Measurement of attitudes and behaviors in public health surveys.

Abstract

IntroductionAttitudes and behaviors are salient factors in most of the current public health issues. Most studies measure attitudes and behaviors using a set ofitems included in a single closed-form questionnaire. Inasmuch as attitudes and behaviors toward the same act are examined, the wording of the items and the response formats should be nearly identical to each other.1-13 Feldman and Lynchl4 implied that a response to a question is likely to be retrieved as a basis for a subsequent response if they are presented consecutively in one questionnaire. Using similarly worded items to measure attitudes and behaviors compounds the problem. In the present paper we address the question of whether attitudes and behaviors can be measured accurately with a single questionnaire. MethodsWe randomly divided a sample of 900 general practitioners into two groups. One group (A) received a questionnaire in which both their attitudes and behaviors were assessed. The other group (B) received a questionnaire measuring behaviors; on receipt of that completed questionnaire, we mailed them a second one which pertained to their attitudes.The response rate for group A was 64.3 percent. The response rate for group B was 73.2 percent. Ninety-two percent of the physicians in group B who answered the first questionnaire also completed the second questionnaire.Twenty-one attitudinal items based on a series of questions (see Appendix) and 21 similarly worded behavioral items were presented to physicians on a sixpoint rating scale. They measured practices that physicians considered appropriate to perform (attitudes) and those which they claimed to perform (behaviors) in three clinical situations:* the extent of information that a physician should provide to chronically ill patients about their illness; * the extent to which a physician should promote non-smoking behavior to patients that consult for reasons other than smoking problems; * the extent to which a physician should discuss weight with an obese patient consulting for reasons other than obesity.The items measuring attitudes and behaviors were worded in the same manner.14 Systematic biases in measuring attitudes and behaviors were assessed by comparing the correlation coefficients between attitudes and behaviors in the two groups of physicians.The correlation coefficients were transformed into z-scores, using the Fisher Transformation, to test for the differencel1 between the association of attitudes and behaviors in the two groups of respondents. A total of 21 tests were performed for each of the comparisons between two groups. A P-plot procedure was used16 to estimate the number N of true null hypotheses. The estimate of N was then used to obtain an overall a-level which was less conservative than the usual Bonferonni level. The value of N was 6 in this study, thus the significant level was set at .05/6 = .008. ResmltsEight correlations between attitudes and behaviors were different in the two groups at the 0.008 level (Table 1). Pearson's correlation coefficients between attitu...

Explaining Delay in Care Seeking for Breast Cancer Symptoms1

“…Ideally, explanatory variables would be assessed before outcomes, but identifying women with symptoms before they contact the health system is difficult. If participants attempted to make their responses to explanatory and delay variables appear consistent, then observed relationships between these variables might be inflated (Beland, Maheux, & Lambert, 1991;Manfred0 & Shelby, 1988). However, study procedures minimized this possibility: Confidentiality was assured, data were collected by a nurse who was not the participants' practitioner, measures of psychosocial variables and delay were separated in the interview, and the study purpose was not stated explicitly.…”

Survey strategies discussed: using more than one questionnaire.

“…All 22 were White; 73% were male. The earliest reported use of GHB was April 1990, but most (64%) began using the drug after October 1,1990. Seventy-one percent described themselves as regular users; 95% used the dose recommended on the bottle, which was one teaspoon (2.5 g) on most bottles.…”