Search
share
Search
Young Educated Poor Childless Males: The Philosophical Profile Survey Analysis on Demographics

Date

author

share

bigstock-Survey-And-Mouse-Sign-37757554

Analyzing the Survey

About 6 months ago, Philosophy News published a Philosophical Profile survey to gather data on the philosophical views of our readers. We wanted to better understand how our readership thinks about philosophy and better understand who you are. Since that time, over 500 users have participated in the survey in one way or another. We believe we have enough data to draw some tentative conclusions. We start with the demographic data.

Note: if you haven’t taken our survey, or completed all parts of the survey, please do so. it won’t take too long and will help provide better data for future studies.

Summary of Findings

  • Nearly 80% of respondents are male
  • Most are between 23 and 60 years old.
  • More than 80% are college educated.
  • 65% have no children.
  • Survey respondents are 43% undergraduates, 28% young workers, 15% established professors and parents, and 13% graduate students/young professors.
  • The attributes of having little money (0-$20K), being young (18-22), having no children, and being enrolled in a college/university are all associated with each other.
  • Those making between $20K and $40K mostly have 0 kids. This increases to 2 children between $40K-$60K and 3 children between $60K-$80K. However, this trend does not continue. Those making between $80K-$100K have 2 kids and those making $100K+ or more have one child on average.
  • Younger respondents are still in their undergraduate programs. They make little money and do not have children. After undergraduate studies, respondents either enter into the work force or they enter into grad school. They are making a little more money and are starting to have children. After graduate school, respondents are making significantly more, having many more children, and are no longer in school.
  • Everyone that believes Philosophy of Language is the most important field of study is male.
  • Nearly everyone that has an “Other” meta-ethical view does not have children. Those that have an “Other” view of meta-ethics and do not have children tend to have more “liberal” views on applied ethical issues compared with those that do have children (those with children have a more “traditional” or “conservative” views on ethics). Perhaps one’s ethical views influence one’s decision to have children or children change how we think about the world.

Quick Stats

Here are some quick stats.

  • 372 / 512 survey takers (73%) responded to the first part of the survey.
  • 79% are male, 20% are female, and 1% are other. This is slightly skewed towards men in comparison with the percentages of men and women in academic philosophy and graduate school (70% men and 30% women graduate school; 75% men and 25% women faculty members).
  • 71% are between 23 and 60 years old; 20% are between 18 and 22.
  • 83% have had college education in some respect; 57% have at least a bachelors degree;
  • 58% are not currently enrolled as a student; 10% are in graduate school; 24% are in undergraduate school
  • 34% are in an “Other” occupation; 21% are teachers or in education; 10% are in engineering and technology
  • 65% have 0 children
  • 57% make less than $40K a year; 37% make less than $20K a year; 20% make between $20K and $40K a year.

Clusters

A cluster is simply a grouping of similar respondents taken from the whole population of respondents. For example, we might form a cluster based around all of those respondents that have children. These would form one group, while those that have no children would form another group (so we would have two clusters). We can let an algorithm do the clustering for us to put each respondent into a cluster with those respondents most similar to him or her based on all answers given in this survey. I use a Simple K Means clustering algorithm to generate 5 clusters (Note: this is the maximum number of clusters where each cluster contains at least 5% of the population). Now let’s look at the defining characteristics of each cluster to determine what sort of grouping each cluster is, and accordingly, give it a name:

Cluster 0: The Undergraduate (37%)

  • These people are, in general, soon-to-be, current, or recent undergraduate students. They have very few children, make little money, are currently/recently in school, are very young, and do not have a specific career path yet.
  • 46% are in undergraduate college; 36% are not currently enrolled
  • 84% have 0 children
  • 39% have some college but no degree; 19% have a bachelors; 28% have a high school degree
  • 75% make less than $20K a year
  • 85% are male
  • 54% are 23-30; 34% are 18-22
  • 44% are Other in occupation; 10% are teaching and education

Cluster 1: The Young Worker (28%)

  • This person has graduated from college and is no longer in school. He or she is more likely to have children. This person is starting to make money, but because he or she is still young, this person has not yet moved into the higher income brackets.
  • 89% are not currently enrolled as a student
  • 50% have 0 children; 23% have two children
  • 64% have a bachelors degree
  • 37% make between $20K and $40K a year; 19% make $40K-$60K
  • 85% are male
  • 54% are 23-39; 34% are 18-22
  • 44% selected Other for their occupation

Cluster 2: The Graduate Student/Young Professor (13%)

  • Most are in graduate school, and the rest are mostly not enrolled. Everyone has a higher education degree, and most have a graduate degree. Nearly all are still fairly young, but not as young as other clusters. Nearly a third are in teaching and education.
  • 58% are in graduate school; 33% are not currently enrolled
  • 72% have no children; 10% have 2 children
  • 83% have a graduate degree; everyone else has a bachelors
  • 41% are making $20K to $40K a year
  • 77% are male
  • 93% are 23-39
  • 33% are other in occupation; 31% are in teaching and education

Cluster 3: The Female Undergraduate (6%)

  • This cluster consists mostly of undergraduate women that already have an associates degree. They have no children and make very little money. They are college-aged.
  • 75% are undergraduates
  • 87% have 0 children
  • 45% have an associates degree; 20% have a high school degree
  • 66% make less than $20K a year
  • 79% are female
  • 70% are 18-22
  • 50% are Other in occupation

Cluster 4: The Established Professor and Parent (15%)

  • The people are no longer students. Nearly all have graduate degrees and are in teaching and education. Since they are older and are making more money, most have children.
  • 94% are not enrolled as students
  • 29% have 0 children; 24% have 2 children; 19% have 3 children
  • 92% have a graduate degree
  • 33% make $60K-$80K; 15% make $100K+
  • 82% are male
  • 59% are 40-59
  • 75% are in teaching and education

In sum, it seems we have something like this breakdown: 43% undergraduates, 28% young workers, 15% established professors and parents, and 13% graduate students/young professors.

Associations

How do the answers to these questions relate? Associations describe the likelihood of respondents to fit into categories based on responses to questions in other categories. For example, if you fit into one or two categories (e.g., low income and young in age), then we can likely say that you fit into another category (e.g., no children). Based on these associations of responses, we can generate rules in the form of if-then statements. For example, if you have low income and are young in age, then you will have no children. Each of these rules has a probability associated with it to say how accurate the rule is.

Let’s look at some of the more interesting association rules, ordered by confidence greater than 0.85 (i.e., ratio of the accuracy of the rule applied to all cases):

  • What is your average yearly income?=0-$19,999 Which category below includes your age?=18-22 ==> How many children do you have?=0.0
    • 100% confidence for 58 respondents
  • Are you currently enrolled as a student?=Yes, at a college/university What is your average yearly income?=0-$19,999 ==> How many children do you have?=0.0
    • 98% confidence for 60 respondents
  • Are you currently enrolled as a student?=Yes, at a college/university Which category below includes your age?=18-22 ==> How many children do you have?=0.0
    • 98% confidence for 57 respondents
  • Are you currently enrolled as a student?=Yes, at a college/university How many children do you have?=0.0 What is your gender?=Male Which category below includes your age?=18-22 ==> What is your average yearly income?=0-$19,999
    • 89% confidence for 41 respondents
  • Are you currently enrolled as a student?=Yes, at a college/university What is your gender?=Male Which category below includes your age?=18-22==> What is your average yearly income?=0-$19,999
    • 87% confidence for 41 respondents
  • Are you currently enrolled as a student?=Yes, at a college/university==> How many children do you have?=0.0
    • 86% confidence for 76 respondents
  • What is your average yearly income?=0-$19,999 What is your gender?=Male Which category below includes your age?=18-22==> Are you currently enrolled as a student?=Yes, at a college/university
    • 85% confidence for 41 respondents

Not surprisingly, we can see that having little money (0-$20K), being young (18-22), having 0 children, and being enrolled in a college/university are all associated with each other. So if you fit into one or two of these categories, then you are likely to fit into all of them.

Predictions

Suppose we want to predict one’s gender, age, or income? Similar to the above association rules, we can use the answers to other questions to predict these values. For example, suppose I know that no one under 20 has a child in my sample of respondents. Then I know that being under 20 and not having children are highly correlated in my sample. I can take age to thus be a decent predictor of whether or not one has children. Then if I get another respondent that is under 20, I can predict that this respondent also does not have any children.

I looked at each of the questions and attempted to find out which other questions best predict the answers to that question. That is, I used the answers to all of the other questions to build models that predict the answers to the question under consideration, and looked at which specific other questions are the most important in making an accurate prediction.

Note: I used the measure of information gain to rank the importance of attributes in prediction. Information gain is a measure of the reduction of uncertainty produced by using an attribute to classify the target attribute. The more information gain, the more important and useful the attribute is in predicting a different attribute.

Gender

  • Because the data is so lopsided towards men, the best model simply predicts that all respondents are men (with 79% accuracy) without using any other questions for input. That isn’t very helpful. Looking at which attributes are best at predicting Gender, the best indicator is the kind of occupation indicated. But that only helps the model by 6%. Until there is more (and balanced) data, there isn’t much we can say about predicting Gender based on other attributes.

Age

  • Age is a little easier to predict. Although a simple model is only 66% accurate, the most important attribute in predicting age is the enrollment status of the respondent. This helps the model by 46% on its own. Second and most important is one’s average yearly income, at 36%. This makes sense as younger people are more likely to be enrolled as students and to make little money, but older people are more likely to be in graduate school or out of school, and to make more money. These two attributes combined produce the best predictive model.

Highest Degree

  • This again is difficult to predict. Age is most important at 28% (the older you are, the higher your degree is likely to be), followed by enrollment status at 27%. A model built on these two attributes is only 57% accurate: hardly illuminating.

Student Status

  • The most important attribute in predicting the enrollment status of a respondent is age (at 46%). Between 18-22, one is likely to be in college. 17 or younger, one is likely to only have a high school degree or equivalent. Over 22, most respondents are no longer in school. Using this logic, the model predicts with 73% accuracy.

Current Occupation

  • The most important predictor of current occupation is one’s highest degree (22% information gain) followed by income (18% information gain). However, with so little data and so many choices of occupation, it is difficult to make an accurate prediction with a model.

Children

  • The best predictors of the number of children one has are age (29%) and income (27%). The model predicts that the young (below 22) have 0 children and those 60 or older have 2 children. In between, most 23-39 year olds have 0 children. With 40-59 year olds, the income is what makes the difference. Those making less than $20K have 0 children. Those making between $20K and $40K mostly have 0 kids. $40K-$60K: 2 kids; $60K-$80K: 3 kids; $80K-$100K: 2 kids; $100K+ or more: 1 kid. The model is 68% accurate.
  • This is very interesting. It is understandable that those making very little money would not have children, and that as money increases, the number of children increases. This indicates more money to take care of more children, and probably also a longer period of time working in one’s career (so a longer period of time available for supporting children). But then it goes down after $80K. Why? Perhaps as the income increases, the job responsibilities increase, so those who are working in higher paying jobs have less time for family obligations and have consequently opted for smaller families. But this is only speculation. We’d need more data for more conclusive analysis. Thoughts?

Income

  • Age (at 36%) and children (at 27%) are the best predictors of income. This is likely due to the reasoning noted above. If you have more children and are older, you likely have more income to support your children and through longer work experience.

Summary

We don’t have a whole lot of data to work with, but we can see some trends forming, and these match with our clusters. Younger respondents are still in their undergraduate programs. They make little money and do not have children. After undergraduate studies, respondents either enter into the work force or they enter into grad school. They are making a little more money and are starting to have children. After graduate school, respondents are making significantly more, having many more children, and are no longer in school. Careers are now more defined with more respondents in teaching and education jobs. All of this is pretty unsurprising and is intuitive.

Relationships to Other Survey Questions

Finally, do any other answers to questions in the overall survey predict these 7 categories better? I took the answers of everyone who had finished each section of the survey. There were 174 respondents that did so. Here is what I discovered, based on their completed answers:

Gender

  • The best overall predictor of Gender is the major philosophical category that one considers to be most important (8% info gain).
  • 100% of respondents that chose Philosophy of Language were male (18 respondents), 85% that chose epistemology were male (35 respondents), and 85% that chose metaphysics were male (23 respondents).
  • Any thoughts on why this is? What is it about Philosophy of Language that makes it generally more appealing and important to men than women?

Age

  • The best predictor of Age (amongst the other survey questions) is the number of years one has been studying philosophy (27% info gain).
  • 79% of 18-22 year olds have been studying philosophy 1-5 years (26).
  • I imagine that these are currently undergraduates who has just started taking college classes in philosophy or are majoring in philosophy.

Highest Degree

  • The best predictors of one’s highest degree depend on answers to the following questions: What is your education in philosophy? (49% info gain), Which early modern philosophy is the second most important? (22%), and Which early modern philosopher is the most important? (22%).
  • 85% of people with a Master’s Degree in philosophy have a graduate degree (20) (Note: this should be 100%, but 3 people that reported having a Master’s in philosophy reported not having a graduate degree as their highest degree. I don’t know why.); 84% of respondents with a Bachelor’s degree in philosophy have that as their highest degree (25).
  • These are pretty obvious associations.

Student Status

  • The best predictor of student status is the number of years one has been studying philosophy (21% info gain).
  • 88% of those that have been studying philosophy for 11 to 20 years are not currently enrolled as students (25); 78% of those that are currently enrolled as students at a college/university have been studying philosophy 1-5 years (37).
  • These are also pretty obvious associations, as one likely discovers philosophy in college, so one must already be out of college if one has been studying it for 11-20 years. Those that are currently in college have only recently discovered philosophy, and so have only been studying it for a short while.

Current Occupation

  • Here is a list of the top 3 questions for predicting current occupation, ordered by info gain:
    • 0.36 What is your education in philosophy?
    • 0.27 Which existential/continental philosopher is the most important in contributing to subsequent philosophical thought?
    • 0.27 Which of the major philosophical categories is MOST important/central/foundational to properly understanding/framing the other philosophical categories?
  • Unfortunately, since there are so many current occupations, the association rules that are generated to predict current occupation aren’t very reliable (below 50%), and generally reflect answers which mostly everyone selected anyway.

Children

  • One’s most important modern philosopher selection contributes an information gain of 28%. One’s position on meta-ethics contributes 24%.
  • 92% of those that have an “Other” meta-ethical view do not have children (24); 72% of those who believe Russell is the more important modern philosopher do not have children (31); 64% of those who believe Wittgenstein to be the most important modern philosopher do not have children.
  • The fact that Russell and Wittgenstein fans do not have children is not surprising because most respondents chose one of these two and most respondents have no children.
  • However, while most respondents do not have children (65%), very few respondents chose “Other” as their meta-ethical view (13%), and they have no children at 92%, which is much higher than average.
    • Why would having an “Other” meta-ethical view lead one to not have children?
      • Cultural Relativists have on average 1.31 children; Emotivists have 0.47 children; “I don’t knows” have 1.85 children; “Others” have 0.0; Realists have 1.3; Subjectivists have 0.87.
      • As a group, “Others” fall outside 3 standard deviations in their average number of children from the average of all of the groups, so there does seem to be something significant going on here, but without knowing more about what “Other” means it is difficult to tell a story about why this is the case.
    • Is it a byproduct of something else in common?
      • No one that chose “Other” for Meta-ethics and do not have children believe that abortion generally is morally impermissible, while those that did choose “Other” and do have children were split on the issue.
      • No one that chose “Other” for Meta-ethics and do not have children believe that capital punishment generally is morally impermissible, while those that did choose “Other” and do have children are generally in favor of capital punishment.
      • Very few that chose “Other” for Meta-ethics and do not have children believe that virtue ethics is the right approach to normative ethics (most favor utilitarianism), while those that did choose “Other” and do have children generally do believe that virtue ethics is the right approach.
      • Very few that chose “Other” for Meta-ethics and do not have children believe that libertarianism is the right approach to political philosophy (most favored liberalism), while those that did choose “Other” and do have children generally do believe that libertarianism is the right approach.
      • All of those with children are not currently in school. Only 31% of those without children are not in school.
      • Most of those with children are older than those that do not have children.
      • Most of those without children are atheists, while those with children are split on the issue.
  • It thus seems that those that do have children are more “traditional” or “conservative” than those that do not have children, particularly on the ethical, political, and religious issues. We will look at these issues more in depth in later analyses. In any case, age and enrollment status are still much better predictors of children than anything else, and since this is a very small sample, we shouldn’t take too much from these observations.

Income

  • Here is a list of the top 3 questions for predicting income, ordered by info gain:
    • 0.26 Which early modern philosopher is the most important in contributing to subsequent philosophical thought?
    • 0.24 Which early modern philosopher is the second most important in contributing to subsequent philosophical thought?
    • 0.23 Applied Ethics: Capital Punishment is…
  • The best association rules found using other survey questions aren’t really that great: 54% of respondents that believe capital punishment is never morally permissible and have been studying philosophy 1-5 years have an average yearly income of $0-$20K (37); 52% of respondents that have been studying philosophy 1-5 years have an average yearly income of $0-$20K. Nothing very reliable here.

Moving Forward

With such a small sample, it is difficult to draw any firm conclusions. However, if more people respond to the survey we can gather more data. So, if you like the sort of results, observations, and analyses done above, and you haven’t taken or completed our survey (upon which this data is based), please do so. We’ll be able to provide better results and more interesting conclusions for you in the future.

Next up: Philosophical Expertise (coming soon)

Thanks,

Andy Carson
pn_logo16x16_thumb1Philosophy News

More
articles

More
news