network and compare its perfor-
This study aimd to: 1) examine the predictors of mance to logistic regression model. parients' financial worries over kids' college educa- About 53% of 9129 Adults had Financial worries
tion in 2017; 2) build a predictive model for parients' over kids' college education, about 56.7% among the
financial worries over kids' college education using female and 48.5% among the male. According to the
logistic regression, the female was more likely than the male to have financial worries over kids' college education. The non-Hispanic adults were less likely to have financial worries over kids' college education than Hispanic adults. The older adults were less likely to have financial worries over kids' college education. Compared to Northeast region, Midwest and South were less likely to have financial worries over kids' college education. Compared to people who were not employed but looking, people who were not employed and not looking were less worried. According to this neural network, the most important predictors were age, Hispanic or not, working or not and gender.
In this study, we identified several important predictors for parents' financial worries over kids' college education in 2017 e.g., gender, age, region and working status. This provided important information for social works to design and implement measures for depression prevention. We built a predictive model using artificial neural network as well as logistic regression to provide a tool for early detection. As to performance of these two models, logistic and artificial neural network regression had a similar discriminating capability.
References:
1. URL: https://www.usatoday.com/story/money/personalfinance/2018/07/11/college-student-ques-tion-cost-education-loan-value/772770002
2. URL: https://news.osu.edu/70-percent-of-college-students-stressed-about-finances
3. URL: https://news.gallup.com/poll/182537/parents-college-funding-worries-top-money-concern.aspx
4. Peng C. J., Lee K. L., Ingersoll G. M. An Introduction to Logistic Regression Analysis and Reporting. The Journal of Educational Research, 96(1). - P. 3-14.
5. Tabachnick B. and Fidell L. Using Multivariate Statistics (4th Ed.). Needham Heights, MA: Allyn & Bacon, 2001.
6. Stat Soft. Electronic Statistics Textbook. URL: http://www.statsoft.com/textbook/stathome.html.
7. Stokes M., Davis C. S. Categorical Data Analysis Using the SAS System, SAS Institute Inc., 1995.
