Integrating five linear trend techniques into performance-criteria-based effect size measurements: Impressions and...

Area of Science:

• Behavioral science
• Quantitative psychology
• Research methodology

Background:

• Effect size measures are crucial for identifying evidence-based practices in single-case experimental designs (SCEDs).
• Existing nonoverlap-based effect size indices have limitations.
• The performance-criteria-based effect size (PCES) was introduced to quantify socially meaningful behavioral change, but its trend correction (PCES_trend) used a basic split-middle method.

Purpose of the Study:

• To enhance the accuracy and reliability of PCES_trend calculations.
• To compare advanced linear trend-fitting techniques for improving PCES_trend.
• To integrate improved PCES_trend into an accessible tool for applied researchers.

Main Methods:

• Analyzed 87 published SCED graphs with therapeutic baseline trends.
• Compared five advanced linear trend-fitting techniques against the original split-middle method.
• Evaluated techniques using metrics like mean absolute scaled error, coefficient of variation, R², visual inspection, and ancillary outcomes.

Main Results:

• Ordinary least squares (OLS) demonstrated superior accuracy and reliability in trend adjustments compared to other methods.
• OLS outperformed other techniques across various performance indicators and ancillary outcomes.
• The enhanced PCES_trend using OLS was integrated into a Google Sheet tool.

Conclusions:

• OLS is recommended as the preferred method for trend correction in PCES calculations.
• The developed Google Sheet tool enhances the accessibility and application of improved PCES measures.
• Accurate effect size measurement in SCEDs is vital for advancing evidence-based practices.