Assessing the Quality of Survey Data by Jörg Blasius

By Jörg Blasius

This booklet introduces the newest tools for assessing the standard and validity of survey information via supplying new methods of reading edition and measuring mistakes. by way of essentially and accessibly demonstrating those innovations, specifically these derived from a number of Correspondence research, the authors enhance screening approaches to go looking for version in saw responses that don't correspond with genuine adjustments among respondents. utilizing recognized foreign info units, the authors convey tips to realize all demeanour of non-substantive edition from reaction types together with acquiescence, respondents' failure to appreciate questions, insufficient box paintings criteria, interview fatigue, or even the manufacture of (partly) faked interviews.

Show description

Read Online or Download Assessing the Quality of Survey Data PDF

Similar research books

Research Agendas in EU Studies: Stalking the Elephant (Palgrave Studies in European Union Politics)

Top students discover the advanced questions coming up from the continued transformation of Europe during the deepening and widening results of eu integration. in response to authoritative analyses, the ebook takes account of the numerous nationwide, transnational and overseas approaches and contexts during which ecu integration has develop into embedded.

Perspectives on Scientific Argumentation: Theory, Practice and Research

Argumentation—arriving at conclusions on a subject via a strategy of logical reasoning that comes with debate and persuasion— has in recent times emerged as a important subject of debate between technology educators and researchers. there's now a company and common trust that fostering argumentation in studying actions can strengthen scholars’ serious considering and reasoning talents, and that dialogic and collaborative inquiries are key precursors to an engagement in medical argumentation.

Ciba Foundation Symposium 134 - Research and the Ageing Population

This overseas symposium makes a speciality of the contributions that biomedical examine is making to the availability of right wellbeing and fitness deal with the expanding numbers of aged humans in all nations, built and constructing. contemporary, encouraging advancements within the components of malignancy, dementia, osteoporosis, and heart problems are emphasised.

Additional info for Assessing the Quality of Survey Data

Example text

More convincingly, Billiet and McClendon (2000) modelled two substantive concepts (political trust and attitudes towards immigrants), each of which was measured with a polarity-balanced set of items, with a common methods factor loading equally on all items of both concepts. They obtained a solid-fitting model, but only by permitting correlated error between the two positively worded political trust items. To assess whether the methods factor really tapped ARS, they created an observed measure of acquiescence (the sum of the agreements to the political trust and attitude towards immigrants items plus one additional balanced pair of items on another topic).

Because y −y xi − x and z2i = i sy sx z1i = and since n r12 = ∑ ( x i − x )( yi − y ) i =1 n n i =1 i =1 2 2 ∑ ( x i − x ) ∑ ( yi − y ) it follows that r12 = 1 T z1 z2 n or in matrix notation, R= 1 T 1 Z Z = A T FT FA = A T A n n The recognition that the correlation matrix can be reproduced by the product of its transposed factor loadings and the factor loadings is known as the fundamental theorem in factor analysis. indd 36 and R ≈ A Tm* A m* ASSESSING THE QUALITY OF SURVEY DATA 25/01/2012 2:29:28 PM Applying eigenvalue decomposition (also known as canonical decomposition) to the correlation matrix results in R = BLBT where B is the matrix of the eigenvectors and L the diagonal matrix of the eigenvalues, with the eigenvalues in descending order λ1 > λ2 > … > λm* >… > λm.

Converse (1976) classified over 300 items taken from several opinion polls on whether the referent was one on which respondents can be assumed to have direct personal experience or involvement. On such topics she found only trivial differences by educational attainment, while on more abstract issues the more educated were consistently more likely (three to seven percentage points) to provide a substantive response. , 2002). Judd, Krosnick and Milburn (1981) found significantly more measurement error among respondents with low education, suggesting that the task of placing themselves on various political orientation scales was more difficult for them.

Download PDF sample

Rated 4.50 of 5 – based on 26 votes