Effectiveness of actions to reduce harm from nuisance calls in Scotland

Annex H: Comparison of UK data from Ofcom and trueCall

H.1 Ofcom surveys

Ofcom conducts surveys designed to have respondents who are representative of the UK in terms of gender, age group, socio-economic group, working status and region of residence. In Figure 43 we identify those of most relevance to this project.

Figure 43: Relevant surveys by Ofcom

Survey	Frequency	Technique
Landline nuisance call (diary survey)	January-February every year since 2013	A sample of about 800 people keep diaries in which they record details about every nuisance call that they receive in four successive weeks
TPS effectiveness (diary survey for a randomised control trial in two waves)	November 2013 and March 2014 only	This resembled the landline nuisance call diary surveys, except that the 800 or so diarists that participated in both 2013 and 2014 were not registered with TPS before the 2013 survey and, unknown to them, half of them were registered before the 2014 survey
Consumer issues (omnibus survey)	January, May and September (and formerly March, July and November too) every year	A sample of about 1,000 people is interviewed as part of an omnibus study (potentially asking about other topics besides telecommunications) to determine their experience of, and attitudes to, telecommunications problems such as nuisance calls
Technology tracker (omnibus survey)	January-February and July-August every year	A sample of about 2,000 people is interviewed to determine whether and how they use telecommunications services and devices

H.2 The proportion of people receiving nuisance calls

The sources of data from Ofcom that are most directly concerned with nuisance calls are the consumer issues surveys and the landline nuisance call surveys (along with the TPS effectiveness surveys of 2013-2014). The consumer issues surveys rely on omnibus interviews in which interviewees recall face-to-face whether they have received nuisance calls in the previous four weeks. The landline nuisance call surveys and the TPS effectiveness surveys rely on diary entries for which respondents keep diaries over four weeks.

The consumer issues surveys persistently find fewer nuisance calls than do the landline nuisance call surveys and the TPS effectiveness surveys ^[132] : typically the proportion of interviewees that recall receiving nuisance calls is roughly 3/4 of the proportion of diarists that report receiving nuisance calls. Possibly this indicates that low numbers of nuisance calls pass unnoticed, or at least unremembered. Figure 44 supports this possibility: in the landline nuisance call surveys the proportion of diarists that report receiving at least three nuisances calls ranges between 63% and 69%, so it is usually close to the proportion of interviewees that recall receiving nuisance calls.

Figure 44: Proportions of users receiving nuisance calls in Ofcom surveys

Year	Proportion of adults with landlines that reported receiving nuisance calls in four weeks…
	During the landline nuisance call diary survey, with the number of nuisance calls received being at least…			In a consumer issues omnibus survey…
				Just before the diary survey	Just after the diary survey
	1	2	3
2017	81%	71%	63%	55%	54%
2016	84%	73%	64%	59%	64%
2015	87%	78%	69%	63%	68%
2014	85% [133]	76% [134]	68%	62%	66%

The landline nuisance call surveys themselves may be affected by human nature. As Figure 45 illustrates, in every year there is a decrease in the number of nuisance calls reported in the surveys after the first week, at a season of the year when evidence (from the consumer issues surveys and standard trueCall units) suggests that there should be an increase. If there was not simply week-by-week variation perhaps the diarists became less conscientious after the first week.

Also, the diary can hold details of only 40 nuisance calls; anyone getting more than that is asked to record them on separate sheets or request another booklet, but this too could lead to under-reporting ^[135] . A diarist may also just overlook noting call details, especially when busy with something else, in a different room from the diary, or getting a series of calls.

Figure 45: Weekly variations in Ofcom landline nuisance call surveys

Year	Proportion of users that reported receiving nuisance calls for successive weeks numbered…				Mean number of nuisance calls received in one week for successive weeks numbered…
	1	2	3	4	1	2	3	4
2017	31%	26%	24%	25%	1.9	1.7	1.6	1.5
2016	30%	26%	27%	24%	2.1	2.0	1.9	1.8
2015	34%	28%	28%	25%	2.5	2.1	2.1	1.9
2014	35%	29%	25%	25%	2.1	2.0	1.8	1.8
2013	30%	24%	24%	22%	2.0	1.7	1.7	1.7

Despite their limitations the consumer issues surveys have been useful, because by occurring frequently and considering mobile phones as well as landlines they can provide an impression of trends, as illustrated by Figure 46. Of course, recent publicity about nuisance calls, as well as their actual level, may affect how many nuisance calls people remember receiving. The network measurements and user device measurements that are now becoming available should provide more objective ways of detecting the trends. They should also avoid some of the problems faced by the landline nuisance call surveys.

Records from standard trueCall units can provide figures analogous to those from the landline nuisance call surveys and consumer issues surveys given in Figure 44. We discuss below how these figures (or indeed other figures derived from network measurements or user device measurements) might be related to those other figures.

H.3 The frequency of nuisance calls

Figure 46 uses the Ofcom consumer issues surveys to show evidence of a seasonal variation in the number of nuisance calls, which rises towards the beginning of the year and falls towards the end, with the highest sometimes being 23% more than the lowest over six months. The variation is more marked for some types of nuisance call (such as automated marketing calls) than for others (such as abandoned calls).

Figure 46: Percentages of adults receiving nuisance calls in four weeks

Source: Ofcom consumer issues survey reports in updates to the joint ICO/Ofcom action plan ^[136]

The Ofcom landline nuisance call surveys occur only once per year, so they are unable to show a seasonal variation in the number of nuisance calls like that in Figure 46. However, trueCall units can do so, as demonstrated in Figure 47.

There is also a pronounced day-of-week variation, as recorded by the landline nuisance call diary surveys and trueCall data ^[137] . It is exhibited very clearly for one calling number in Figure 66.

Figure 47 distinguishes between trueCall “standard” units and trueCall “vulnerable” (or “secure”) units. Both the standard units and the vulnerable units admit calls from callers on the white list, reject calls from callers on the black list and intercept calls from callers on neither list. The standard units tell callers on neither list to say their names if they want the calls to proceed; the vulnerable units tell them to press a key, provide a code already assigned to them or contact someone else (such as a relative, carer, neighbour or warden). The vulnerable unit users are likely to be vulnerable people and typically had their units supplied free of charge by their local authority or other care agency. The different numbers of nuisance calls received by the standard units and the vulnerable units provide evidence that vulnerable people are especially targeted by nuisance callers.

Figure 47: Nuisance calls per trueCall unit per month, UK, 2014-2017

Even the standard units report higher numbers of nuisance calls than do the landline nuisance call surveys: the proportions of users receiving higher numbers of nuisance calls are higher for standard trueCall units than for Ofcom landline nuisance call surveys, with the effect that the mean number of nuisance calls received is much higher for the standard units report than for the landline nuisance call surveys. Figure 48 consolidates the reports for 2014-2017 to illustrate this.

Figure 48 Proportions of landline users receiving given numbers of nuisance calls, 2014-2017

Figure 49 shows this in another way ^[138] .

Figure 49: Summary distribution of landline nuisance calls per user, 2014-2017

Data source	Proportion of recipients receiving in four weeks a number of calls in the range…									Mean number received in four weeks
	0	1-5	6-10	11-15	16-20	21-25	26-30	31-35	36-40
Ofcom landline nuisance call surveys	15%	37%	22%	11%	7%	3%	2%	1%	0%	7.6
Standard trueCall unit records	9%	16%	15%	13%	10%	8%	5%	5%	4%	18.0

H.4 Comparisons between sources of data

The consumer issues surveys and landline nuisance call surveys are designed so that the respondents are representative of UK adults in terms of gender, age group, socio-economic group, working status and region of residence. By contrast, the owners of standard trueCall units are not representative, as:

• They are motivated to buy the units, perhaps by receiving high numbers of nuisance calls.
• They could be disproportionately in older age groups (as suggested by a small trueCall user survey in 2013), so they would encounter disproportionate numbers of nuisance calls.
• They are not spread uniformly throughout the UK, but there are regional variations in nuisance call patterns (as indicated in Annex K).
• They can accidentally or deliberately fail to activate the central collection of records.

However, related comments apply to the landline nuisance call surveys. For instance:

• The respondents might be motivated to participate either by receiving very few nuisance calls (in which case any participation fee would be easily earned) or by receiving very many nuisance calls (in which case anything that might reduce the nuisance could be welcomed).
• The respondents can be mistaken or forgetful, and, as demonstrated by Figure 45, some appear to be less conscientious in reporting nuisance calls after the first week of a survey.
• The surveys use samples that are not large enough to permit detailed inferences about cross-tabulated groups (such as the elderly in Scotland).
• The surveys occur at particular times of year, but there are seasonal variations in nuisance call patterns.

In any event, counting nuisance calls by using standard trueCall units (or indeed by using any equipment that intercepts them before the users receive them) is likely to produce a higher number than counting them by using the diarists of the landline nuisance call surveys, who need to be at home to receive them. In fact the diarists count fewer calls if they go out to work; Figure 50, based on the 2017 landline nuisance call survey, illustrates this. In it, the mean number of nuisance calls in four weeks per diarist who is not working (8.4) or even per diarist who is retired (10.1) is still well below the figure obtained from standard trueCall units in Figure 49. However, many of the diarists share their homes with other adults who are likely to receive some calls, even if not as many as the diarists, so the average number of nuisance calls per household will be higher than the number of nuisance calls per diarist. The assumption made in Annex C is that the diarists receive only the proportions of the nuisance calls due according to the sizes of their households (so in a two-person household the diarist would receive half the nuisance calls), but this is questionable.

Figure 50: Landline nuisance calls received in four weeks, by working status

	Working status						All
	Full time employed	Part time employed	Home-maker	Student	Tempor-arily out of work	Retired	Working	Not working
Mean number of nuisance calls in four weeks per diarist	5.3	6.5	6.9	4.4	6.7	10.1	5.6	8.4
Proportion of diarists receiving nuisance calls in four weeks	76%	82%	85%	60%	82%	92%	78%	85%

Source: Ofcom landline nuisance call surveys

Understanding how the results from the landline nuisance call surveys and those from the standard trueCall unit records differ requires us to look at the distributions, not just the means. With suitable c information about these distributions, we could, for example, infer the distribution of nuisance calls to a two-person household, given the distribution of nuisance calls to a diarist. We could take similar account of the age groups of trueCall unit purchasers and of the difference between the first week and the subsequent weeks of the landline nuisance call surveys. Our explorations so far show that several factors need to be taken into account when resolving the differences; for instance, just considering working status or age group on its own is insufficient.

We do not have adequate data to determine these distributions. We note, however, that the distribution of the landline nuisance call surveys appears to align best with that of the standard trueCall unit records if it is offset by one, two or three nuisance calls ^[139] ; this amounts to assuming that people are motivated to buy standard trueCall units only if they receive at least one, two or three nuisance calls in four weeks (on average). However, the alignment between the distributions remains poor even with this offset.