Erhard Meier

Worldwide Readership Research Symposium Valencia 2009 Session 1.1

Synopsis

This paper describes salient trends in readership research since the last Symposium two years ago. It is based on information collected for the new edition of the “Summary of Current Readership Research”, which tabulates detailed descriptions of the methodological features of 98 national readership surveys from 74 countries. The particular details, some large, some small, which together constitute the measurement system of a given survey, can vary considerably from survey to survey. An increasing tendency of conducting parts of a readership survey using online interview methods and the confusing variety of existing systems of measuring audiences of online editions, all producing different results, are also noted.

Special Note

As in previous years, our grateful thanks are due to all our correspondents who provided the information. In addition to the 98 total audience measurement surveys listed in the main part of the “Summary”, there are over 60 other readership surveys listed in a special section of the “Summary”, most of which measure the audiences of other populations.

Introduction

As in previous editions of the “Summary of Current Readership Research” – in fact since the first edition of 1981 which we prepared for the first Symposium in New Orleans – our definition of a national readership survey remains unchanged. It denotes a total audience measurement survey which aims to provide the “currency” for the print medium in a given country. However, this means that we list more than one survey for a given country when different surveys are used as currencies. In most such cases, many of them in Latin America, others in China and India and some in Europe, they are commercially competing surveys, while in other cases, for example those in Canada and France, they are different surveys providing the currencies for, respectively, newspapers and magazines.

The 2009 edition lists three new surveys amongst its total of 98 national readership surveys. These are the Latin American EGM (Estudio General de Medios) surveys from Bolivia, Uruguay and Venezuela.

When concluding my overview of trends at the last Symposium, I found that most surveys seem to be in a “wait and see” position with regard to the various demands made on their methodological capacities and budgets, caused by steady increases in the number of publications, drastic falls in advertising revenues and in circulations, the needs of multi-media planners, the need to enhance basic audience counts with quality of reading, engagement and other added value data, the problems of respondent co-operation and, finally, the general increase in the use of online and other new media.

Today, while trying to meet new demands, the old demands stay unaltered. These are the need to provide data acceptable to all sides of the industry, to provide stable data but equally to be flexible enough to accommodate new needs, and to provide data to the highest possible standards compatible with being cost effective and affordable.

The information collected for the “Summary” demonstrates the variety of methodological solutions which, however, are all designed to serve the same end: namely to achieve reliability of response. That is to say, designing a readership survey – this includes both sample design and questionnaire – is a balancing act between what is desirable and what is practicable to ensure that responses given by respondents will be as reliable and the results for magazines and newspapers as fair to each other as possible.

Seemingly insignificant methodological details are all important and will affect the results. For example, when we decide to use a twelve months as opposed to a six months filter question, and if in the case of the former we decide to say “in the past 12 months” instead of “in the past year”, we know that these details will make a difference. A readership survey design is based on a multitude of such seemingly small details.

In the following, I will demonstrate such variety on only a few examples – these being interview methods, means to reduce title confusion and order effects, and types of “when last” and frequency of reading questions. This will be followed by a note about response rates and a look at the current state of the increasing importance of online samples for the readership currency. Quite a separate subject is that of measuring online reading as part of our national readership surveys, a subject which we will look at as well.

A review of recent developments country by country around the world concludes this paper.

Methods of Readership Measurement

Thirty to forty years ago, when I started to be involved with media research, the main debate concerning readership measurement was about the relative merits of the Through the Book (TTB) versus the Recent Reading (RR) methods. This was very much in the forefront at the first Symposium in 1981 and many of the subsequent events. Technical awareness about the inner workings of each method was then of greater concern and spread much wider amongst users than it seems today. The fact that Recent Reading has survived Through the Book as our principal measurement model is probably the reason for the seemingly decreasing interest, but this does not mean that Recent Reading is without flaws. The Recent Reading model bias, as expressed by the terms parallel and replicated reading, and much debated in earlier Symposia, still remains. So does disagreement about other technical aspects, for which there are a variety of possible and often contrary solutions.

Let me give a few examples of some of these technical aspects. First, with regard to the model, there seems nearly universal agreement, as said. The new “Summary” identifies 96 surveys which use the Recent Reading model, two which use Specific Issue Readership (SIR) – in Japan and the Netherlands – and none which uses Through the Book, with the exception of the Australian Roy Morgan survey which employs TTB for business titles, whilst other publications are treated according to the RR model.

I should make it clear that this analysis refers to the method of obtaining the main currency measure, average issue readership (AIR), not additional readership surveys like the Issue Specific Study conducted in the USA by MRI. The MRI National Study’s currency is based on using Recent Reading.

Mode of Interview

As to the mode of interview, personal face to face interviewing is still the main method of collecting readership data around the world, some are telephone and others self completion surveys. However, an increasing number of surveys now use mixed methods, as may be seen from the following:

Personal – pen and paper – 55 surveys

Personal – CASI, CAPI or DS-CAPI – 12 surveys

Telephone – CATI – 8 surveys

Self-completion – leave behind or postal – 11 surveys

Mixed methods – 12 surveys

There are 12 mixed methods surveys compared with five two years ago. The actual combinations of methods used are different for each one, as may be seen below:

Pen and paper + CAPI (India-IRS)

Pen and paper + CASI (Germany-ma)

Pen and paper + self completion (Norway-NM)

Pen and paper + telephone (Ukraine-NRS)

Pen and paper + online (Hong Kong-NMI)

CASI + CAPI (Austria-MA)

CASI + online (Netherlands-NOM)

CAPI + telephone (Spain-EGM)

DS-CAPI + self completion (UK-NRS)

Telephone + self completion (Norway-F&M) Telephone + online (Portugal-BI)

Self-completion – online (UK-TGI)

As may be seen, four surveys use online for part of their interviews. Not one of the 98 surveys uses online samples for all its interviews at present. (See also Using Online Samples further below.)

Reducing Title Confusion and Order Effects

The Recent Reading model employs a recognition method – publications are prompted to respondents. Most national readership surveys use visual stimuli as recognition aids in the form of mastheads. How these mastheads are presented is important. The aim is to reduce title confusion as much as possible. The major variations are shown below:

Grouped titles mastheads – 12 surveys

Single title mastheads – 63 surveys

Recent covers of magazines – 2 surveys

No mastheads (typescript or telephone) – 19 surveys

No information – 2 surveys

The two surveys showing covers of magazines are the SIR surveys of the Netherlands and Japan.

Concerning the reduction of order effects, as a general rule, the titles of newspapers and magazines tend to be rotated within their respective publication groups. Nearly all surveys do this. In most cases publication groups are defined by their frequencies, that is to say, dailies, weeklies, fortnightlies, monthlies and quarterlies. However, it is a different matter when it comes to rotation of publication groups. Whether publication groups themselves should be rotated is still open to debate. As may be seen from the following, research practitioners are fairly equally divided on this subject:

Publication groups fixed – 51 surveys

Publication groups rotated – 46 surveys

No information – 1 survey

It is interesting to note that of those 51 surveys that keep their publication groups fixed, the majority start with dailies followed by weeklies followed by monthly publications. However, eight surveys use the reverse order, starting with monthlies followed by weeklies followed by dailies. Knowing that these details have bearing on the results, it would be instructive to rehearse the arguments pro and contra each version. Alas this paper is not the right place for this. Consulting the Symposium archive, available on the Symposium’s website, would be a good way to look for answers in papers discussing these issues.

The Recency Question

Another fine point, but one which arguably can make an important difference, is the way in which the recency question is administered. With the Recent Reading model, whether “read or looked at within the publication interval” defines the average issue readership currency. This means “yesterday” for dailies, “last 7 days” for weeklies, “last 4 weeks” for monthlies etc.

The question here is whether respondents should be shown what the response categories are or whether they should answer unprompted. In our terminology, if the pre-codes of the relevant time categories are prompted, they are “disclosed”, if they are not shown to respondents (and the interviewer enters the appropriate codes according to the un-prompted answers given), then they are “hidden”. The numbers of surveys using either method are shown below:

Pre-codes “disclosed” – 70 surveys

Pre-codes “hidden” – 22 surveys

Depends on publication group – 4 surveys

Not applicable (frequency question only) – 2 surveys

As may be seen, most surveys use the “disclosed” method, but a substantial minority of 22 surveys prefers the “hidden”, unprompted method.

The Frequency Question

The frequency question is important because in combination with recent reading results, it enables us to calculate probabilities, cumulative readerships and campaign schedules. Some surveys, notably Sweden’s Orvesto Consumer, only use a frequency question to calculate average issue readership.

Concerning the finer details of the frequency question, as in previous years, we find there is little agreement across the 98 surveys of our “Summary”. For one thing, the number of scale positions varies greatly from survey to survey. We find some surveys have 3, some 4, others 5, some 6, others 7 or 9 or even 12 points on their scale. Arguably, there must be differences in results.

Another variation is whether a “verbal” or a “numeric” scale is used. Where a verbal form is used, this is in most cases accompanied with numeric explanations as in the example of the British NRS. This survey has four scale positions: “almost always (at least 3 issues out of 4), quite often (at least 1 issue out of 4), only occasionally (less than 1 issue out of 4)” and “not in the past 12 months”. An example of the numeric variety is given by the German ma survey: “How many of the last 12 issues did you read? 1 issue, 2-3 issues, 4-5 issues, 6-7 issues, 8-9 issues, 10-11 issues, 12 all issues”. As may be seen from the analysis below, most surveys use the numeric variety. However, the verbal form is used by the quite substantial number of 19 surveys:

Frequency scale “verbal” (most with numeric explanations) – 19 surveys

Frequency scale “numeric” – 73 surveys

Mixed scale – 2 surveys

No frequency question asked – 4 surveys

Unfortunately it is beyond the scope of this analysis to investigate the hypothesis that regional, cultural, educational and social connections are possibly behind a national readership survey’s particular choice of reading frequency scale.

Response Rates

Response rates are a crucial factor contributing to the statistical accuracy and stability of a readership survey. Sadly, as we have noted over the years, declining response rates are a problem for many of our surveys. As on previous occasions, I have asked my “Summary” correspondents to tell me their response rates, this time for the two years 2007 and 2008. I don’t know what caused it, but replies to this question were given for only 27 individual surveys compared with 43 surveys for the previous edition of the “Summary”. In addition to the 27, two of my correspondents replied in general terms relating to the surveys conducted by their respective companies in different countries within their geographical regions.

While I can’t regard the results as representative, the picture that emerges is not rosy, but at least there are six surveys which report a slight improvement in their response rates between 2007 and 2008. These are Finland-KMT from 70% to 71%, GreeceBARI from 37% to 38%, Poland-PBC from 43% to 44%, Puerto Rico-TGI from 62% to 63%, Thailand-Nielsen Media Index from 64% to 66% and UK-NRS from 51% to 53%.

Response rates are not always calculated in the same way, survey methods differ and there are substantial cultural and social differences. However, what is clear is that, while a certain number of surveys still report relatively high response rates, there is a wide range from very low to very high. Without identifying individual surveys, the range of reported response rates was as follows:

Under 30% – 3 surveys

30%-39% – 4 surveys 40%-49% – 5 surveys 50%-59% – 2 surveys 60%-69% – 7 surveys

70%-79% – 3 surveys

80% and over – 3 surveys

Using Online Samples

As we have seen, an increasing number of surveys use different modes of interview in addition to their main samples. Many of these undoubtedly do this in order to improve response rates in a cost effective way. Four surveys use online interviews, but there are considerable variations in the proportions of online and main samples, as follows:

Hong Kong-Nielsen Media Index – pen and paper + online (% not stated)

Netherlands-NOM – 40% CASI + 60% online

Portugal-Bareme Imprensa – 94% telephone + 6% online UK-TGI – 90% self-completion + 10% online.

Because of the difficulties of achieving satisfactory representative samples of the population with online sampling, no national readership survey at present relies exclusively on online interviews. However, special add-on surveys and readership surveys amongst special populations (like the American Issue Specific, the Finish Decision Makers, the French Premium or the new Swedish Orvesto Day surveys) make total use of online interviewing. Others, like Norway’s NMI, Russia’s NRS, Slovenia’s NRS or Taiwan’s Nielsen Media Index report that they plan to embark on tests to investigate the feasibility of measuring print audiences using online samples. In the case of Canada’s NADbank, two tests have been completed, both with discouraging results.

Measurement of Online Reading

When it comes to online reading – that is reading or looking at publishers’ websites – the situation is fairly complicated. Very briefly, we identify two basic concepts for measuring the audiences of publisher’s websites. One is the recall method – as for our standard readership surveys – which involves asking respondents of sample surveys whether they have read or looked at a given website, and if so, when last.

The other concept is internet site measurement, involving the counting of “clicks” relating to a publication’s website. This method sub-divides into site-centric and user-centric measurement. Site-centric involves the measure of total activity on the site and includes page requests regardless of where they come from geographically or from what sort of device. User-centric is sample-based, making use of panels providing demographic information of individual panel members. There are also hybrid systems in place which combine site- with user-centric approaches.

Different systems have different advantages and disadvantages. Internet site measurement systems are all of great technical complexity and their definitions are often not very clear to the outsider. Not surprisingly, they tend to show different results to each other and to recall measures of online reading.

Of great interest is, of course, whether definitions and results of passive online reading measurement systems and of Recent Reading recall measures relating to printed editions are comparable and whether they can be sensibly combined into an integrated measure of “portfolio” reach.

I asked our correspondents whether their surveys include measures of online readership, whether they publish integrated print and online audiences or conducted experiments during the last two years. The answers indicate that 38 of our 98 surveys currently publish title specific online reading results, most of these based on recall methods and combined with the standard average issue readership results of printed editions. The numbers of websites covered range from just a handful (an example is Canada’s PMB, which includes eight websites) to over 100 (as for example in Finland’s KMT).

Three surveys – included in the above figure of 38 – publish integrated results based on passive site measurements which are fused with main survey results. These are Hungary’s Media Analysis, Netherlands’ NOM and Sweden’s Orvesto Consumer.

Some correspondents report that experiments concerning online reading had been conducted, but that the results have not been published or are “on hold”. The surveys here include Belgium’s CIM, Italy’s Audipress, South Africa’s AMPS, Switzerland’s MACH Basic and the UK’s NRS.

New Developments by Country

The following is a review of recent developments as reported to us by our correspondents around the world.

Australia-Roy Morgan. The Australian Roy Morgan Readership Survey has harmonized its demographics during the last two years to enable analysis of all media. They also report that they are in discussions on engagement/involvement measures for Print as for TV. Reading online is now separated from hard copy reading, and print and website multi media scheduling is made possible through the introduction of new software called WebScheduler. Other new developments include cinema scheduling, word of mouth measures and even richer and more detailed profiling than before, according to our correspondent.

Austria-MA. In 2008 the Austrian Media-Analyse changed its sampling procedures relating to the selection of individuals in selected households. Prior to that both selection of addresses and selection of individuals, employed random methods – the Kish-grid method was used for individuals. Now individuals are selected by quota. Also new is that two “free” addresses are issued in addition to the original eight. It is interesting – perhaps astonishing – to note that their reported overall response rate for 2007 was 80%. Now, with quota, they have no response rate any more. My Austrian correspondent adds that, with the new procedures, 77.5% of interviews were conducted at original addresses and 22.5% at the nearby additional addresses.

Bahrain-NMA. Our correspondent for the Near and Middle East National Media Analyses covers Bahrain, Egypt, Iraq, Jordan, Lebanon, Oman, Qatar, Saudi Arabia, Syria and the United Arab Emirates. For all these countries he reports increases in the number of publications to be included in these surveys. (Iraq NMA only covers newspapers.)

Belgium-CIM. The CIM Survey Press and Plurimedia of Belgium changed its title rotation procedures for the 2007/08 survey. While the method of rotation of publication groups remains the same as before, the alphabetical orders of titles within groups are now reversed for half the sample. Previously there was a random starting point. This had some impact on publications with close to similar titles, according to our correspondent.

Brazil-Estudos Marplan EGM. As a sample quality measure, the Brazilian EGM started contacting potential respondents living in high-class condominiums by telephone in order to arrange face-to-face interviews at locations convenient to respondents.

Canada-PMB. As from autumn 2009, the Canadian PMB survey will increase its reporting frequency from annual to semi annual.

Canada-NADbank. From the Canadian newspaper measurement survey NADbank we hear of a test concerning the readership model. They are conducting a parallel test of their telephone questionnaire in one major market, changing the recency question in the test cell from their traditional standard issue specific model (“Did you read or look into yesterday’s issue of…?”) to the recent reading model (“When did you last read of look into any issue of…?”). At the time of writing, the results of this experiment were not yet available.

Croatia-MEDIApuls. For clients of the MEDIApuls survey, new software will be available during 2009 based on web applications, including advertising expenditure and multi-media consumption involving print, radio and the web.

Croatia-TGI. TGI Croatia now includes socio-economic levels of the population plus Media Neutral Planning segmentation.

Finland-KMT. Four separate developments were reported from Finland. First, concerning the readership interview, which is conducted by telephone, the groupings of publications has been changed in 2008 for the 12-months filter question from grouping by frequency to grouping by topic. Our correspondent reports that this made the interview more fluent and slightly shorter in time without notable affects on the currency. Second, the sample now includes the five per cent of the Finnish population who can only be reached through their employers’ mobile phones. Previously these could only be contacted if their home addresses were known. This change affected business publications in particular. Third, portfolio reach was reported as from 2008, involving publications whose websites are measured via the survey’s postal questionnaire. Finally, word-of mouth (WOM) questions for eleven product categories were added to the postal questionnaire in 2008, while for 2009 new decision making and information search questions were developed for each product category.

Germany-Media-Analyse. Since our last review, the German ma developed an estimation model for newspapers at the local level, called L-value, comprising two types of values for the database: plausible efficiency values concerning coverage and populations in municipalities, and person related efficiency values instead of household coverage as previously used. Secondly, a methodological test was conducted in autumn 2008 with 331 individuals as part of their Media Scan Project. This project involves the scanning of barcodes in an add-on survey amongst a sample of ma respondents, with the objective of providing time-based information on print usage, including exposure accumulation, duration of reading, amount of reading per issue, multiple exposures and more. A larger real-life test is planned to start in September 2009 with the intention of linking the resultant data with the currency of the ma, using modeling and simulation algorithms still to be optimized.

Greece-BARI. The Sesame software has recently been introduced by the Focus company of Greece for the BARI/National Readership Survey.

Guatemala-Estudio Multimedios. This survey’s sample size has been increased from 6,000 to 10,000 per annum.

Hong Kong-Nielsen Media Index. The booster samples of the Nielsen Media Index of Hong Kong have been re-defined and doubled in size. Since October 2006, the booster sample consists of respondents with monthly personal incomes of HK$20,000 or more. Furthermore, since January 2008 Chinese booster interviews are conducted online.

Hungary-Media Analysis. The Hungarian survey – conducted with CAPI – has introduced an audio effect into the questionnaire: a voice names the newspapers and magazines when their respective mastheads appear on screen.

India-IRS. The Indian Readership Survey is now partly conducted with CAPI. Following an encouraging parallel pilot in 2008, CAPI is currently implemented in four top metropolitan areas.

Indonesia-Nielsen Media Index. Two updates are given regarding the Indonesian Media Index. Its coverage now includes the Greater Bandung and Greater Semarang (urban and rural) areas, and an online readership measure has been introduced.

Iran-TGI. The Iranian TGI questionnaire has introduced three new questions, one about daily use of SMS, a second about attention to advertising on SMS and a third about attention to advertising on the internet.

Ireland-JNRS. The Joint National Readership Survey of Ireland is conducting preliminary pilots during 2009 to evaluate the DS-CAPI methodology.

Italy-Audipress. Audipress introduced two experimental topics to the Italian readership survey during the autumn of 2008, one designed to examine the relationships between printed and e-papers, the other dealing with the engagement of readers towards newspapers and magazines of their choice. The results to these questions are taken into account for the definition of a planned new model of readership which is under discussion in Italy at present.

Kenya-KARF-ES. Our Kenyan correspondent reports that the Audience Research Foundation’s Establishment Survey – formerly AMPS (All Media and Product Survey) – has increased its sample size from 2,000 to 6,000, with sampling frames provided by the Central Bureau of Statistics.

Netherlands-NOM Print Monitor. The practitioners of the Dutch NOM survey, which uses the SIR (Specific Issue Readership) model, further enhanced their method by introducing two new measures. One is the provision of reading probabilities for specific daily issues of newspapers (e.g. Monday issues) in addition to AIR, and the other is accumulation of readership in time, allowing time planning for magazines. Another development is the fusion of print and internet data, published for the first time in 2009 following testing in 2008.

New Zealand-Nielsen NRS. Our New Zealand correspondent reports an increasing demand to include media engagement measures within the Nielsen NRS.

Norway-Norsk Medieindeks. The Norwegian NMI survey, experiencing a growing number of publications, declining response rates and difficulties in establishing representative samples within tight budgets, is actively investigating online interviewing using Synovate’s own access panel in addition to the face-to-face method. Our correspondent adds that fair treatment of all magazines is all important – he regards web interviewing as being close to face-to-face interviewing concerning stimuli and rotations.

Russia-NRS. Three recent changes were reported regarding the Russian NRS. Firstly, since 2008, all Russian interviews are now conducted with CATI. Previously only Moscow and St. Petersburg were sampled by telephone, while the rest of the country was sampled face-to-face. Secondly, the Moscow sample and media list has since 2009 been divided into six subsamples, with each of 260 Moscow titles being presented to half of respondents. Previously each title was presented to twothirds of respondents. As before, the missing title data are re-constructed through statistical fusion methods. And thirdly, the St. Petersburg sample is being subdivided into three sub-samples with each title being presented to two-thirds of respondents, as was the case previously in Moscow.

Slovenia-NRS. The Slovenian NRS is currently experimenting with web-interviewing of respondents who are not willing to participate in the survey face-to-face. This is planned to be a standard feature of the NRS as from January 2010, when about 10% of survey respondents are expected to be interviewed that way, using a shorter self-completion online questionnaire than the standard CAPI questionnaire.

South Africa-AMPS. Three developments are reported from South Africa. The All Media and Product Survey has moved from CAPI to DS-CAPI. The sample universe is now defined as adults 15 years and over instead of 16 and over. And the survey is now conducted in two national waves instead of one national and one large urban wave.

Spain-EGM. The Spanish Estudio General de Medios has now a very large total sample size of 95,000 respondents – the largest such sample in Europe. It consists of three parts: 30,000 CAPI interviews for a central multimedia study, 45,000 CATI interviews for a newspaper extension study, and 20,000 CAPI interviews for a magazine extension. These sample sizes have changed in 2008 – the multimedia CAPI interviews have been reduced from previously 43,000, the CATI interviews have increased from 32,000, while the magazine extension CAPI interviews are new. Thus magazine results are now based on a total sample size of 50,000 CAPI interviews. An additional development reported to us is the introduction of fusion processes to obtain a single database which will integrate the different mono-media newspaper, magazine, radio and television studies with the existing multimedia study.

Sweden-Orvesto Consumer. From Sweden, we can report two noteworthy developments. One was the launch in August 2008 of Orvesto Day, a daily measurement of newspaper reading amongst the Swedish population aged 15-79, based on 1,000 online interviews per day. It provides daily reach measures using the receny model. It allows contact guarantees to be given to advertisers and agencies and provides the opportunity for recall and effectiveness studies. The other was the launch, also during 2008, of Orvesto Internet, combining Orvesto Consumer with a user-centric online panel – the panel consisting of recruits from the Orvesto Consumer survey, thus providing a single source multi-media database and, using the Sesame software, allowing for cross-media planning involving print, television, internet, cinema, direct mail and radio.

Switzerland-MACH Basic. Maintaining response rates was a major issue in Switzerland over the last two years. A further increase in contact attempts and optimization of contact schemes were introduced. Our correspondent adds that maintaining each year the same sample quality is more expensive than the year before.

Taiwan-Nielsen Media Index. The Media Index of Taiwan changed from using pen and paper questionnaires to CAPI (Computer Assisted Personal Interview).

Tanzania-AMPS. This survey increased its annual sample size from 2,500 to 4,900.

Thailand-Nielsen Media Index. This survey increased its annual sample size from 8,000 to 9,000 interviews and changed its reporting frequency form twice yearly to four times per annum.

Uganda-AMPS. This survey increased its sample size from 2,500 to 5,790.

UK-NRS. Two experimental surveys concerning online and integrated print and web readership have been carried out by the British National Readership Survey. One, conducted between November 2004 and September 2008, consisted of additional questions within the standard CAPI interview relating to 17 publications and their websites. The other, conducted between June 2007 and September 2008, was based on a self-completion follow-up questionnaire relating to 65 publishers’ websites. The results of neither study have been published.

UK-TGI. Regarding the UK’s TGI we are informed that about fifteen per cent of its interviews have been completed online. Regarding its media list, as from October 2008, all readership questions have been moved to the front of the questionnaire. For the Northern Ireland version, logos are shown in colour alongside each title.

USA-MRI. In 2009, MRI introduced AdMeasure, a new product which links the MRI Starch print ad effectiveness studies with both the MRI National Study and the MRI Issue Specific Study. This allows print ad effectiveness information to be seen in the context of a magazine and a particular issue audience.

Conclusion

In conclusion, whilst the variety of solutions concerning the many details of our standard readership measurement methods seems as great as ever, we are at least in agreement about the basic definition of reading and of average issue readership of printed newspapers and magazines. When it comes to the measurement methods of online editions and of portfolio reach, there are many solutions on offer, but from my perspective, we seem much further away from agreement about general standards and comparability than we are, or ever were, with print research. There is still much to learn.