Generalizability theory is a modern, powerful, and broad framework used to assess the reliability, or dependability, of measurements. While there exist classic works that explain the basic concepts and mathematical foundations of the method, there is currently a lack of resources addressing computational resources for those researchers wishing to…

In institutional research, modern data mining approaches are seldom considered to address predictive analytics problems. The goal of this paper is to highlight the advantages of tree-based machine learning algorithms over classic (logistic) regression methods for data-informed decision making in higher education problems, and stress the success of…

Missing data patterns are the combinations in which the variables with missing values occur. Exploring these patterns in multivariate data can be very useful but there are few specialized tools. The current paper presents a plot that includes relevant information for visualizing these patterns. The plot is also dynamic-interactive; so, selecting…

Predictive studies have been widely undertaken in the field of education to provide strategic information about the extensive set of processes related to teaching and learning, as well as about what variables predict certain educational outcomes, such as academic achievement or dropout. As in any other area, there is a set of standard techniques…

Hierarchical linear modeling (HLM) is a useful tool when analyzing data collected from groups. There are many decisions to be made when constructing and estimating a model in HLM including which estimation technique to use. Three of the estimation techniques available when analyzing data with HLM are maximum likelihood, restricted maximum…

Researchers and data analysts are sometimes faced with the problem of very small samples, where the number of variables approaches or exceeds the overall sample size; i.e. high dimensional data. In such cases, standard statistical models such as regression or analysis of variance cannot be used, either because the resulting parameter estimates…

The Delphi survey technique is an iterative mail or electronic (e-mail or web-based) survey method used to obtain agreement or consensus among a group of experts in a specific field on a particular issue through a well-designed and systematic multiple sequential rounds of survey administrations. Each of the multiple rounds of the Delphi survey…

Often it is infeasible or unethical to use random assignment in educational settings to study important constructs and questions. Hence, educational research often uses observational data, such as large-scale secondary data sets and state and school district data, and quasi-experimental designs. One method of reducing selection bias in estimations…

Many assessment and evaluation studies use statistical hypothesis tests, such as the independent samples t test or analysis of variance, to test the equality of two or more means for gender, age groups, cultures or language group comparisons. In addition, some, but far fewer, studies compare variability across these same groups or research…

Large surveys often use probability sampling in order to obtain representative samples, and these data sets are valuable tools for researchers in all areas of science. Yet many researchers are not formally prepared to appropriately utilize these resources. Indeed, users of one popular dataset were generally found "not" to have modeled…

Yule (1903) and Simpson (1951) described a statistical paradox that occurs when data is aggregated. In such situations, aggregated data may reveal a trend that directly contrasts those of sub-groups trends. In fact, the aggregate data trends may even be opposite in direction of sub-group trends. To reveal Yule-Simpson's paradox (YSP)-type…

Although they test somewhat different hypotheses, analysis of gain scores (or its repeated-measures analog) and analysis of covariance are both common methods that researchers use for pre-post data. The results of the two approaches yield non-comparable outcomes, but since the same generic data are used, it is possible to transform the test…

Power computations for one-level experimental designs that assume simple random samples are greatly facilitated by power tables such as those presented in Cohen's book about statistical power analysis. However, in education and the social sciences experimental designs have naturally nested structures and multilevel models are needed to compute the…

When attempting to determine if an intervention has a causal impact, the "gold standard" of program evaluation is the randomized controlled trial (RCT). In education studies random assignment is rarely feasible at the student level, making RCTs harder to conduct. School-level assignment is more common but this often requires considerable resources…