Abstract
Purpose: :
Conjoint analysis is a technique for eliciting preferences regarding the relative importance of different attributes of choices. The purpose of this study was to validate a method of conducting conjoint analysis among patients with visual impairment using a computer to display colored-bars to indicate the levels of the different attributes of the choice.
Methods: :
Fifty-three patients with visual impairment presenting at a low vision clinic were asked to indicate which of two hypothetical rehabilitation regimens and results they would prefer. Attributes of the regimens included difficulty with reading, transportation, and television viewing, the number of rehabilitation sessions, and out-of-pocket expenses. Each attribute had three levels. An interviewer read a description of the levels of each attribute for each choice. Respondents were also shown a screen with two columns; each contained five colored bars. Green bars represented the most favorable level (lowest cost, fewest visits, least difficulty). Red bars represented the least favorable level. Yellow bars represented the intermediate category. Each respondent was asked to indicate his or her preference for 12 choices. Data were analyzed using a random effects conditional logit in which the characteristics of the choices (levels of attributes) were the explanatory variables.
Results: :
The coefficients on all variables in the regression analyses suggested that less favorable categories were preferred less, although not all differences were statistically significant. All coefficients for variables indicating the least favorable categories were statistically significantly different from zero (p<0.001 in all cases). For reading (p=0.004) and out-of-pocket expenses (p=0.002), the estimated coefficient for the variable indicating the moderately favorable category was statistically significantly different from zero. The results suggest that patients would pay $1514 to improve from somewhat dissatisfied with reading to reading satisfactorily.
Conclusions: :
Administering a conjoint analysis questionnaire to low vision patients assisted by a visual display using colors to indicate levels of the attributes of the choices yields valid results and should be considered for preference elicitation in this population in the future.
Keywords: low vision • clinical (human) or epidemiologic studies: health care delivery/economics/manpower