Final Project Two
1.Research Question
Does educational attainment impact ideal family size for US Americans?
Many countries in the world are struggling with their fertility levels. In some regions of the world, the fertility rate is extremely high, and governments struggle to serve incredibly young populations, as well as curb population growth. On the other hand, countries such as Italy and Japan are witnessing their population shrink as their fertility rate drops below replacement level. It is critical to understand the different factors that impact fertility rates. One of the factors that impacts fertility rates is womens’ desired fertility. Many programs focus on education impact on fertility in developing countries, but I am curious to see how education affects ideal family size in my home country of the USA. Understanding trends in ideal family size by education levels is one way of figuring out which groups would be most effective to target with fertility interventions– whether to encourage increased child rearing or by assessing if there is a discrepancy between ideal family size and actual fertility rate in some families.
According to previous research, the majority of Americans consider 2 children to be the ideal family size. I expect for the majority of my data to indicate this. I am interested in seeing where the family size preference is slightly higher or lower than 2. I expect those with higher education to prefer slightly smaller families, as with more education they may begin to see a larger trade off with each additional child.While I do not expect to find any large, shocking differences among education groups, I believe the lack of difference is also an important finding. If families across all education levels have similar ideal family size, further studies could look at the difference between ideal and realized (how many children do they actually have). If highly educated Americans actually want two children, but are only able to have one because of timing constraints with higher education, than fertility policy might look into creating more family friendly PhD environments, for example. 2. Control Variables
Educational attainment and ideal family size are correlated with many other characteristics, and thus it is important to control for these other factors to ensure that my regression only looks at the relationship between education and ideal family size. I will control for these variables:
Age: I will control for age as age is correlated with education, as in people are more likely to have higher education if they have been alive for longer. Older people may have certain beliefs about ideal family size that is actually caused by their age, rather than the amount of schooling they’ve completed.
Race: Race may have an impact on educational attainment, while it is also possible that people of different races have different ideal family size due to cultural preference. Controlling for race ensure that any differences in idea family size that are correlated with educational attainment, but that are actually due to racial differences, will not be included in the relationship between educational attainment and ideal family size.
Amount of children: I will control for the amount of children the respondent says they have because it is likely that the amount of children will not exceed the ideal size– ie families of three children are not likely to report an ideal family size of only two. This will help ensure the ideal family sizes reported are not due to a different factor like the current amount of children.
Religion: Some religions may have different ideas about ideal family size. Those religions could also be associated with higher education levels because of the cultural value of education in those religious communities. This is important to control for so that we can see if the different types of religions may impact ideal family size, but appears to be from educational levels.
Social class: Social class plays a role in both educational attainment, as higher social classes may have more access to educational resources that are not free in the USA. Social class could also impact ideal family size, as lower social classes may not feel they have the resources to provide for many children. Controlling for social class helps us make sure we are looking solely at the impact of education, not of the effects of social class masked as education attainment.
Question 4. Regression Equation
ideal family size=β0+β1educ+β2race0+β3race1+ β4race2+β5relig0+β6relig1+β7relig2+β8relig2+β9relig3+β10relig4+β11relig5+β12relig6++β13class+β14age+β15female+ε, where education (educ) is a continuous variable that looks at years of education completed, where race is using 0 as the reference (white), 1 as black and 2 as other (binary variables for yes or no for each subsection), religion is measured with dummy binary variables where 0 is christian (all types), 1 is Jewish 2 is Buddhist, 3 is Muslim, 4 is Hindu, 5 is other and 6 is none, class is a continuous variable of 0 for lower class, 1 is for working class, 2 for middle class, and 3 for upper class,age is measured as a continuous variable based on respondent’s self reported age, and lastly, sex is measured as a binary where female is 0 and male is 1.
Question 5.
This regression assesses if there is an impact of educational attainment on ideal family size. The coefficient β1 measures the different educational levels of respondents, and the other variables are controls for sex, age, religion, social class, race and amount of children.
Question 6.
## Warning: package 'readxl' was built under R version 4.3.2## Warning: package 'rmarkdown' was built under R version 4.3.2## Warning: NAs introduced by coercion## Warning: package 'dplyr' was built under R version 4.3.2##
## Attaching package: 'dplyr'## The following objects are masked from 'package:stats':
##
## filter, lag## The following objects are masked from 'package:base':
##
## intersect, setdiff, setequal, union## Warning: package 'rstatix' was built under R version 4.3.2##
## Attaching package: 'rstatix'## The following object is masked from 'package:stats':
##
## filter## # A tibble: 4 × 10
## variable n min max median iqr mean sd se ci
## <fct> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
## 1 chldidel 1661 0 7 2 1 2.42 0.871 0.021 0.042
## 2 age 1661 19 89 54 29 52.8 17.2 0.421 0.826
## 3 childs 1658 0 7 2 2 1.61 1.38 0.034 0.067
## 4 educ 1661 0 20 15 4 14.7 2.85 0.07 0.137## # A tibble: 8 × 11
## sex variable n min max median iqr mean sd se ci
## <chr> <fct> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
## 1 FEMALE chldidel 872 0 7 2 1 2.41 0.874 0.03 0.058
## 2 FEMALE age 872 19 89 54 27 52.2 16.6 0.561 1.10
## 3 FEMALE childs 872 0 7 2 3 1.72 1.40 0.047 0.093
## 4 FEMALE educ 872 0 20 14 4 14.4 2.8 0.095 0.186
## 5 MALE chldidel 789 0 7 2 1 2.42 0.869 0.031 0.061
## 6 MALE age 789 19 89 54 31 53.6 17.8 0.633 1.24
## 7 MALE childs 786 0 7 2 2 1.50 1.36 0.049 0.095
## 8 MALE educ 789 0 20 15 5 14.9 2.89 0.103 0.202## # A tibble: 13 × 2
## relig n
## <chr> <int>
## 1 Buddhism 20
## 2 Catholic 350
## 3 Christian 34
## 4 Hinduism 17
## 5 Inter-nondenominational 6
## 6 Jewish 33
## 7 Muslim/islam 10
## 8 Native american 1
## 9 None 489
## 10 Orthodox-christian 16
## 11 Other 21
## 12 Other eastern religions 1
## 13 Protestant 663## # A tibble: 3 × 2
## race n
## <chr> <int>
## 1 Black 176
## 2 Other 174
## 3 White 1311## # A tibble: 2 × 2
## sex n
## <chr> <int>
## 1 FEMALE 872
## 2 MALE 789## # A tibble: 4 × 2
## class_ n
## <chr> <int>
## 1 Lower class 141
## 2 Middle class 852
## 3 Upper class 76
## 4 Working class 592One interaction that I want to test is the interaction between age and education attainment. It makes sense to test this because social class may impact educational attainment differently at different “levels” of social class. This means that social class may cause education levels to increase sharply for upper class individuals, but the difference between working class and middle class might be much smaller, since education is expensive for nearly everyone except for the upper class. So the jump from lower class to working class might not impact educational levels very much, but when we jump from middle to upper our educational levels skyrocket. This will require more research and analysis, but I am excited to see the results of this interaction variable.