Judgment of Two Variable Relationship
A lot of what we have discussed so far relates to comparisons between means, which is typically what we do when we use experimental methodology. But in a range of other research situations we are interested in assessing the relationship between two variables. For example, how is height related to weight? How is stress related to heart disease? This type of question can be asked in experiments (what is the relationship between the amount of training and memory?), but is more typically addressed in surveys, where the researcher has multiple values of each variable. Suppose we are working on the concept of attraction, which occurs at many levels. We might have data recording both people’s attraction to their partners and the amount of time they have spent apart from them, our interest lying in whether higher levels of attraction are associated with higher levels of time spent apart, or whether high levels of attraction are associated with lower levels of time spent apart, or whether there is no clear relationship between the two variables. This type of data is described as bivariate, as opposed to univariate. [bivariate the relationship or association between two variables (‘variate’ is another word for variable)] One useful way to set about answering this type of question is to draw a scatterplot – a two-dimensional graph displaying each pair of observations (each participant’s attraction to their partner and the time spent apart). [univariate relating to a single variable] A negative correlation would be obtained when one value decreases as the other increases. Note that the stronger the relationship, the less scattered the various points are from a straight line, and the more confidently we can estimate or predict one variable on the basis of the other. In this example, it becomes easier to estimate from someone’s attraction how much time they have spent apart from their partner,
or to estimate level of attraction from the time spent apart.
A lot of what we have discussed so far relates to comparisons between means, which is typically what we do when we use experimental methodology. But in a range of other research situations we are interested in assessing the relationship between two variables. For example, how is height related to weight? How is stress related to heart disease? This type of question can be asked in experiments (what is the relationship between the amount of training and memory?), but is more typically addressed in surveys, where the researcher has multiple values of each variable. Suppose we are working on the concept of attraction, which occurs at many levels. We might have data recording both people’s attraction to their partners and the amount of time they have spent apart from them, our interest lying in whether higher levels of attraction are associated with higher levels of time spent apart, or whether high levels of attraction are associated with lower levels of time spent apart, or whether there is no clear relationship between the two variables. This type of data is described as bivariate, as opposed to univariate. [bivariate the relationship or association between two variables (‘variate’ is another word for variable)] One useful way to set about answering this type of question is to draw a scatterplot – a two-dimensional graph displaying each pair of observations (each participant’s attraction to their partner and the time spent apart). [univariate relating to a single variable] A negative correlation would be obtained when one value decreases as the other increases. Note that the stronger the relationship, the less scattered the various points are from a straight line, and the more confidently we can estimate or predict one variable on the basis of the other. In this example, it becomes easier to estimate from someone’s attraction how much time they have spent apart from their partner,
or to estimate level of attraction from the time spent apart.
No comments:
Post a Comment