nonresponse


Also found in: Dictionary, Wikipedia.

nonresponse

a problem in the social sciences caused by people not completing QUESTIONNAIRES, refusing to be interviewed, etc. Nonresponse is a common problem in sociological research, especially in those instances where a postal questionnaire is being used, where a RESPONSE RATE of above 50% is generally considered to be good.

In order to discuss ways of increasing the response rate it is necessary to consider briefly why people do not return questionnaires. Some people have died or moved before the questionnaires are sent out, whilst others simply refuse to complete questionnaires. Apart from people who have moved house, died, etc., probably the main reason why questionnaires are not returned is due to forgetfulness and inertia. This problem can be minimized in a number of ways. In constructing the questionnaire care should be taken in writing and ordering the questions to make it as simple and easy to complete as possible. The covering letter should be written so that the purposes of the study are explained and assurances should be given concerning the confidentiality of the data. A few days after sending the questionnaires a reminder card should be distributed, and second and even third copies of the questionnaire may be sent to nonrespondents, depending on finances. A shortened version of the questionnaire, asking key questions only, may be sent, or this may be administered by telephone.

A number of techniques exist which attempt to calculate the extent to which nonrespondents might differ from respondents. Depending on the data source, that data which is known about nonrespondents can be coded (see CODING) for analysis and comparison with respondents. For example, a study of young people using addresses obtained from careers offices may be able to build up a picture of nonrespondents which includes data on gender, age, examination results, employment status, etc. Also, replies can be weighted on the basis that respondents who reply to the final reminder have more in common with those from whom no response has been obtained than with those who replied earlier. It may also be possible to estimate some characteristics of the total population by using alternative data sources (e.g. other questionnaire studies, CENSUS material, etc.) and then the data can be weighted accordingly.

References in periodicals archive ?
To validate the relevance between TAZ expression and treatment response in NSIP, we also compared TAZ expression in fibroblasts, bronchiolar cells, and alveolar cells between the good response and nonresponse groups.
The specific number of respondents varies from question to question, owing to item nonresponse and to the fact that, in the cases of several items, we randomly divided the sample into multiple groups in order to examine the effect of variations in the way questions were posed.
Among patients who received antibiotic treatment escalation within 2 days of starting treatment, most (90%) had nonresponse on treatment day 1, but only 5% had inappropriate initial therapy.
Teacher samples are weighted to correct for unequal selection probability and nonresponse.
Respondents with nonresponses were included in the denominators when calculating prevalence estimates.
Previous studies have used comparisons between early and late responders to look at nonresponse bias (Dooley & Lindner, 2003; Jordan, Walker, Kent, & Inoue, 2011), a method based upon the assumption that late respondents are most similar to nonrespondents (Armstrong & Overton, 1977; Miller & Smith, 1983).
In the case of item nonresponse for sendout, the residential and commercial sectors default to the reported
These include failure to use sample frames that provide adequate coverage of the population being studied (coverage errors), imprecision in the sample(s) drawn from a sample frame (sampling error), errors associated with failure to contact or complete interviews with all sampled respondents, and failure to obtain answers to all questions included in a survey instrument (nonresponse errors), as well as failure to make adequate adjustments for complex sample designs and survey nonresponse (adjustment errors).
2) Bias in estimates can result from measurement error, processing/editing error, coverage error, and nonresponse error (Federal Committee on Statistical Methodology [FCSM] 2001).
In randomized control trials (RCTs) of educational interventions, there is a growing literature on impact estimation methods to adjust for missing student outcome data using such methods as multiple imputation, the construction of nonresponse weights, casewise deletion, and maximum likelihood methods (see, for example, Allison, 2002; Graham, 2009; Peugh & Enders, 2004; Puma, Olsen, Bell & Price, 2009; Schafer & Graham, 2002).
Thus, it would be premature to repeat renal denervation or switch to alternative therapy because of blood pressure nonresponse at 6 months, Dr.
In the previous example, readers would have a better understanding of the type of missing data and the rationale that went into addressing the missing values if the researcher reported overall nonresponse rates both for the study and for each of the three variables.