Figure 3 - uploaded by Andy Peytchev
Content may be subject to copyright.
BMI Reduction in Treatment and Control Groups under Different Treatment of Nonresponse.
Source publication
Over 3,000 subjects were recruited in 3 U.S. regions for a randomized experiment of an online weight management intervention. Participants were sent invitations to web survey reassessments after 3, 6, and 12 months. High and increasing nonresponse to the three follow- up surveys created the potential for nonresponse bias in key program outcomes. A...
Context in source publication
Context 1
... complete case estimates, presented in Figure 3, show that if we completely ignore nonresponse by neither launching a second phase nor adjusting for nonresponse covariates, BMI reduction is 1.3, or 72% larger in the treatment than the 0.76 in the control group (t-test, p<.05). This method assumes that the nonrespondents have the same weight loss pattern as the respondents. ...
Similar publications
Objective General practitioners (GP) are uniquely placed to guide their patients on nutrition and physical activity. The aims of the present study were to assess: (i) the extent to which GP guide on nutrition and physical activity; (ii) the determinants that cause GP to give guidance on nutrition and physical activity; and (iii) the extent to which...
This article examined the influence of urban homeowners’ general wildlife value orientations (i.e., mutualism, domination) and their specific positive beliefs about coyotes on perceived risk from coyotes. Data were obtained from a mail survey of residents in the Greater Chicago Metropolitan Region during 2012 (n = 1,624, response rate = 34%). As pr...
Criticisms of normative explanations of helping behavior are examined, and an explanation responsive to these criticisms is proposed. This explanation specifies conditions which affect the activation of personal norms and hence their influence on behavior. One hypothesis based on the explanation was tested: the impact of norms on behavior is a func...
This paper aims to examine the influence of innovation on competitive advantage in foods manufacturing SMEs in Malaysia and the moderating effects of firm age on innovation-competitive advantage relationship. Given the correlational nature of research, the researchers adopted a random sampling technique in Malaysian foods manufacturing SMEs. Mailed...
Citations
... An emerging issue of concern for Internet-based health behavior change interventions is high levels of attrition [1,[14][15][16]. Participants use Internet-based resources differently than they do other modalities such as group and in-person programs, and for Internet programs serving large numbers of participants that do not heavily screen users or conduct studies under efficacy conditions, it has proven difficult to obtain follow-up information on a high percentage of initial participants [1,[14][15][16]. ...
... An emerging issue of concern for Internet-based health behavior change interventions is high levels of attrition [1,[14][15][16]. Participants use Internet-based resources differently than they do other modalities such as group and in-person programs, and for Internet programs serving large numbers of participants that do not heavily screen users or conduct studies under efficacy conditions, it has proven difficult to obtain follow-up information on a high percentage of initial participants [1,[14][15][16]. At present, it is unclear what patient characteristics are related to active engagement with Internet program content and to participation in follow-up assessments and whether these characteristics are similar or different. ...
... Of the 1796 nonrespondents, 913 were sent a mail survey; of these, 586 returned the completed survey, for a 64% response rate to the mail follow-up. (This compares to 56% for a mailed follow-up and 59% for a telephone follow-up in the earlier evaluation of the Balance program [15].) This follow-up effort boosted the unweighted 12-month response rate from 22.3% to 47.6% and the weighted rate to 72.7%. ...
Research increasingly supports the conclusion that well-designed programs delivered over the Internet can produce significant weight loss compared to randomized controlled conditions. Much less is known about four important issues addressed in this study: (1) which recruitment methods produce higher eHealth participation rates, (2) which patient characteristics are related to enrollment, (3) which characteristics are related to level of user engagement in the program, and (4) which characteristics are related to continued participation in project assessments.
We recruited overweight members of three health maintenance organizations (HMOs) to participate in an entirely Internet-mediated weight loss program developed by HealthMedia, Inc. Two different recruitment methods were used: personal letters from prevention directors in each HMO, and general notices in member newsletters. The personal letters were sent to members diagnosed with diabetes or heart disease and, in one HMO, to a general membership sample in a particular geographic location. Data were collected in the context of a 2x2 randomized controlled trial, with participants assigned to receive or not receive a goal setting intervention and a nutrition education intervention in addition to the basic program.
A total of 2311 members enrolled. Bivariate analyses on aggregate data revealed that personalized mailings produced higher enrollment rates than member newsletters and that members with diabetes or heart disease were more likely to enroll than those without these diagnoses. In addition, males, those over age 60, smokers, and those estimated to have higher medical expenses were less likely to enroll (all P < .001). Males and those in the combined intervention were less likely to engage initially, or to continue to be engaged with their Web program, than other participants. In terms of retention, multiple logistic regressions revealed that enrollees under age 60 (P < .001) and those with higher baseline self-efficacy were less likely to participate in the 12-month follow-up (P = .03), but with these exceptions, those participating were very similar to those not participating in the follow-up.
A single personalized mailing increases enrollment in Internet-based weight loss. eHealth programs offer great potential for recruiting large numbers of participants, but they may not reach those at highest risk. Patient characteristics related to each of these important factors may be different, and more comprehensive analyses of determinants of enrollment, engagement, and retention in eHealth programs are needed.
We consider data with a continuous outcome that is missing at random and a fully observed set of covariates. We compare by simulation a variety of doubly-robust (DR) estimators for estimating the mean of the outcome. An estimator is DR if it is consistent when either the regression model for the mean function or the propensity to respond is correctly specified. Performance of different methods is compared in terms of root mean squared error of the estimates and width and coverage of confidence intervals or posterior credibility intervals in repeated samples. Overall, the DR methods tended to yield better inference than the incorrect model when either the propensity or mean model is correctly specified, but were less successful for small sample sizes, where the asymptotic DR property is less consequential. Two methods tended to outperform the other DR methods: penalized spline of propensity prediction [Little RJA, An H. Robust likelihood-based analysis of multivariate data with missing values. Statist Sinica. 2004;14:949-968] and the robust method proposed in [Cao W, Tsiatis AA, Davidian M. Improving efficiency and robustness of the doubly robust estimator for a population mean with incomplete data. Biometrika. 2009;96:723-734].
Little and An (2004, Statistica Sinica 14, 949-968) proposed a penalized spline of propensity prediction (PSPP) method of imputation of missing values that yields robust model-based inference under the missing at random assumption. The propensity score for a missing variable is estimated and a regression model is fitted that includes the spline of the estimated logit propensity score as a covariate. The predicted unconditional mean of the missing variable has a double robustness (DR) property under misspecification of the imputation model. We show that a simplified version of PSPP, which does not center other regressors prior to including them in the prediction model, also has the DR property. We also propose two extensions of PSPP, namely, stratified PSPP and bivariate PSPP, that extend the DR property to inferences about conditional means. These extended PSPP methods are compared with the PSPP method and simple alternatives in a simulation study and applied to an online weight loss study conducted by Kaiser Permanente.
