Scale Construction and Development - Online Course

First, we will cover current theory and best practices in scale construction. We will begin by discussing how to create an item pool. We will then review current research on:

• • The impact of vaguely worded or negatively worded items on scale reliability and validity.
  • The optimal length of a survey.
  • How many scale points to include.
  • Whether scale points should be labeled or unlabeled.
  • Whether to include a neutral option.
  • How item order effects may impact responses.

Next, we will cover the use of exploratory factor analysis (EFA) in the scale development and revision process. We will focus on basic EFA analysis and the interpretation of model parameters, with an emphasis on best practices in using EFA-based information to inform scale development. We will also discuss common issues in EFA, such as method factors, weak factors, and highly correlated factors. There will be several examples with real data sets, using SPSS, SAS, R (psych), and Mplus for the analysis. These include:

• • Factor analysis of dichotomously scored data.
  • Parallel analysis.
  • Sources of EFA model misfit.

Finally, we will introduce confirmatory factor analysis (CFA) and its use in scale development and the revision process. We will discuss estimation of CFA models and interpretation of CFA model parameters. There will be a particular emphasis on reasons for CFA model misfit, including issues with redundant and similarly worded items, cross-loading items, and method effects. We will also discuss the relation between model fit and the homogeneity of item intercorrelations. These methods will be illustrated with the Mplus and R (lavaan) programs. Sample code will be provided for analyses that include:

• • CFA models.
  • Bifactor models.
  • Calculation of coefficient omega.
  • Tests of item parallelism.

SPSS, SAS, and R will be used for scale construction examples. SPSS, SAS, R, and Mplus will be used for exploratory factor analysis (EFA) examples. R and Mplus will be used for all confirmatory factor analysis (CFA) examples. Prior knowledge of these programs is not essential. During the seminar, you are welcome to use a computer with any of these packages installed, although this is not required. Syntax and output for all examples, with comprehensive explanatory annotation, will be provided in the materials.

There is now a free version of SAS, called SAS OnDemand for Academics, that is available to anyone.

Those using R will also need the psych, lavaan, tidyverse or tidyr, dplyr, rlang, and e1071, packages, as well as readxl for reading in Excel files. The tidyverse package includes haven, which can be used to read in SPSS or SAS files.

If you’d like to use R for this course but don’t yet have much experience with that package, here are some excellent online resources for building your R skills.

If you’d like to familiarize yourself with Mplus basics before the seminar begins, we recommend reading through UCLA’s short guide here.

This seminar is designed for researchers interested in developing attitude, personality, opinion, or other noncognitive scales for use in research studies, needs assessments, program evaluations, or other purposes. You should be familiar with the basic principles of measurement theory, such as reliability and validity. You should also be familiar with basic statistics such as correlations, descriptive statistics, and introductory inferential statistics. No prior knowledge of EFA or CFA is required, although a basic knowledge of these methods will be helpful.

Day 1

Theoretical underpinnings of noncognitive scale development

• • Theories of response processes for noncognitive items
  • Developing noncognitive scales to optimize the response process

Practical issues in noncognitive scale development

• • Use of vague wording
  • Negatively keyed items
  • Number of scale points
  • Labeling vs not labeling response points
  • Inclusion of a neutral option
  • Item order effects

Day 2

Use of exploratory factor analysis (EFA) in scale development and revision

• • Basic EFA analyses and interpretation of model parameters
  • Use of EFA-based information in scale development and revision
  • Common problems in EFA and what to do about them
    • Obtaining more or fewer factors than hypothesized
    • Redundant and similarly worded items
    • Cross-loading items
    • Method effects

Day 3

Use of confirmatory factor analysis (CFA) in scale development and revision

• • Basic CFA analyses and interpretation of model parameters
  • Use of CFA-based information in scale development and revision
  • Reasons for CFA model misfit
    • Relation of model fit and homogeneity of item correlations (violations of proportionality constraints)
    • Redundant and similarly worded items
    • Cross-loading items
    • Method effects
  • Additional analyses
    • Bifactor models
    • Coefficient omega
    • Tests of parallelism

The fee of $995 USD includes all course materials.

PayPal and all major credit cards are accepted.

Our Tax ID number is 26-4576270.