Self‐reported surveys are used to assess dietary and physical activity (PA) behavior changes among low‐income Expanded Food and Nutrition Education Program (EFNEP) adult participants. A multi‐state research project (NC2169) has been established to create a new EFNEP survey that is consistent with America's most recent dietary and physical activity guidelines. The aim of this portion of the project's work was to examine the test‐retest reliability for the survey's PA questions, for which face validity had previously been verified via cognitive interviews. Due to pragmatic concerns (i.e. literacy levels, participant burden and time constraints), questions used for EFNEP evaluation need to be relatively quick and easy to complete. Thus, a single item was used to address each of three PA concepts: Item 1: moderate to vigorous aerobic PA; Item 2: muscle strengthening/resistance training; and Item 3: bodily exertion to intentionally develop and maintain fitness via everyday activities (e.g., taking the stairs instead of the elevator, parking far from a destination to walk more, etc.). Items 1 and 2 were measured on a 7‐point scale, (“0” to “7” days); Item 3 was assessed using a 5‐point Likert‐type scale (“Never” to “Always”). Sixty‐eight low‐income women from Colorado, New Jersey, and Nevada were asked to respond to the questions on two separate occasions, 7–10 days apart. Care was taken to administer questions during periods for which consistent weather patterns had been predicted, as inconsistent weather has been correlated with changes in individuals' activity patterns. Spearman's correlation coefficient, intra‐class correlation coefficient (ICC) and paired t‐test were used to examine test‐retest reliability using SAS (version 9.3). Demographic information was also collected, and analyzed using descriptive statistics (frequencies, means and standard deviations). Of the 68 participant recruited, 64 completed both the test and retest; most were Black (56.3%), had a high school degree or lower (51.5%), and received SNAP benefit (68.0%). Neither test nor retest responses differed significantly by age, race or education levels. Spearman correlation coefficients ranged from 0.60 (Item 3) to 0.88 (Item 2), suggesting a “strong” (0.60–0.79) to “very strong” (0.80–1.00) correlation between the test and retest responses. ICC for Items 1–3 were 0.71, 0.77, 0.64, respectively, which also indicated “good” reliability (ICC= 0.61 to 0.80). Paired t‐tests showed no significant differences between test and retest responses, which, in the absence of external influences, reflected their reliability. In conclusion, these PA questions demonstrated satisfactory test‐retest reliability for assessing low‐income women's physical activity levels. Research to examine the questions' validity remains necessary.Support or Funding InformationFunding resource: United States Department of Agriculture, New Jersey Agricultural Experiment Station Grant Number: NJ 14211; (Multi‐State Project NC‐2169)
Read full abstract