Homework Assignment #3 -- Feedback

What you were supposed to do:

• poverty
• education
• health
• delinquency
• alcohol abuse

How you should have done it:

Sample solution:
Let us consider the concepts of poverty and of education. First, we need to think about the unit that we want to consider. The basic choice is between "individuals" (people, students, ..) as unit of analysis or "aggregates" (neighborhoods, schools, cities, countries, ..). We will do one hypothesis for each type of unit. [That, by the way, was not required in the assignment.]

Variables are traits or characteristics or properties that units have. So, the choice of unit has consequences for the construction of variables ("operationalization"). Let's start with individuals:

• independent variable: Financial status of parents
• dependent variable: College attendance

Note that it is preferable to use one of the two common formats to state such hypotheses. One format (the one used here) is: "The more xyz ....., the more (or less) abc ....". The other format is the "if  -- then" phrasing. In our example:
"If a child has parents living in poverty, then it is less likely that he/she will get a college education."
Using one of these two format forces you to think more clearly about cause and effect, about independent and dependent variable. Finally, it is preferable, to make explicit reference to the probabilistic nature of the hypotheses by using phrases like "more likely", "better chances", etc. However, reading scholarly articles you may not always find such wording, because it is taken for granted that there are no deterministic explanations in sociology and all (nomothetic) hypotheses are to be understood as probabilistic.

Finally, the assignment called for a brief explanation of why you expect the hypothesis to be "true", to be consistent with empirical data. The key was brief. Something like:
"Based on my own experience from living in a less affluent neighborhood, I think the number of kids my age that go to college is very low."  OR:
"In a course on sociology of education we have discussed what role 'social capital' of the parents plays in determining the educational attainment of their kids." OR:
"Since poor parents can hardly afford to send their kids to private schools and it is difficult to get an education that truly prepares students for college-level work in NYC's public school, the consequence appears to be obvious."

Now to second hypothesis. For simplicity we stay with the same concepts, but consider neighborhoods (like Bensonhurst, Canarsie, Harlem, etc.) as units.

• independent variable: Poverty rate (=number of people per 1000 with income below the official poverty line)
• dependent variable: Percentage of people age 25 and over with at least a bachelor degree

Hypothesis: The higher the poverty rate (in a neighborhood), the lower the percentage of people with at least a bachelor degree.

"In my statistics class, we look at data from 1990 Census for the NYC neighborhoods and we found an association between the two variables." OR:
"If there is a link on the individual level (see above), then this link usually also shows on the aggregate level."

DONE!
 

Some examples of hypotheses with problems:

1. "Inner city children who come from low-income families are more likely to have a high incidence of delinquency"
   There are at least three variables involved here: place of residence, family income, delinquency. Fix: Drop "inner city".
2. "The higher the level of poverty in Third World countries, the lower the level of education completed."
   The first part clearly indicates that the intended units are countries (aggregates), yet the dependent variable does describe a trait countries can have. Fix: use "average level of education" or "high school graduation rate" or "illiteracy rate" as the dependent variable.
3. "The more commonplace alcohol abuse is in the United States, the more expensive health care gets."
   Not clear at all what the units are. Two possible fixes, one relating to the individual level, one to the aggregate.
   "The more alcohol someone drinks, the higher his/her medical expenses will be in the long run" (unit: individuals)
   "The less alcohol abusers a state has, the less money the government has to spend on health care." (unit: states)
4. "Poor Black and Latino children that in live in rural and inner-city areas have less of a chance of getting a good secondary education that has adequate government and local funding."
   Variables involved: race, place of residence, quality of education, source of education funding. Also what is the basis for comparison? This is a general thesis not a hypothesis of the kind discussed in textbook that seeks the test the impact of one specific factor (possible cause) on a particular outcome (effect). A thesis of this kind must be broken down into simpler hypotheses in order to subject it to an empirical test. But this requires a bit more insight into more advanced statistics. So, while this thesis is thoughtful (and while there is indeed empirical evidence to support it) it is a bad response to the assignment -- missing it.
5. "Lower education level may lead to delinquency"
   Again, the unit referred to is not clear. "Groups of people" is a wrong answer. Can be fixed on both the individual and the aggregate level. But I will leave this to you as an exercise.