Readership accumulation study: methodology

NRS Survey of Readership Accumulation over Time:
Summary of method

In essence, a sample of 7,000 people filled in diaries of their newspaper and magazine reading, for one week each. Every day, for each publication read, respondents entered the cover date of the issue as shown on the front page, indicated whether it was their own/household copy, and stated whether or not it was the first time that particular issue had been read. The subsequent analysis of the data then measured the difference in time between the appearance date of that issue and the date on which it was first read. By combining all such measurements for a given publication, across all its issues, it was possible to plot the rate at which readership of that publication built up through time.

The sample

The sample of adults was recruited through the NOP Random Location Omnibus Survey, phased across the period October 2001 to May 2002. In a representative sub-sample of Omnibus interviews, respondents were asked to keep a diary of their reading of newspapers and magazines for one week. 70% agreed to do so. At the end of the week interviewers collected the completed diaries, or where it was not possible to collect them the diaries were posted back to NOP. Among those who agreed to participate, 60% successfully returned a usable diary. This amounted to 7,001 diaries.

Editing and weighting the diary data

The completed diaries were shipped to New York for analysis by NOP’s sister company MRI, who had recently conducted a similar survey of their own in the USA.

The diaries first went through an editing process in which logical consistency checks were made on the entries.

There was no demographic weighting of the data, but weights were applied for primary and secondary readership. There was evidence from earlier work that diaries may tend to under-represent out-of-home reading, which tends to accumulate more slowly than in-home reading. Consequently the diary data were weighted so that, for each title, the profile of primary/secondary readership recorded for each title matched the NRS profile of primary/secondary readers.

Creating accumulation curves: (1) minimum base of 130

The basic measure being analysed was ‘first time reading’ (FTR) – that is, a reading event that the diarist recorded as being the first time he or she had read that particular issue. For every ‘first time reading’ event, a record was made of the number of days that had elapsed since that specific issue first came on sale.

130 ‘first time reading’ events (FTRs) was set as the minimum sample size on which to base an accumulation curve. Any publication which achieved 130 or more FTRs would have a curve based solely on its own data. A publication with less than 130 FTRs would use all of its own FTRs and the shortfall below 130 would be made up by a contribution from the curve for the publication group to which it belonged (such as women’s weeklies). Any title with no FTRs of its own would simply use its group curve.

For example, if a publication had 50 FTRs, its curve would be based on those 50 plus the equivalent of 80 FTRs from the group curve. This means each title’s published curve would use all of the available data for that title but would also have the stability of a sample size of 130 minimum.

Creating accumulation curves: (2) defining the axes

The diary data, being very different in kind from the standard NRS readership measures, were not intended to produce readership levels identical to those of the NRS in an absolute sense. Instead, the diary data simply had the objective of being a means of distributing through time the exposures which the NRS was reporting. Consequently the accumulation curves were designed so that on the vertical scale they eventually finished at 100%, where 100% is the NRS readership figure for a given publication.

The horizontal axis for the curves represented the number of days since the title first appeared. Day 1 was defined as the on-sale or appearance day. All curves started at Day 1. In some cases a very small proportion of exposures took place slightly before this, and such exposures were allocated to Day 1.

At the other end of the horizontal axis, cut-off points were set for curves, where accumulation had reached or almost reached 100% and beyond which the number of FTR claims per title were too few and too thinly spread to permit meaningful distinctions between one publication and another. Any small shortfall below 100% at the cut-off point was shown as a single figure which was attributed to the final point on the curve. For national newspapers, the final day was day 8; for newspaper supplements it was day 15; for the television monthly it was day 113 (6 weeks); for all other magazines it was day 182.

Creating accumulation curves: (3) modelling the curves

With the minimum sample set at 130 FTRs, and the scale for the vertical and horizontal axes determined, curves could be plotted.

Curves plotted using the raw data showed some uneven-ness, as expected, because of sampling variation on relatively modest sample sizes. Therefore MRI modelled accumulation curves for all publications and publication groups, thus smoothing out the bumpiness of the raw data, in order to produce curves which represent good predictions of future behaviour.

The modelling used a number of different mathematical formulae, choosing the one that best fitted the raw data for each specific curve.

 
To obtain a copy of the full technical report, email guy@consterdine.com