|
|
MRS members may request copies of these papers by emailing info@mrs.org.uk
International Journal of Market ResearchVolume 54, Issue 1 2012: Summary of ArticlesThe growing efficacy of telephone canvassing at the 2005 UK General Election, Charles Pattie and Ron Johnston (University of Sheffield)
Our Call for Papers in 2010 focussed on research related to measuring voting intentions. Pattie and Johnston examine the impact of canvassing on voting behaviour – in particular, whether telephone canvassing encourages voters to be more likely to vote for the party making the call. The data sources are the British Elections Survey (BES) for 1997. 2005 and 2010, and, the Campaign Internet Panel Study (CIPS) for 2010. The authors discuss the recent history of canvassing with the emergence of ‘phone based methods first measured in the 1997 BES, and the challenges faced in analysing the impact of canvassing on subsequent behaviour. They describe the data used in the research, the control variables and the modelling process developed to measure the impact. Their findings show that the lack of influence shown in 1997 has been confirmed in their latest analysis, but the results for 2005 and 2010 show that political parties gained an advantage through telephone canvassing. The authors think this is due to the development by the parties of databases to target their canvassing through this channel and using professional call centres to replace declining local party member resources, especially as there is no indication that the volume of such telephone calls has increased over time. However, the authors feel that this may be a short-term strategy due to the ability by voters to screen out unsolicited calls, and, the rapid growth in mobile technology making it harder to target residents by constituency. They conclude that parties will need to further re-think their canvassing methods before the next election. An improved, practical model of consumer choice, Len Marchant (independent consultant), Phil Prescott (University of Southampton), Nic Jackson A:Cet Ltd)
As the authors state, a fundamental challenge for marketing practitioners is to better understand brand choice by consumers. For market researchers, the challenge is to devise more effective methods for measuring this that deliver robust data to improve decision making using easy to present methods. From a potential respondents’ perspective, the research process needs to keep them fully engaged throughout the interview by using straightforward questions that focus on the essential issues – in essence it requires each respondent to identify the brands which constitute his or her repertoire and then rate these relative to the leading four brands in the market according to their buying preferences, a process which the authors consider is straightforward and user friendly and can be administered via a self-completion interview using CAPI. A further requirement identified by the authors is that the method needs to be capable of being seamlessly applied in qualitative and quantitative situations. The Monte Carlo based model method developed by the authors is based on earlier models, including a heuristic based approach which the authors now believe contains flaws addressed in the new approach. The development of the model is described in detail, including how it enables competition between well differentiated brands to be explored, an issue which the authors believe is less well addressed by applying the Dirichlet model. The building of the model is described in detail, and the application is illustrated through a discussion of brand strategy scenarios in the breakfast cereal market. Using card-based games to enhance the value of semi-structured interviews, Jennifer Rowley (Manchester Metropolitan University), Rosalind Jones, Sonya Hanna (University of Bangor) and Magda Vassiliou (Manchester Metropolitan University)
‘Gamification’ is a new term in market research, used to describe the application of the creative methods used in computer games to engage more effectively with respondents in online research, for example through scenarios or question design (see ‘Improving online surveys’, Pulston J, Conference Notes, IJMR Vol. 53 Issue 4, 2011). However, as Rowley et al remind us, games have a long history in qualitative research methods, and can help reduce the ‘research effect’ by enabling participants to explore issues and express their views with minimal prompting from the researcher. In addition, the authors argue that whilst game based methods are commonly used in qualitative research, there is relatively little discussion of these methods in the literature and text books covering this field. In the paper the authors describe the role of card based methods to provide qualitative validation of theoretical models, exploring issues such as definitions, priorities, processes, challenges, issues, difficulties, views on the future and critical success factors. Rowley et al primarily describe in detail three applications of card based methods in projects based on semi-structured interviews with experts covering, firstly, the marketing practices of small software firms; secondly, ‘place’ (geographic) brand management; thirdly, adoption and management of ebooks in academic libraries. The authors conclude by summarising the outcomes and the benefits derived from applying card games in these projects, and in particular the richer data obtained from the interviews than might otherwise have been the case using more direct methods. International segmentation: toward a third path between global and national, Belvaux Bertrand and Guibert Nathalie (University Paris II – Sorbonne Universités)
The increasingly globalisation of markets leads marketers to hope that they can develop segmentations that transcend national borders, enabling global strategies to be developed. Bertrand and Nathalie investigate this topic, and propose a methodology to compare national and global level segmentations, to help determine whether a global solution is viable. The authors use the η² test to assess the quality of a global level solution against a national approach and apply Means End Chain (MEC) theory to compare consumer motivations and consumption patterns from data collected in a three country survey on wine consumption. Their findings identify differences by country which lead to the conclusion that marketers need to consider national and international issues when developing segmentations if their strategies are to be successful. They also conclude that this method provides a deeper understanding of consumption patterns. Adjusting self-reported attitudinal data for mischievous respondents, Michael Richard Hyman and Jeremy Sierra (New Mexico State University)
Bias and error can arise for a number of reasons in responses in self-completion surveys, as described by Hyman and Sierra. They content that best practice survey design on its own will not eliminate those situations where respondents attempt to purposefully deceive the researcher by providing phoney responses – what the authors call ‘mischievous respondents’. The authors describe the issues generally leading to bias and error in self-completion surveys, drawing on the extensive literature in this field. They discuss the reasons that might cause respondents to deliberately provide incorrect answers, distinguishing faked responses – where respondents try to provide the researcher with what they perceive will be favourably received answers – from those where the objective is malicious. The authors discuss the methods commonly used to detect sources of error and bias, proposing a new algorithm to identify and purge mischievous respondents from polychotomous attitudinal datasets, and how their method differs from outlier analysis. They then describe the tests of this solution undertaken on the responses from four surveys. The paper ends with recommendations for extending the research in this field. Survey methods in an age of austerity: driving value in survey design, Joel Williams (TNS-BMRB)
At the 2010 IJMR Research Methods Forum conference, Richard Bartholomew (Joint Head Government Social Research, UK) described the reasons why major government commissioned surveys continued to be based on random probability sampling methods and face to face interviews. However, Bartholomew described how the current economic climate was leading to survey methods being reviewed in an effort to reduce costs, but that any methodological changes should not compromise reliability or the capability to track trends over time. This is the challenge addressed by Williams in a revised version of his ’ Research 2011’ conference Best Paper award winning paper. The author calls this ‘smart’ research design, exploring possibilities that provide either the same level of quality but at reduced cost, or a revision of quality levels to achieve cost savings – and the compromises that result. As Williams describes, ONS and the ESRC have instigated methodological reviews in response to the current situation, and research agencies are also exploring options, the theme of this Forum article. The author describes research conducted by TNS BMRB to explore whether online access panels could provide a lower cost alternative to more traditional methods. This exploratory work used questions drawn from three government commissioned surveys (British Crime Survey, Taking Part and the Scottish Household Survey), administered in a single study to a sample drawn from an access panel. Four hypotheses were tested in the research: home internet access only plays a minor role in differences; demographic calibration of online samples will not eradicate bias; subject specific variables will not reduce bias; relationship statistics are less biased than absolute statistics. Williams concludes that firstly, the relationship between home internet access and the likelihood of being an online panellist is quite weak. However, too little data on internet behaviour is currently collected on these particular surveys to fully explore this issue. Secondly, demographic weighting will not eradicate differences, but could suffice where required accuracy is low. Thirdly, it is unlikely that a suitable small group of ‘hook’ questions can be identified to cover sufficient topics. Fourthly, the factors underlying correlations cannot easily be identified (e.g. the presence and impact of flat-lining), but the real problem is in generating reliable absolute data. Finally, differences might be mode effects, but in some cases, online self completion may produce more honest answers. more widely in the Spanish market.
|
|
What's New - Membership - Company Partner Service - Members' Area - Code/Guidelines - Qualifications - Training - Awards © Copyright 2012 MRS - Privacy Statement - Terms and Conditions - Legal Information | ||
