Part D How might you reasonably explain these differences over time? Provide support for your hypotheses.

VARIABLES AND HYPOTHESES

Begin with stating the research question, the purpose of the research, the resources needed, and a plan for the research, including a model of the phenomenon under study.

Where do research ideas come from?  Curiosity; experience; need for deciding or acting; job; school; building on or contesting existing theory; available funding; etc. 

A preliminary research proposal, in one or two pages, a.  states the research question b.  states the purpose of the research c.  sketched the initial model d.  discusses (explains) the initial model e.  identifies pertinent background literature (bibliography)

A model shows how different elements are linked by relationships.  The elements for a model can be drawn from personal experience, consulting with key players, published literature, asking experts, existing data sets, and pilot studies.  Generally a model is fixed at the beginning of the research; it may be altered as a result of the data analysis.

A model is a visual representation of how something works; it both describes and explains some phenomenon.  The advantages and drawbacks of models are: 

  to understand the research projectMay over-simplify the problemExplains the idea to othersMay not meet the client's needsGuides the research processMay not be well-suited to application

Elements of the model are variables.  Variables are measurable characteristics or properties of people or things that can take on different values.  In contrast, characteristics that do not vary are constants.

A hypothesis states a presumed relationship between two variables in a way that can be tested with empirical data.  It may take the form of a cause-effect statement, or an "if x,...then y" statement.

The cause is called the independent variable; and the effect is called the dependent variable.

Relationships can be of several forms:  linear, or non-linear.  Linear relationships can be either direct (positive) or inverse (negative).

In a direct or positive relationship, the values of both variables increase together or decrease together.  That is, if one increases in value, so does the other; if one decreases in value, so does the other.

In an inverse or negative relationship, the values of the variables change in opposite directions.  That is, if the independent variable increases in value, the dependent variable decreases; if the independent variable decreases in value, the dependent variable increases.

In a non-linear relationship, there is no easy way to describe how the values of the dependent variable are affected by changes in the values of the independent variable.

If there is no discernable relationship between two variables, they are said to be unrelated, or to have a null relationship.  Changes in the values of the variables are due to random events, not the influence of one upon the other. 

To establish a causal relationship between two variables, you must establish that four conditions exist:1)  time order:  the cause must exist before the effect;2)  co-variation:  a change in the cause produces a change in the effect;3)  rationale:  there must be a reasonable explanation of why they are related;4)  non-spuriousness:  no other (rival) cause for the effect can be found.

To establish that your causal (independent) variable is the sole cause of the observed effect in the dependent variable, you must introduce rival or control variables.  If the introduction of the control variable does not change the original relationship between the cause and effect variables, then the claim of non-spuriousness is strengthened.

Commonly used control variables for research on people include sex, age, race, education, and income.  Commonly used control variables for research on organizations include agency size (number of employees), stability, mission, budget, and region of the country where located.

For example, consider the placement rates for three training programs.  The independent variable is the type of training, and the dependent variable is the placement rate.

Vocational education has a placement rate of 30%; on-the-job training has a rate of 40%; and work-skill training has a rate of 35%.  It would appear that on-the-job training is the best program, followed by work-skill training, with vocational education last.

However, when education is introduced as a control variable, it can be seen that the effect of the independent variable (type of training) on the dependent variable (placement rate) is quite different for people with different levels of education. 

 Level of Education Vocational EdOn-the-job trainingWork-Skill TrainingLess than high school30%20%50%High School 60%45%15%More than high school20%60%10%Overall rate30%40%35%  

 (Note that there are different numbers of people in each educational category, and different numbers of people in each training program, so the overall rate is not simply the average of the rates for each educational category within each training program).