Search

Reduction in nonresponse bias has been a key focus in responsive and adaptive survey designs, through multiple phases of data collection, each defined by a different protocol, and targeting interventions to a subset of sample elements. Key in this approach is the identification of nonrespondents who, if interviewed, can reduce nonresponse bias in survey estimates. From a design perspective, we need to identify an appropriate model to select targeted cases, in addition to an effective intervention (change in protocol). From an evaluation perspective, we need to compare estimates to a control condition that is often omitted from study designs, in addition to the need for benchmark estimates for key survey measures to provide estimates of nonresponse bias. We introduced a bias propensity approach for the selection of sample members to reduce nonresponse bias. Unlike a response propensity approach in which the objective is to maximize the prediction of nonresponse, this new approach deliberately excludes strong predictors of nonresponse that are uncorrelated with survey measures and uses covariates that are of substantive interest to the study. We also devised an analytic approach to simulate which sample members would have responded in a control condition. This study also provided a rare opportunity to estimate nonresponse bias, using rich sampling frame information, prior round survey data, and data from extensive nonresponse follow-up. The bias propensity model yielded reasonable fit despite the exclusion of the strongest predictors of nonresponse. The intervention was found to be effective in increasing participation among identified sample members. On average, the responsive and adaptive survey design reduced nonresponse bias by more than one-quarter—almost one percentage point—regardless of the choice of benchmark estimates. Effort under the control condition did not reduce nonresponse bias. While results are strongly encouraging, we argue for replication with varied populations and methods.

Response rates in household surveys are declining, increasing the risk of nonresponse bias in survey estimates. Survey costs are increasing. As a result, design features such as higher monetary incentives are needed but often cannot be afforded. Two or more survey protocols could be implemented in parallel, where some have lower nonresponse while others have lower cost, as long as the data can be combined in a way that reflects the reduced potential for nonresponse bias under the more intensive protocol. We describe the main barrier to the use of such an approach—that traditional methods ignore the expected lower bias in one condition. The proposed approach includes random assignment of sample members to a data collection protocol and adjustment of survey estimates to the superior protocol, based on key survey variables—if differences in estimates are found. This represents a major departure from the current practice in constructing nonresponse adjustments and leverages the use of the same sampling design, survey instrument, and measurement procedures in each condition. An illustrative example is presented using data from a national survey. Methods to address both bias reduction and variance estimation are described. We end with limitations and suggestions for future research.

Survey length and monetary incentives are design features that have been found to affect survey participation in opposing ways. Declining response rates make these key design features even more important. However, survey length has received surprisingly little attention, possibly being viewed as a static survey attribute rather than an alterable design feature. In addition, experimental tests of the interaction between survey length and monetary incentives could not be found, despite calls for the need for such research and theories that support such an interaction.Leverage-salience theory leads to opposing expectations, depending on the relative importance people give to different factors affecting survey participation. Making the survey request in parts, to reduce burden, may increase participation. Yet, it may instead decrease participation relative to the status quoof offering the full incentive for the entire survey if incentives play a larger role than survey length. When factoring both burden and motivation, the results for survey length could likely depend on the structure of monetary incentives. We designed an experiment to test a modularized survey, ending one part of the survey with a request to complete another part. Respondents in the modular survey design were also assigned to proportionate or disproportionate incentives—offering a larger proportion of the incentive for the initial survey request. Surprisingly, the two-part survey request led to not lower but greater nonresponse. There was increased potential for nonresponse bias attributable to nonresponse to the second request, without reduction in item nonresponse and measurement error. Disproportionate allocation of incentives across modules seemed to increase overall participation compared with proportionate allocation. The lack of a beneficial effect of a request for a shorter survey and the results on incentive allocation have important implications for survey practitioners. Furthermore, the inconsistency in findings on survey length requires further research to identify mediating factors.

This article presents results from an experimental study in Germany designed to test the effectiveness of a novel protocol for matching participants in a national panel survey with interviewers employing computer-assisted telephone interviewing (CATI) on selected sociodemographic features, including sex, age, and education. We specifically focus on the ability of the protocol to engender close matches between respondents and interviewers in terms of these features, using both theory and empirical evidence to suggest that this type of matching will improve cooperation rates in surveys employing CATI. We also focus on indicators of “success” at first contact (defined as a successful interview or establishment of an appointment for an interview) as a function of whether the matching protocol was in use on a given day and whether specific types of matches generated higher rates of success overall. We find strong evidence of the protocol effectively establishing close matches, and we also observe that matches based on education proved especially effective for rates of “success” in a panel survey that focused primarily on labor market topics. We conclude with thoughts on practical implementation of this approach in other settings and suggested directions for future work in this area.

Surveys routinely offer incentives to motivate respondents and increase the likelihood of their participation, yet surprisingly little is known about the effectiveness of interviewer incentives. If interviewer incentives increase interviewers’ success in gaining cooperation, they could help address declining survey response rates. In this article, we present the results of an experiment testing the effectiveness of interviewer incentives in the form of cash bonuses for each successfully completed field interview. We did not find evidence that higher payments to interviewers for each completion led to increased effort on the part of interviewers nor did they lead to higher levels of success in securing respondent cooperation. These findings suggest that per complete interviewer incentives may not be cost effective in reducing survey nonresponse.

Rapidly decreasing coverage of landline surveys is increasing the need to implement dual-frame surveys for inference to the adult U.S. population. Vast differences between the way cell phones and landlines are used, and the populations using them, require separate data collection designs. Yet research comparing cell phone surveys to landline telephone surveys is scarce with respect to operational outcomes. The authors test hypothesized differences between cell phone and landline interviewing through experiments on survey topic and length and find that these factors may not have the same impact in cell phone surveys. To help optimize calling cell phone numbers in future studies, the authors present self-reported cell phone use patterns and other factors affecting the probability of contact and sampling design.To inform the inclusion of adults with both a cell and landline phone, they compare cell phone use among cell phone only and cell with landline cases. The authors found notable differences between the cell only and cell with landline respondents in terms of cell phone use. Implications and directions for future work are discussed.

The paper discusses the principles and the algorithm of granular analysis of data in a specific context of urban traffic monitoring and control (EIS). The proposed granular information processing enables extraction of information on the pattern of journeys from the detailed traffic counts. This facilitates progression from the local optimisation of traffic on individual crossroads to the more holistic optimisation of traffic in a road network. The proposed EIS makes use of readily available stop-line queue data, which is used for adaptive tuning of traffic signals, and adds a data processing layer referred to as granular analysis. It is argued that granular analysis is preferred to statistical data processing since it does not require any assumptions about statistical characterisation of traffic. The granulation algorithm has two distinctive features: (1) the information granules are formed by means of hierarchical optimisation of information density, and (2) the granules are created as hyperboxes thus being readily interpretable in the pattern space. The granular estimates of turning movements are calibrated using an HUTSIM micro-simulator.

Response rates in household surveys have been declining, which increases the risks of bias in the survey results. Nonresponse weight adjustments can reduce these risks to some extent; however, for many studies—particularly in random digit dial (RDD) telephone surveys—data needed to perform the adjustments are scant. A source of data that is being used with increasing frequency by statisticians is census geocoded (CG) data where block group- or tract-level census data are appended to each sample case for use in the nonresponse adjustment process. Though this method may have some merit, its effectiveness for reducing bias remains unknown. This article reports results of an evaluation of its effectiveness for general household surveys and investigates the conditions that favor its use in the nonresponse adjustment process. Using a unique data set from a national household survey augmented by CG data, we empirically evaluate: (1) the effectiveness of CG data for removing bias in nonresponse adjustments, (2) the effect of using aggregate data and effect of using matching by telephone number on the ability to adjust for nonresponse bias, (3) any potentially adverse effects on the variances of survey estimates and trade-off with bias reduction, and (4) the conditions that enhance or limit its effectiveness. We conclude that aggregate census data have very limited utility in nonresponse adjustment unless (a) the census characteristics used in the adjustment are highly geographically clustered and (b) outcome variables exhibit good correlation with these census characteristics. We provide specific recommendations regarding (a) and (b).

Imatinib 400 mg/day represents the current standard treatment for de novo as well as pre-treated CML patients in chronic phase (CP). Recent randomized phase III trials revealed conflicting results concerning the potential higher efficacy of dose-increased imatinib in de novo treated CP-CML.We here present the final analyses including response data, OS, EFS and PFS of the multicenter, randomised, 2-arm phase III CELSG “ISTAHIT” trial evaluating imatinib high dose (HD) induction (800 mg/day, 6 months) followed by 400 mg/day as maintenance (experimental arm B) compared to continuous imatinib standard dose (400mg/day; arm A) in pre-treated CP CML patients. ClinicalTrials.gov Identifier: NCT0032726.From a total of 243 patients screened for inclusion, 16 patients were not eligible (mainly due to non sufficient numbers of metaphases obtainable from the bone marrow before the start of the study). Of the remaining 227 patients, 113 patients were randomized into arm A and 114 patients into the experimental arm B. Subsequent data are presented as per protocol. No significant differences between treatment groups were observed regarding sex (55.5% female, 44.5% male), age (median: 46.3 years, range 18 –76), Sokal scores at diagnosis (30% low, 41% intermediate, 16% Sokal high risk, 13% unknown) and different pre-treatments, which included hydroxyurea (96%), interferon (72%), busulfan (17%) and “others” (26%; mainly Ara-C). The median observation time was 673 days. Cytogenetic responses were generally higher in the experimental arm B and revealed statistically significant differences in major cytogenetic responses (MCyR) at 3 and 6 months (month 3: 25.8% arm A, 48.3% arm B, p=0.002; month 6: 41.9% arm A, 58.8% arm B, p=0.029) as well as in complete cytogenetic responses (CCyR) not only during imatinib HD therapy (month 3: 7.5% arm A, 29.9% arm B, p<0.001; month 6: 20.4% arm A, 47.4% arm B, p<0.001) but also thereafter (month 12: 31.8% arm A, 52.9% arm B, p=0.006). The primary endpoint of the study, the achievement of an improved MCyR at 12 month was, however, not significantly different (56.8% arm A, 64.4% arm B). In line with improved cytogenetic responses, major molecular response (MMRIS) rates were also significantly better at 3, 6 and even at 24 months in the HD arm B (month 3: 3.7% arm A, 15.9% arm B, p=0.003; month 6: 9.4% arm A, 34.6% arm B, p<0.001; month 24: 26.5% arm A, 42.5% arm B, p=0.034). Surprisingly, however, this impressing improvement in cytogenetic and molecular remissions in patients achieving high dose imatinib as induction therapy did not translate into a better OS and PFS, both of which were comparable in the two treatment arms (OS: p=0.25; EFS: p=0.37). Moreover, the EFS was even significantly worsened in the experimental arm B (p=0.014). Grade 3/4 non-haematological toxicities during the first 6 months of therapy were comparable, whereas grade 3/4 haematological toxicities were significantly more common in the imatinib HD arm B.Although high dose imatinib induction induces more rapid and higher cytogenetic and molecular remission rates in pre-treated CP CML patients, OS as well as PFS were not improved and EFS was even worsened in the high dose induction arm B. Therefore we conclude that imatinib 400mg/day remains the standard of care for pre-treated CP-CML patients.Petzer: Novartis: Honoraria, Membership on an entity's Board of Directors or advisory committees, Research Funding. Off Label Use: Imatinib 800mg is not licensed as the initial therapy of chronic phase CML. Lion: Novartis: Honoraria, Research Funding. Bogdanovic: Novartis: Consultancy, Honoraria, Membership on an entity's Board of Directors or advisory committees. Griskevicius: Novartis: Research Funding. Kwakkelstein: Celgene: Employment. Rancati: Novartis: Consultancy, Employment, Equity Ownership, Honoraria, Membership on an entity's Board of Directors or advisory committees, Research Funding, Speakers Bureau. Gastl: Novartis: Honoraria, Membership on an entity's Board of Directors or advisory committees, Research Funding. Wolf: Novartis: Honoraria, Membership on an entity's Board of Directors or advisory committees.

Abstract 2271

We have recently reported results from the first planned interim analysis of a multicenter, randomised, 2-arm - phase III study comparing imatinib standard dose (400 mg/day; arm A) with imatinib high dose (HD) induction (800 mg/day; HD arm B) followed by imatinib standard dose maintenance (400 mg/day) in pretreated Ph+/BCRABL+ CML patients in chronic phase (CP; Blood,110,1048a). The first planned interim analysis was performed after 50% of the patients had been treated for 12 months (mo) since randomisation. A total of 227 patients were randomized. Pretreatments included hydroxyurea (96%), interferon (72%), busulfan (17%) and “others” (26%; mainly AraC +/− combinations). Although rates of complete haematological responses did not differ significantly between the 2 arms at 3, 6 and 12 mo, significantly more patients achieved a major (MCR) and a complete cytogenetic response (CCR) at 3 mo (MCR: 21% arm A, 37% arm B, p=0.01; CCR: 6% arm A, 25% arm B, p<0.001) and 6 mo (MCR: 34% arm A, 54% arm B, p=0.009; CCR: 20% arm A, 44% arm B, p<0.001). At 12 mo, following dose reduction of imatinib to 400 mg/d for “maintenance” at mo 6 in the HD arm B, the rates of MCR (the primary endpoint of the study) were comparabel (59% arm A, 57% arm B). Nevertheless, there was still a clear trend to higher rates of CCR (37% arm A, 48% arm B) in the HD arm B, but at the time of the interim analysis these values did not reach statistical significance. In contrast to non-hematological toxicities, grade 3/4 hematological toxicities were significantly more common in the HD arm B (anemia: 2% arm A, 14% arm B; leukopenia: 24% arm A, 46% arm B; thrombocytopenia: 15% arm A, 39% arm B; p≤0.02). Updated results reveal, that the cumulative median doses of imatinib were 400 mg (arm A) and 767 mg (arm B), respectively. The median days of dose interruptions during the first 6 mo were not significantly different (12.5 days arm A, 13 days arm B; p=0.78). Nevertheless, significantly more patients (65.1%) remained on the initial 400mg dose of imatinib in arm A as compared to the initial 800mg dose in the HD arm B (45.6%; p=0.009). 35% of the patients that had to be dose reduced in arm B were reduced to 600mg/day, 63% to 400mg and 2% to 300mg. The cumulative rates of MCR in patients with no dose reduction/interruption were higher in the HD arm B (71%) compared to arm A (59%, p=0.232). Moreover, the cumulative rates of CCR were significantly better for patients with no dose reduction/interruption in the HD arm B (61%) compared to the patients with no dose reduction/interruption in arm A (36%, p=0.014). In addition, in the HD arm B the cumulative rates of CCR were significantly better for patients with no dose reduction/interruption (61%) compared to patients in the HD arm B with an interruption and/or dose reduction (34.9%; p=0.017). Conclusions: These updated phase III data not only support the concept of more rapid and higher rates of cytogenetic remissions (MCR, CCR) when higher doses of imatinib are applied, they also suggested that patients that are capable to tolerate continous high dose imatinib may have a better outcome.

We have recently reported results from the first planned interim analysis of a multicenter, randomised, 2-arm - phase III study comparing imatinib standard dose (400 mg/day; arm A) with imatinib high dose (HD) induction (800 mg/day; HD arm B) followed by imatinib standard dose maintenance (400 mg/day) in pretreatedPh+/BCRABL+CML patients in chronic phase (CP; Blood,110,1048a). The first planned interim analysis was performed after 50% of the patients had been treated for 12 months (mo) since randomisation. A total of 227 patients were randomized. Pretreatments included hydroxyurea (96%), interferon (72%), busulfan (17%) and “others” (26%; mainly AraC +/− combinations). Although rates of complete haematological responses did not differ significantly between the 2 arms at 3, 6 and 12 mo, significantly more patients achieved a major (MCR) and a complete cytogenetic response (CCR) at 3 mo (MCR: 21% arm A, 37% arm B, p=0.01; CCR: 6% arm A, 25% arm B, p<0.001) and 6 mo (MCR: 34% arm A, 54% arm B, p=0.009; CCR: 20% arm A, 44% arm B, p<0.001). At 12 mo, following dose reduction of imatinib to 400 mg/d for “maintenance” at mo 6 in the HD arm B, the rates of MCR (the primary endpoint of the study) were comparabel (59% arm A, 57% arm B). Nevertheless, there was still a clear trend to higher rates of CCR (37% arm A, 48% arm B) in the HD arm B, but at the time of the interim analysis these values did not reach statistical significance. In contrast to non-hematological toxicities, grade 3/4 hematological toxicities were significantly more common in the HD arm B (anemia: 2% arm A, 14% arm B; leukopenia: 24% arm A, 46% arm B; thrombocytopenia: 15% arm A, 39% arm B; p≤0.02). Updated results reveal, that the cumulative median doses of imatinib were 400 mg (arm A) and 767 mg (arm B), respectively. The median days of dose interruptions during the first 6 mo were not significantly different (12.5 days arm A, 13 days arm B; p=0.78). Nevertheless, significantly more patients (65.1%) remained on the initial 400mg dose of imatinib in arm A as compared to the initial 800mg dose in the HD arm B (45.6%; p=0.009). 35% of the patients that had to be dose reduced in arm B were reduced to 600mg/day, 63% to 400mg and 2% to 300mg. The cumulative rates of MCR in patients with no dose reduction/interruption were higher in the HD arm B (71%) compared to arm A (59%, p=0.232). Moreover, the cumulative rates of CCR were significantly better for patients with no dose reduction/interruption in the HD arm B (61%) compared to the patients with no dose reduction/interruption in arm A (36%, p=0.014). In addition, in the HD arm B the cumulative rates of CCR were significantly better for patients with no dose reduction/interruption (61%) compared to patients in the HD arm B with an interruption and/or dose reduction (34.9%; p=0.017).

227 patients with pretreated Ph+/BCR-ABL+ CML were randomized into this 2-arm phase III trial comparing standard dose Imatinib (400 mg/d; arm A) with high dose (HD) Imatinib (800 mg/d for 6 months) for induction followed by Imatinib 400 mg/d for maintenance (400 mg/d; arm B). The interim analysis presented here was performed on all patients after 50% of the patients had been treated for 12 months since randomisation. There were no significant differences between treatment arms regarding sex, age and different pretreatments. Rates of complete hematological responses did not differ significantly between both groups at 3, 6 and 12 months (82% arm A, 90% arm B). However, significantly* more patients achieved a major (MCR) and a complete cytogenetic response (CCR) at 3 months (MCR: 21% arm A, 37% arm B; CCR: 6% arm A, 25% arm B) and 6 months (MCR: 34% arm A, 54% arm B; CCR: 20% arm A, 44% arm B). Moreover, significantly* more patients achieved a major molecular response (MMR) at 6 months in the Imatinib HD arm B compared to arm A (20% vs 7%). At 12 months, following dose reduction of Imatinib to 400 mg/d for maintenance at month 6 in the HD arm B, the rates of MCR (the primary endpoint of the study) were comparable (57% arm A, 59% arm B). Nevertheless, there was still a clear trend to higher rates of CCR (37% arm A, 48% arm B) and MMR (16% arm A, 21% arm B) in the HD arm, but at the time of the interim analysis these values did not reach statistical significance. In contrast to non-hematological toxicities, grade 3/4 hematological toxicities were significantly* more common in the HD arm B [anemia: 2% (A), 12% (B); leukopenia: 23% (A), 41% (B), thrombopenia: 14% (A), 34% (B)]. In conclusion, this is the first randomized phase III trial demonstrating significantly* higher rates of MCR, CCR and MMR in chronic phase CML during therapy with HD Imatinib. *p<0.05

227 patients with pretreated Ph+/BCR-ABL+CML were randomized into this 2-arm phase III trial comparing standard dose Imatinib (400 mg/d; arm A) with high dose (HD) Imatinib (800 mg/d for 6 months) for induction followed by Imatinib 400 mg/d for maintenance (400 mg/d; arm B). The interim analysis presented here was performed on all patients after 50% of the patients had been treated for 12 months since randomisation. There were no significant differences between treatment arms regarding sex, age and different pretreatments. Rates of complete hematological responses did not differ significantly between both groups at 3, 6 and 12 months (82% arm A, 90% arm B). However, significantly* more patients achieved a major (MCR) and a complete cytogenetic response (CCR) at 3 months (MCR: 21% arm A, 37% arm B; CCR: 6% arm A, 25% arm B) and 6 months (MCR: 34% arm A, 54% arm B; CCR: 20% arm A, 44% arm B). Moreover, significantly* more patients achieved a major molecular response (MMR) at 6 months in the Imatinib HD arm B compared to arm A (20% vs 7%). At 12 months, following dose reduction of Imatinib to 400 mg/d for maintenance at month 6 in the HD arm B, the rates of MCR (the primary endpoint of the study) were comparable (57% arm A, 59% arm B). Nevertheless, there was still a clear trend to higher rates of CCR (37% arm A, 48% arm B) and MMR (16% arm A, 21% arm B) in the HD arm, but at the time of the interim analysis these values did not reach statistical significance. In contrast to non-hematological toxicities, grade 3/4 hematological toxicities were significantly* more common in the HD arm B [anemia: 2% (A), 12% (B); leukopenia: 23% (A), 41% (B), thrombopenia: 14% (A), 34% (B)]. In conclusion, this is the first randomized phase III trial demonstrating significantly* higher rates of MCR, CCR and MMR in chronic phase CML during therapy with HD Imatinib. *p<0.05