What Causes Gender Inequality?
     ... Analytical Strategies

What Causes Gender Inequality? ... Analytical Strategies

SOC-GA 2227

Robert Max Jackson

~~~~~  Notes: Isolating Causes  ~~~~~

~~  Competing or Complementary  ~~

When research or theoretical arguments try to judge the value of two or more explanations for some outcome, we may usefully distinguish between the cases when the alternative explanations are competing from those that are complementaryCompeting explanations are mutually exclusive, so they suggest analyses trying to decide which are valid and which are not.  Complementary explanations do not contradict each other, and analyses of them try to ascertain their relative importance in explaining outcomes.

If we have two or more possible causal explanations of some social phenomena (such as why women’s average incomes are lower than those of men or why some groups perceive Muslims as a threat and others do not), they may be strictly competing or loosely complementary. 

Before going further, let me say that these are not official or formal terms.  We are just applying the common meanings of these words here.  The goal is to understand how the goals and opportunities available from comparing alternative explanations depend on their distinctiveness as causes.

We can think of strictly competing explanations as ones that disagree – they cannot both (or all) be correct.  Competing explanations will produce contradictory predictions under some circumstances.   Let’s consider a simple example.  While considering how the educational achievements of parents (in heterosexual marriages) might influence the educational aspirations of their children, different authors might offer these arguments: (1) the level of the higher achieving parent is decisive, lower achievement by a second parent has little effect; (2) the level of the same-sex parent is the main influence, with just a slight effect of the opposite-sex parent; (3) the educational aspirations of children mainly respond to the household economic status and the appearance of an influence by parents’ education is merely a spurious effect due to the correlation between parents’ educational levels and household income.  In generally, only one of these three explanations could be correct (they could all be wrong, of course).  This means that with a good research design or theoretical argument, an analysis can hope to discredit at least two of these explanations.

Complementary explanations imply that two or more independent causal processes could coexist.  Returning to the previous example, the alternate simple arguments might be that children’s educational aspirations are a result of: (1) parents’ education levels, (2) the quality of schools attended while young, or (3) inherent intelligence.  While we may expect that any or all of these proposed causal conditions, or variables, have some direct or indirect causal links (correlated in the data), the propositions linking each to the outcome variable of children’s educational aspirations are  largely independent.  Any combination of them could be true. 

With competing explanations, our research tries to discover which is valid.  This is a classic exercise in the furthering of science.  The goal is to show that some possible causal relationship does not exist, implying a flaw in theories that assert it does.  One fundamental  criterion for a good social theory or explanation is that it be sufficiently unambiguous and precise that it leads to empirical predictions that we can, at least hypothetically, test. 

In contrast, research on complementary explanations usually does not concern their validity, but their explanatory importance or power under the circumstances being studied.  Most social conditions or processes are affected by varied influences, and each influence is validly considered a contributing cause.  How much influence each cause has over the outcome will depend in part on the way the system of causes works, how all the relevant processes and conditions connect to each other.  The relative influence of various causes also depends on the outcome’s sensitivity to variations in each cause and the amount that each cause varies.  To return to our hypothetical example, assume we suspect that parents’ education, quality of elementary schools, and inherent intelligence all influence educational aspirations.  Sometimes, from a practical viewpoint, assessing how much each of these causes influences the outcomes may be just as, and sometimes even more, important than identifying their locations in a good causal model. 

Realistically, complementary explanations are far more common in sociological scholarship.  For this reason, conclusions to scholarly research articles very often offer as their final claim that the findings “are consistent” with one or more explanations, or that the findings “offer greater support” for one explanation (or more) compared to others.  Such articles rarely attempt claims that one explanation is valid and others are not. 

Research comparing competing explanations usually has different goals than research with complementary explanations.  From a scientific perspective, research aimed at competing explanations is more compelling because it has the clear potential to advance science by showing an explanation or theory is untenable.  From a practical perspective, most social science research aims at complementary explanations, because such research designs are far easier to achieve but still worthy.

By itself, the distinction between competing and complementary explanations does not tell us how valuable a research project might be or what statistical procedures will be a good choice.  The distinction does potentially allow us to think more clearly about the aims of a research project and, therefore, what will be a good design for it.