Daniel has taught Public Health at the graduate level and has a Ph.D. in Behavioral Sciences & Health Education.
Variables in I/O Psychology Research
Data analysis is an integral part of the research process in industrial and organizational psychology. There are many different kinds of statistical methods that are used in the field. So many, in fact, that sometimes it can feel a little overwhelming to try to differentiate between them.
However, one way we can differentiate between statistical methods is by looking at the specific types of variables involved in each method. To do this, we first need to define some basic terminology when it comes to variables.
A predictor variable is a variable that is used to predict another variable in statistical analyses. It is sometimes referred to as an independent variable when it is manipulated in an experiment, rather than just measured. An outcome variable, or dependent variable, is a variable whose value depends on the predictor variable.
Imagine that a business manager, Sheila, gives her employees a survey. The survey asks employees to rate their job satisfaction as low, medium, or high. Sheila wants to understand whether job satisfaction is related to employee age. In this example, employee age is a predictor variable, and job satisfaction is an outcome variable. In general, all variables are either measured on a continuous scale, or they measure categorical or discrete characteristics. In our example, age is a continuous variable and job satisfaction is a categorical variable.
Models & Observed Predictor Variables
The most frequently used statistical methods in I/O psychology involve observed, or directly measured, predictor variables, like the age variable in our example. Let's take a look at a few of them.
Analysis of variance is a statistical method that we can use to analyze the relationship between a categorical predictor variable and a continuous outcome variable. Analysis of variance provides a statistical test of whether or not the means of a continuous outcome variable are equal across different groups, or categories.
We can use multiple regression to analyze the relationship between two or more predictor variables and a continuous outcome variable. Unlike analysis of variance, multiple regression can be used with either continuous or categorical predictor variables.
Logistic regression is another type of regression model that we can use when our outcome variable is categorical. Logistic regression uses one or more predictor variables, which can be either continuous or categorical.
Models & Unobserved Predictor Variables
Sometimes we want to know about variables that we cannot see or directly measure. For example, imagine that our business manager, Sheila, wants to understand whether her employees' job satisfaction is related to their commitment to the organization. Organizational commitment is somewhat of an abstract concept that has several different dimensions to it, which makes it difficult to measure directly.
In statistics, latent variables are variables that are measured indirectly using other observed variables. Latent variables can be inferred through advanced statistical modeling. Factor analysis is a statistical method that helps us find underlying (latent) constructs that describe variability among observed, correlated variables.
So, suppose Sheila included a set of five different questions in her survey that ask about different dimensions of organizational commitment. She could then use factor analysis to see whether the variability in employees' responses to those five questions can be described by an underlying factor.
In statistics, we also have structural equation modeling, which is an advanced statistical method that is closely related to factor analysis. We can use structural equation modeling to examine relationships between two or more latent variables.
Models & Multilevel Predictor Variables
Data in industrial and organizational psychology are commonly organized at multiple, hierarchical levels. For example, let's assume that Sheila is a regional manager for her company, and she oversees employees working at several different offices. If she surveys all of her employees, her data would inherently be organized at two hierarchical levels: the employee level and the office level.
Now Sheila wants to know whether employees who work at offices open on Saturdays have lower job satisfaction than employees who work at offices closed on Saturdays. Whether or not an office is open on Saturdays is an office-level variable, not an employee-level variable. So, Sheila needs to use a statistical method that accounts for this.
Hierarchical linear modeling is one such model. Hierarchical linear modeling is an advanced form of regression that we can use when data are organized at more than one level. Hierarchical linear modeling allows us to model predictor variables at multiple levels.
We've covered quite a few different statistical methods in this lesson. So, let's review how each of them are used:
- Analysis of variance analyzes the relationship between a categorical predictor variable and a continuous outcome variable.
- Multiple regression analyzes the relationship between two or more categorical or continuous predictor variables and a continuous outcome variable.
- Logistic regression analyzes the relationship between one or more categorical or continuous predictor variables and a categorical outcome variable.
- Factor analysis examines unobserved (or latent) variables among observed, correlated variables.
- Structural equation modeling examines relationships between two or more latent variables.
- Hierarchical linear modeling is an advanced form of regression that models predictor variables at multiple levels.
To unlock this lesson you must be a Study.com Member.
Create your account
Register to view this lesson
Unlock Your Education
See for yourself why 30 million people use Study.com
Become a Study.com member and start learning now.Become a Member
Already a member? Log InBack