Quantifying the presence of evidential value and selective reporting in food-related inhibitory control training: a p-curve analysis. Issue 3 (3rd July 2019)

• Record Type: Journal Article
• Authors: Carbine, Kaylie A.; Larson, Michael J.
• Is Part Of: Health psychology review. Volume 13:Issue 3(2019)
• Page Start: 318
• Page End: 343
• Publication Date: 2019-07-03
• Subjects: P-curve -- selective reporting -- evidential value -- inhibitory control training -- food; Clinical health psychology -- Periodicals; 613.019
• DOI: 10.1080/17437199.2019.1622144
• Languages: English
• ISSNs: 1743-7199

Abstract

ABSTRACT: Meta-analyses suggest inhibitory control training (ICT) may be effective for reducing food intake. However, psychological research has come under scrutiny for lack of reliability. Selective reporting (only reporting significant results) is one factor contributing to the lack of reliability in published research. Therefore, estimates of food-related ICT effects may be inaccurate. We conducted p -curve analyses to assess the presence of selective reporting, evidential value, average effect size, and power in the food-related ICT literature. Extracted p -values were selected from articles included in food-related ICT meta-analyses and an updated literature search. Four p -curve analyses resulted in 'U'-shaped distributions, suggesting evidence for both a true underlying effect and selective reporting in the food-related ICT literature. Robust analyses suggested the evidence for an underlying effect was primarily driven by the smallest p -value. The average effect size from included studies was small ( d = 0.04 to 0.25). Average power to detect this effect was also small (7% to 18%). Results suggest no clear support for or against a true effect for food-related ICT. Low average effect size and power across studies suggests estimated effects are likely inflated in published literature. Higher-powered, pre-registered, longitudinal food-related ICT studies are needed to test for the presence and magnitude of ICT effects.