Sociology 498: Homelessness

Five questions to ask about causal hypotheses:

• For survey research on homelessness (see Rossi or the HUD report), we want to compare homeless people with some group of non-homeless.
• Less often done, but probably better, we could also look among the homeless and compare times when they weren't homeless versus times when they were homeless.
• For research on cities, we compare cities with high homelessness with cities with low homelessness.
• For comparisons over time (see Jencks), we compare years with higher homelessness with years with low homelessness

To test a causal hypothesis, we look for some relationship between homelessness and the supposed cause. For example, if we want to test whether drug abuse is a cause of homelessness, we might compare homeless people with the non-homeless to see if drug abuse rates are higher among the homeless. Or we might compare years over time to see if increases in homelessness are associated with increases in drug abuse.

As we will see, it is hard enough to find good data to make these comparisons. But collecting the data and finding a relationship between homelessness and some possible cause, is only the first step. We need to ask at least five other questions about the relationship of homelessness to this possible cause:

1. How big a relationship?

Are some causes more important, or more common, than others?

A related question is: Is the effect "statistically significant"? This just asks whether the relationship found in the data is bigger than what we might expect to find by chance.

2. Does causality go in both directions?

e.g., homelessness aggravates mental illness so they are associated in part because mental illness is a consequence of homelessness and in part because mental illness may be a cause of homelessness.
 
3. Is there some third factor that causes both?

e.g., maybe family breakdown is related to homelessness only because both family breakdown and homelessness increase with poverty and it is poverty, not family breakdown that is a cause of homelessness:


	poverty	->	family breakdown
			|
	poverty	->	homelessness


 
4. How are the causes linked together in causal chains?

e.g., mental illness -> joblessness -> homelessness
 
5. Is it true at both the macro and micro levels?

e.g., mental illness may explain who becomes homeless, but it may be less helpful in explaining why rates of homelessness have changed over time or are higher in some cities than others.