Imagine two children with asthma—one whose first language is English, the other whose first language is Spanish. Each child is responding to the same health screening questionnaire (translated), but each child’s responses are based on differences in the way she/he perceives the item being measured—a perception often influenced by culture or even language nuance. Decisions related to treatment, health resource allocations, and interventions are often based on data generated from standardized questionnaires.
Is there a more accurate approach to evaluate questionnaires for their appropriate use for between-group comparisons? Karen Sousa, PhD, RN, FAAN, thinks there is.
Registered nurses and nurse researchers often use questionnaires to compare differences between groups, explains Sousa, associate dean for research and extramural affairs at the CU College of Nursing. But even a well developed instrument can be called into question if it has been developed using a homogenous sample.
“Over the years we’ve been studying measurement instruments that have been translated, and we’re finding that a lot of these instruments put the English speaking person in the best light,” Sousa notes.
While she was at Arizona State University, Sousa and her research team were contacted by the Phoenix Children’s Hospital Breathmobile to evaluate the English- and Spanish- language versions of the Pediatric Asthma Quality of Life Questionnaire for measurement equivalence. Sousa using a sample of 607 underserved children living with asthma, ages 6-18, who were given the Quality of Life questionnaire in their language of preference, evaluated the measurement equivalence in the English and Spanish versions using a set of hierarchical statistical tests.
“If a questionnaire assesses one thing in one group and something different in another group, comparing the two groups is like comparing apples to oranges,” Sousa wrote in the paper, which has been submitted to the journal Nursing Research.
“As we become more sophisticated in how we measure quality of life, we need to have a better way to measure group differences in a way that’s not biased.”
Analyzing instruments for measurement invariance to determine the appropriateness for using instruments to compare groups is new in nursing but has been used in other fields, such as psychology.
“When we evaluate the effectiveness of interventions in different groups, we need instruments that are conceptually equivalent within the groups of interest,” Sousa explains. “We want to draw conclusions as to how an intervention may affect each of the separate groups. Valid comparison of self-reported measures requires that the constructs have similar meaning across groups. If the instruments, in this case an English versus a Spanish version, are not equivalent, then group comparisons are highly suspect.”
Although the Phoenix sample of English and Spanish-speaking asthma patients focused on language variances and their implied biases, the research has implications for other cultural variances as well, such as age (children vs. teenagers; elderly vs. younger adults, for example) and gender.
An NIH Academic Research Enhancement Aware (AREA) funded the evaluation. The AREA grants are small-scale research projects in the biomedical and behavioral sciences conducted by faculty and students. These grants must include plans to expose students to hands-on research; Sousa’s students at Arizona State University and students at CU assisted with data gathering and entry.