Summary
Online research has experienced astonishing growth over the past 15 years. To keep up with this growth, researchers have developed new ways of accessing and utilising respondents. Nevertheless, they can still find it difficult to complete the needed number of interviews on time, particularly when the target population is rare or in high demand. For this reason, it is common today for researchers to use more than one sample source for some types of project, such as a tracking survey that measures change over time. Adding one or more sample source to the original might address the need for more respondents, but some evidence suggests that it might also decrease sample representativeness and reduce response accuracy. In this paper, we introduce a new methodology that enables researchers to select potential survey respondents from either a single sample source or multiple sources based on how well their characteristics match an appropriate, evolving standard with demonstrated evidence of external validity. We also present evidence suggesting that, in the aggregate, respondents who are selected through the new methodology are more representative of the target population than respondents selected by other means. Finally, we consider possible implications of the new methodology on methods other than online research with non-probability samples.

Summary
In the absence of studies addressing how to sample advertising in content analysis in a reliable manner, this paper aims to determine the most efficient way to select from a year’s worth of issues for conducting content analysis of advertising in terms of sample size and type. This article specifically examines the cases of three Chilean consumer magazines: Qué Pasa (a weekly news magazine), Cosas (a bimonthly celebrity/interview magazine), and Paula (a monthly women’s magazine). Results show that the most efficient sample size depends on the magazine type/frequency (ranging from 6 to 12) and that stratified random sampling is more efficient than the use of a simple random method.

Summary
This paper outlines how Coca-Cola changed its approach to consumer behaviour in order to deal with an increasingly complex global beverage market. It outlines how the company collects and analyses its consumer research data from over 100 countries, as well as providing advice on how to respond to pressures the market research industry faces from new entrants, methodologies and technologies. It also emphasises the importance of its "protocol custodian", who oversees the application of the newly designed global framework and ensures it remains locally relevant without compromising global comparability. The paper also provides recommendations for the research industry and examples of the type of analysis that is conducted by Coca-Cola each month.

Summary
The internet is increasingly being used for cross-sectional surveys and online panels. Although internet accessibility is growing across developed and developing countries, it seems unlikely that the internet alone will ever provide complete coverage of the general population. Given the incomplete coverage and imbalanced penetration rate of the internet across segments of the population, it is pertinent both for survey companies and academics to assess the potential of mixing the internet with other survey modes as part of a strategy to assure validity of inferential samples when surveying general populations. The purpose of this research was to evaluate to what extent coverage error in internet surveys can be reduced by surveying the offline population via telephone. We use data from Eurobarometer collected in the EU27 member states to simulate first an internet-based survey and then a mixed-mode survey combining the internet with the telephone. Comparisons are made to identify differences in the socio-demographic characteristics of internet households and those of non-internet households with telephone. Coverage error is also estimated in each survey design. Findings reveal significant socio-demographic differences and although the coverage error is reduced in the mixed-mode survey design, it cannot be completely eliminated. Moreover, the outcomes are not homogeneous across countries.

Summary
This paper outlines new evidence on what happens when questions from major social surveys are asked of online survey panellists. The paper shows how difficult it is to control for 'panellist bias' and produce unbiased population estimates but also that, for some statistics, panel data can provide a surprisingly close match to the gold standard surveys of government.

Summary
This presentation teaches how to generate the social media research results you desire regardless of what the true results are. It will show how to gather social media data from the internet using inappropriate sampling methods and how to select the wrong pieces of data and code it incorrectly. The topics of sampling, weighting, data quality, sentiment analysis, and text analysis are highlighted to highlight the full range of options for mistreating data. The ultimate goal will be to create set of data that reflects your predispositions towards a topic as opposed to reality.

Summary
Standard correspondence analysis plots are readily misinterpreted by research users. This paper presents a new plot, called the moon plot, which is less susceptible to misinterpretation. Row points are plotted in the traditional way. Column points are plotted equidistant from the origin, with their directions from the origin as in traditional correspondence analysis plots, and the information traditionally communicated by the distance of the points to the origin instead communicated by the size of the fonts of the labels.

Summary
Market segmentation is a widely accepted concept in marketing research and planning. Although cluster analysis has been extensively applied to segment markets in the last 50 years, the ways in which the results were obtained have often been reported to be less than satisfactory by both practitioners (Yankelovich & Meer 2006) and academics (Dolnicar 2003). In order to provide guidance to those undertaking market segmentation, this study discusses the critical issues involved when using cluster analysis to segment markets, makes suggestions for best practices and potential improvements, and presents an empirical survey that seeks to provide an up-to-date assessment of cluster analysis application in market segmentation within a six-stage framework. Analyses of more than 200 journal articles published since 2000, in which cluster analysis was empirically used in a marketing research setting, indicate that many critical issues are still ignored rather than addressed adequately.

Summary
The Net Promoter Score (NPS) and the American Customer Satisfaction Index (ACSI) are metrics used to predict sales, profit and share price change. We identify problems with the design of both the NPS and the ACSI. In particular, we find that the NPS does not measure negative word of mouth effectively, and we argue that the ACSI is similarly insensitive to dissatisfaction. This is because ex-customers and never-customers are not sampled in these metrics, and these are the people who express most of the negative sentiments about brands/companies. We propose a method of measuring the effect of word of mouth using the volume and mean impact on purchase probability of both the positive and the negative word of mouth expressed by users of the category.

Summary
This paper discusses whether the methodologies of "gold standard" UK government research programmes like the British Crime Survey could modified to reduce costs whilst maintain their necessary rigour. In particular, it examines four hypotheses, mainly about the implications and feasibility of introducing an element of online research.