Effects Of Mobile Phones On Telephone Survey Practices And Results

Effects of Mobile Phones on Telephone Survey Practices and Results Vesa Kuusela, Matti Simpanen Statistics Finland Survey Research Unit

During recent years, the telephone coverage and its structure in Finland have undergone a dramatic change, mainly as a consequence of rapid increase in the popularity of the mobile phone. The number of mobile phones has increased since the first half of the 1990s while the number of conventional, fixed phones has decreased simultaneously. The proportion of households with mobile phones only has also been increasing since the mid-1990s (see e.g. Kuusela and Vikki, 1999). The change of the telephone coverage has been very rapid, but not uniform. In fact, the popularity of mobile phones – and conventional, fixed phones as well – varies considerably between population groups: while in some groups 99% of people have a mobile phone, in some others the proportion may only be about 10% . The mobile phone only households in Finland are mainly new ones, but certain other types of households are also switching their fixed phones to mobile phones. There are great differences between population groups in this respect, too. The restructuring of the telephone coverage in Finland seems to continue, and the next generation of mobile phones may bring along new features still. An expected consequence of the changes in the telephone coverage has been an increase in the number of mobile phone interviews in all telephone surveys of Statistics Finland. The most important reason, of course, is the popularity of the mobile phone but it is partly also due to sampling practice: samples are drawn from the Population Register and telephone numbers are subsequently searched for each respondent or household. The proportion of mobile phone interviews has grown steadily for several years and roughly 50% of the CATI interviews at Statistics Finland will be conducted over the mobile phone by the end of the year 2002. The proportion will probably still grow because of the continuing structural change. The mobile phone has several differences when compared to the fixed phone. Maybe the most important is that the mobile phone is a personal appliance and is carried all the time like a wristwatch, whereas the fixed phone is for the whole household and is always kept in the same place. A respondent may be virtually anywhere and in almost any situation when he or she answers the mobile phone. The question was raised whether the popularity of the mobile phone has either indirect or direct impacts on telephone survey practices, or on survey parameters or results, such as survey costs, non-response, estimation, and even factual distributions of research variables. Sampling is another area where significant methodological problems may occur. For example, how do the applied sampling methods comply with the new situation? Do mobile phones influence sampling-related issues if the sample is drawn from, say, an address register or the Population Register; or what if households are the sampling units but the sample frame is a list of telephone numbers or random digit dialling (RDD) is applied? 1

Telephone survey methods have been developed for fixed phones (see e.g. de Leeuw and Collins, 1997, Groves et al.,1988 or Piekarski, 1996). There have been a number of pieces of research and studies concerning the potential effects of fixed phones on interviews. However, very few reports refer to the potential impacts of mobile phones. For example, the bibliography made by Khurshild and Hardeo (1995) had no entries that dealt with mobile phones, and as late as in 1999, Collins only made a brief comment on the growing popularity of mobile phones in the future. Nathan (2001) and Nicolaas and Lynn (2002) recognise the problem mobile phones introduce in household surveys, especially if random digit dialling (RDD) is applied. Fuchs (2000) conducted an experimental survey to compare non-responses in mobile phone and fixed phone surveys. It is not straightforward to ascertain what consequences the extensive use of mobile phones and their unequal distribution among the population have to surveys. At Statistics Finland, the potential effects of mobile phones on the quality of surveys in terms of non-response, duration of interview, etc., has been followed for roughly three years. For example, the type of telephone by which the interview or appointment was done has been recorded in CATI surveys since spring 2000. In addition to the continuous data collection, a specific survey has been carried out three times in Finland (in 1996, 1999 and 2001) to obtain an accurate estimate of telephone coverage and on its structural changes. It should be noted that telephone coverage and its structure, not to even speak about the usage of phones, could be estimated only through a survey designed for this purpose. This report has two parts: the first part consists of a description of the structural changes of the telephone coverage in Finland. Reportedly, the spread of the mobile phone has been one of the fastest in Finland and it is possible that a similar change has taken place, or will take place, in other countries, too. The second part of the report deals with the consequences that the growing popularity of the mobile phone has produced in survey research. Some methodological problems and challenges are also discussed.

Characteristics of mobile phone installations and usage in Finland The installations of mobile phone systems differ from country to country and the applied tariffs are also different. Both influence the popularity and daily use of mobile phones. The differences may be so extensive that direct comparison between countries might be biased.In 1999, nearly 90% of the mobile phones in Finland were privately owned and that has not changed much since. Nearly all mobile phones are subscribed to under monthly contracts. The use of pre-paid packages is very rare (less than 2%). In addition, almost 90% of the mobile phones were listed either to their actual users or to someone else in the same household. Both fixed and mobile phone numbers in Finland are stored in a single database where the numbers are available for everyone at a small charge. In Finland, most operators have a network of their own and all networks are nationwide. There are no hindrances to calling from one mobile phone to another. Calls between the networks are always possible but charges are slightly higher between networks than within one network. Mobile phone numbers have a specific prefix so that the number always tells the type of the phone and the network. In Finland, short messaging service (SMS) is nearly as popular as normal speech communication and there are a lot of different services that are based on this technology.

2

Receiving a call in Finland is always free of charge, both for mobile and fixed phones, i.e. the calling party pays all the costs. However, when a mobile phone is outside Finland the receiver pays half of the roaming costs. Calls from and to mobile phones cost more than calls between fixed phones but calls between two mobile phones are only slightly more expensive than calls between fixed phones, especially within the network of the same operator. There are also a number of different special price arrangements, e.g. for calls between the phones of one household.

3

1. Change of telephone coverage in Finland Figure 1 displays the trends of mobile phone and fixed phone coverages of households in Finland1. The change of the structure started already in the first half of the 1990s, but accelerated considerably in the second half of the decade. Fixed phone coverage has been decreasing since 1993. The popularity of the mobile phone started to increase rapidly two or three years later. The growth accelerated considerably when second generation mobile phones were introduced in the mid-1990s. Their important features were that the new models were considerably smaller and their batteries held their charge longer. Mobile phone penetration was higher than fixed phone penetration for the first time in July 1999 and there is no reason to believe that the previous situation would return. The trends shown in Figure 1 will probably continue. The number of mobile phones will still continue to increase and fewer households will have a fixed phone. The structural change of the telephone coverage from 1999 to 2001 is displayed in Figures 2 and 3. The most noteworthy change has been the growth in the number of households with mobile phones only and the diminished number of households with a fixed phone only. Socially the most important change has been the diminished proportion of households with no phone, only 1.2% in September 2001 (see Figure 3). Obviously, this is close to the theoretical minimum, because 100% coverage is probably not possible: there are always households who do not buy a phone in principle, and some that are in a socio-economic situation where they cannot afford a phone, or are not given a phone.

Figure 1: Trends of mobile phone and fixed phone coverages in Finland.

Figure 2: Structure of telephone coverage of households in May 1999.

The number of mobile phones in Finland is actually much higher than the number of households with a mobile phone because 45% of households have more than one mobile phone (or 70% if one-person households are excluded).Because the mobile phone Figure 3: Structure of telephone coverage of should be mainly considered as a personal appliance, households in September 2001. mobile phone penetration should also be evaluated per person. At the moment, nearly 90% of people in Finland have a mobile phone. In some population groups, mobile phone coverage is even 1

Mobile phone penetration here means that there is at least one mobile phone in the household.

4

higher: e.g. practically all (99%) men and women aged from 20 to 40 have mobile phones of their own.Mobile phone coverage is already so high in Finland that it cannot increase very rapidly any more. However, 18% of the households without a mobile phone in 2001 had considered getting one. One third of them would buy a mobile phone to replace the fixed phone. The penetration of fixed line phones will probably decrease even further. Roughly 20% of the households with a fixed phone in September 2001 had considered giving it up, and the main reason for 34% of them was the fact that they had a mobile phone. (For nearly 40% the main reasons were economic.) Consequently, there are good enough reasons to believe that the trend as shown in Figure 1 will continue for some time. The decrease of the proportion of households without any phone is mainly a results from the increase in the number of mobile phones. This conclusion is based on the fact that in September 2001 nearly 90% of households had at least one mobile phone but in April 1999 the proportion was only 74%. On the other hand, nearly all (96%) of the households without a fixed phone had at least one mobile phone, as shown in the table below. Mobile phone ownership Households ...

Has a mobile Does not have phone a mobile phone

with a fixed phone

85%

15%

100%

without a fixed phone

96%

4%

100%

An obvious reason why mobile phones have lessened the number of households without any phone is that the mobile phone is more convenient for many who did not have any phone earlier, typically people in unstable life situations. The nearly 10% increase (from 21% in 1999 to 30% in 2001) in the proportion of households with only a mobile phone, and the over 10% decrease (from 23% to 10%) in the proportion of households with only a fixed phone can also be regarded as a remarkable change in the telephone coverage. In other words, at the end 2001 only one household out of ten was without a mobile phone but nearly every third household was without a fixed phone.

Figure 4: Proportion of those households who did not own a fixed phone but who once had owned one, i.e. had given up the fixed phone. The rightmost bar is shorter partly because nearly all older households had a fixed phone.

The number of households with a fixed phone has decreased starting from 1993, and in July 1999 more households had a mobile phone than a fixed one. Three different processes have been causing this change: some households simply give up their fixed phone and buy a mobile one, some households moving from one dwelling to another no longer get a fixed phone, and many newly established households do not get a fixed phone at all because they already have one or more mobile phones. In the

5

last case, a fixed phone is a kind of luxury unless it is needed for some other purpose, like Internet access. Nearly half (46%) of the households without a fixed phone had previously owned one but had abandoned it2. The rest (54%) had never owned a fixed phone. Age structure of households partly explains the changes in the telephone coverage. Figure 4 shows how the frequency of abandoned fixed phones increases with the age of the oldest member of the household. Roughly 30% of young people’s households, but only four out of five older people’s households, without a fixed phone had owed one at some point in time. Figure 4 also shows the popularity of the mobile phone by age of the oldest member of household. The fact that young people have never owned a fixed phone indicates that they prefer to get a mobile phone as their primary means of communication and do not get a fixed phone at all.

Structural changes in different households The change of the telephone coverage has been fairly divergent in different types of households. Different population groups have different lifestyles and situations in life in many respects and, therefore, their needs also differ. The importance or status attached to products also varies between people. These differences have some bearing on the kinds of telephones each household has. Figure 5 shows that the most dramatic change has taken place in one-person households: while in 1996, 70% of them had a fixed phone only, in 2001, 80% had a mobile phone and nearly 50% had a mobile phone only. Mobile phone coverage has also increased considerably among other types of households, but they have not given up their fixed phones to the same extent as one-person households have. More than one third of the households in Finland were one-person, or single adult, households in 2001, and their proportion has been growing slowly. There are two very different types of one person households: young adults who have moved from their parents’ home e.g. for studies; and older widows who have not remarried. The lifestyles and living conditions of these two types vary considerably, and they are

Figure 5: Change of telephone coverage from 1996 to 2001 by type of household. The “other households” are similar to families but have more than two adults.

Figure 6: Structure of telephone coverage in single adult households in Finland, by gender and age.

2

The question did not ask whether households had lived in same apartment all the time. Consequently, it is not known how often mobile phones were given up when moving from one dwelling to another.

6

different from other types of households as well. Figure 6 shows how telephone coverage changes by age. Roughly 90% of the youngest people, both males and females, living alone have mobile phones only and none have a fixed phone only. The proportion of mobile phone only households decreases steadily with age, faster with women than with men. In the oldest age groups, fixed phone only households become more frequent. Especially older females living alone rarely have a mobile phone and only very few have a mobile phone only. It is quite obvious that the age structure of a household is connected with the types of telephones it has. Use of the mobile phone is characteristic of young people while few older people have mobile phones. Figure 7 shows the change in the telephone coverage in relation to the oldest member of household. For example, in households where all members were aged under 35, 70% had a mobile phone only and 1% a fixed phone only. Figure 7 also shows that the proportion of Figure 7: Change of telephone coverage in households from households with mobile phones only 1999 to 2001 by age of oldest member of household. Single decreases as the age of the oldest member adult households are excluded. of household goes up. The frequency of fixed phones increases in households with older members. Older people have mobile phones, too, but they are not their only means of communication. There are several causes, which in Finland explain the popularity of the mobile phone on the one hand, and the popularity of the fixed phone, on the other. Probably one of the most determining factors is the stability of the life situation of a household. Figure 8 displays the structure of the telephone coverage by type and tenure of dwelling. Especially the tenure reflects the stability of the life situation. Figure 8 shows that nearly 80% of the households owning their dwelling have a fixed phone irrespective of the type of the dwelling, and that mobile phone only households are relatively rare in owner-occupied dwellings. On the other hand, more than half of the households living in rented dwellings have mobile phones only. It should be borne in mind, however, that these characteristics are not isolated but depend on each other. For example, young people may not yet have permanent dwellings but older people and families have probably lived in the same dwelling for a long time and therefore they have a fixed phone they have not given up. The type of the building where a dwelling is located is, naturally, strongly connected both with the wealth of a household and its life situation. For example, households living in single-family houses are usually fairly wealthy. On the other hand, greater variety of households lives in blocks of flats. Figure 8: Change of telephone coverage from 1999 to 2001 by the type of the building and ownership

7

Figure 9 shows that the proportion of mobile phone households increases as the total household income gets higher but the proportion of mobile phone only households is much higher among low income households. That is, people with lower incomes tend to have only mobile phones rather than fixed ones, whereas affluent people tend to have both. The increase from 1999 to 2001 in the number of the mobile phone only households was also much greater in the lowest income group. However, there has also been a slight tendency among affluent households to give up fixed phones.

Figure 9: The change of telephone coverage in households from 1999 to 2001 by the total income of the household.

In addition to the housing conditions, the region where a household is living influences the popularity of the fixed and mobile phone. In general, the mobile phone is more popular in cities whereas most households in rural areas have a fixed phone. The Appendix contains tables showing the structure of the telephone coverage in Finland in more detail.

8

2. Influence of mobile phones on survey research The changes in the structure of the telephone coverage have naturally many different impacts on society, especially on those areas where telephone operations are significant. In everyday life, the mobile phone has proven very practical in a number of ways and the mobile phone systems are developing very fast. New features and services are introduced all the time, and some of them are very practical while others are fairly imaginative. Apart from the practical improvements to everyday life, the popularity of the mobile phone also demands changes and adaptations in areas in which telephones have been utilised heavily. Telephone survey methods and practices have been created for fixed phones (see e.g. Groves et al., 1988, Piekarski, 1996). The popularity of the mobile phone requires making alterations in both. The scope of the required changes largely depends on local conditions. There are great differences from country to country in the infrastructures of mobile phone installations and tariffs. Consequently, the usages of mobile phones also differ. The effects of mobile phones may be divided into sampling and non-sampling ones. Mobile phones raise the need to make changes in some sampling-related issues, especially in the socalled catalogue frame samples3. Their non-sampling effects are composed of the practicalities of survey work and survey costs. In the CATI centre of Statistics Finland, the proportion of mobile phone interviews has increased steadily, although there is also seasonal variation in this (see Figure 7). In July 2001 already more that 50% of interviews were conducted over the mobile phone. The obvious reason for this is that July is the most popular holiday month by far in Finland and a large number of respondents were not reached by fixed phone at home but by mobile phone at e.g. their summerhouse. Earlier, when mobile phones were not so popular, non-response was much higher during the holiday season that at other times. Now the holiday season seems to be an even better time for interviews than the rest of the year, thanks to mobile phone.

Figure 7: The proportions of mobile phone interviews in CATI centre at Statistics Finland since 1999.

The proportion of agreed appointments over the mobile phone is slightly higher than that of completed interviews. This is a natural consequence because interviewers have been instructed to try to make appointments in many situations when calling mobile phones.

The proportion of interviews completed over the mobile phone varies considerably depending on the age of the respondent. Approximately 83% of the interviews of respondents aged between 20 and 29 but only 23% of those of respondents older that 64 were done over the mobile phone,. Respondents’ educational level also has an influence on this. For example, in the first half of 2002, 55.8% of interviews with respondents with basic level education,

3

Catalogue frame sample here means a procedure where the sample is drawn from an explicit or implicit list of telephone numbers. That is, the primary sampling unit is a telephone number like in random digit dialling.

9

48.9% of those with respondents with middle-level education, and 40.2% of those with respondents with high-level education were done over the mobile phone.

Non-sampling effects The increased number of mobile phone interviews has two immediate consequences to telephone surveys: survey costs go up and survey practices need to be redesigned. The everyday use of the mobile phone is in many ways different from that of the fixed phone. For example, a mobile phone is carried nearly all the time. Therefore, a respondent may answer a call virtually anywhere, which may have an impact on how he or she will answer. Technical problems may impair the interview. Survey costs A call to a mobile phone is more expensive than a call to a fixed phone. Therefore, an increase in the number of mobile phone interviews increases the telephone expenses of a survey. For example, at Statistics Finland the telephone costs of interviewing rose 42% in the three years from 1999 to 2001. However, calling from a mobile phone to another mobile phone is cheaper than calling from a mobile phone to a fixed phone. Therefore, a router4 in the telephone exchange may sometimes moderate the rise. In any case, survey budgets will have to be calculated on a different basis than previously, when most or all interviews were conducted over the fixed phone. Survey practice At Statistics Finland, samples are usually drawn randomly from the Population Register and telephone numbers for the sample units are drawn later from the telephone operator’s database, either in batches or manually. The general practice at Statistics Finland is to record all the numbers listed for a particular household. A telephone number can usually be found for approximately 90% of a sample. Interviewers are instructed to always try to call the fixed phone when a respondent is known to have one. However, respondents are often easier to reach by the mobile phone even when they have a fixed phone. Therefore, interviewers need special instructions for mobile phone interviewing. For example, when a call is made to a mobile phone, the interviewer should always ask whether the respondent is in a place where he or she can speak freely. Even if a respondent agrees to be interviewed, the interviewer should suggest an appointment if the respondent is abroad, because half of the cost of the call will be borne by the respondent, or if the respondent is driving a car or is in a noisy environment or it can otherwise be assumed that the respondent would not be able to concentrate properly. Survey results Analysing whether mobile phones influence survey results, that is, substantial matters, is very difficult. The population groups using primarily the mobile phone or the fixed phone are so segregated that observed differences may either be caused by the type of phone or by the differences between the population groups. Up to now, no reason has been found to support the assumption that the results would be different because of the different technical 4

A router is a device attached to the telephone exchange, which directs a call to a fixed line or to a mobile system basing on the type of the dialled number.

10

interviewing devices. This observation has also been made in earlier studies (see Kuusela and Notkola, 1999). Apart from whether the phone type as such has an impact on the results, it induces indirect effects that originate from the structural differences in the telephone coverage. Figure 8 displays the distribution of answers to a question about the economic situation of the household. The question asked whether the household was able to save money or were running debts. The question was included in the Figure 8: Answers to question, what is the monthly Consumer Barometer Survey and the economic situation of the household by distribution represents the answers of nearly respondents’ educational level and phone type of 4,000 respondents. Only respondents aged interview. Only respondent between 30 to 49 years of age were included. between 30 and 49 were included in order to obtain a sub-sample that was as homogenous as possible with regard to economic situation. In every educational level category, respondents who were interviewed over the fixed phone said more often that they were able to save money. However, the reason obviously was that households in a more stable situation are more often interviewed over the fixed phone because they have them more frequently and are more often at home. Figure 8 shows only one example. However, similar results were obtained in this survey with most economic category variables. Length of interviews A few years ago, when mobile phone devices were technically less developed than they are today, it was suspected that mobile phones would have an effect on the lengths of interviews. The batteries of the first models did not last long and, therefore, respondents might have answered hastily. The fact that respondents could be virtually anywhere was also suspected of having an influence on the lengths of interviews. However, even the first analysis in 1999 failed to reveal any observable differences in the lengths of interviews between phone types (see Kuusela and Notkola, 1999). Later results have confirmed this early finding. Figure 9 shows the interquartile ranges of interview durations in the Consumer Barometer Survey. No consistent differences can be observed. However, the interview was very short, lasting only a few minutes, and the real effects may surface in a much longer interview. The length of interviews is also a difficult topic for research because respondents’ different life situations may mean that they are asked very different sets of questions.

Figure 9: Interquartile rages of interview lengths in Consumer Barometer Survey on consecutive months 2002. The question pattern was practically the same for all respondents

11

Non-response When the popularity of the mobile phone started to increase, it was feared that respondents would refuse an interview by mobile phone more frequently than one by fixed phone. There is no indication of this, however. Although the total refusal rate seems to remain practically unchanged, there is faint evidence of interaction: when contacting young respondents by mobile phone the refusal rate tends to be lower than that obtained with fixed phone. The situation is reversed when contacting older respondents. Apart from refusals, those respondents who previously were difficult to reach (e.g. young men, unemployed people, students, etc.) now have mobile phones and are reachable. The high non-response rates obtained during the holiday season have also improved considerably thanks to the mobile phone. A similar phenomenon could probably also be observed in catalogue frame surveys, although it is not as easy to demonstrate. Fuchs (2000) conducted an experimental study to compare non-response rates in mobile phone and fixed phone interviews. He observed that the refusal rate was slightly lower in mobile phone interviews than in fixed phone interviews but more call attempts were necessary to reach an adequate overall response rate.

Sampling related effects The sampling related problems vary between different sampling methods. If the sampling units are something else than telephone numbers, e.g. persons or households, the main problem is finding a telephone number (or numbers) for each unit. In this case, the pivotal question is how comprehensively mobile phones are listed and whether they are listed to the persons who are actually using them. In general, in such population or household samples mobile phones probably assist the survey work if mobile phones are listed comprehensively. If the sample frame is a list of telephone numbers or numbers are generated, like in the RDD method, both sampling and estimation may become much more complicated than before. In these catalogue frame surveys sampling problems may be difficult and new methodological research will be needed to solve them. Frame coverage The proportions of population with mobile phones only and fixed phones only are segregated: roughly speaking young people in a more or less unstable life situation usually have mobile phones only, while elderly people having a stable life situation have fixed phones only, and middle-aged people with families very often have both. If either of the phone types is left out of the frame, serious undercoverage will be introduced. Undercoverage will be a source of bias in all telephone surveys regardless of the applied sampling method. On the other hand, if both phone types are included, overcoverage is unavoidable. Overcoverage only hampers catalogue frame surveys. A mobile phone is usually a personal and private appliance whereas a fixed phone is for the entire household. Therefore, broadly speaking, sampling from a mobile phone number frame makes a population sample while sampling from a fixed phone number frame results in a household sample. The problem is that the frames overlap for certain parts of the population while for some others they do not, and the frames are different in nature. Therefore, the conventional dual frame approaches are not valid.

12

In some telephone surveys, all members of a household should be interviewed. Mobile phones make this design very difficult because the members of a household may be dispersed. There are practically no means (in a survey organisation) to find out the whereabouts of a mobile phone. The number usually does not indicate where the user of the phone lives or works, and even if it did the phone could still be anywhere. Accordingly, all spatial sampling designs are unfeasible. If the number space is homogenous in the sense that it is not possible to distinguish the phone type from the number itself, estimation and weighting get even more difficult. Inclusion probabilities In catalogue frame surveys, mobile phones will change the inclusion probabilities of households in both population and household surveys. In large households, there are presumably more mobile phones, in addition to a fixed phone, than in smaller households. The inclusion probabilities will increase linearly with the number of phones in the household. Consequently, large households will be overrepresented in samples. The different inclusion probabilities are maybe not a big problem in population samples if population statistics concerning household sizes are available. However, significant intrastrata correlation may cause bias. In household samples, the varying inclusion probabilities may cause serious bias and rectifying it calls for special weighting schemes. The construction of the weighting scheme is not straightforward because of the different statuses of the phones. Profiles of mobile phone users Mobile phones do not present as homogenous a group as fixed phones as to their ownership and usage. In Finland, for example, although most of the mobile phones are privately owned, there are also mobile phones provided by employers to their employees who may use them for private purposes as well. They may be listed to the person or to the company. However, many people have both a mobile phone of their own and a company mobile phone. The owners of private mobile phones are often not their users. For example, the owner might be the head of the household but the user is a child. The phone may be listed either to the owner or to the actual user. Many mobile phone users are young, some very young (under 10 years). Therefore, it has to be confirmed that the respondent is eligible if it is not known who he or she is. So-called pre-paid mobile subscriber packages are popular in many countries. This means that the phone number is only valid for the period the prepaid fee covers. In some cases, it is possible to pay to keep the number operational but in others the number is only valid for as long a period as the paid fee covers.

13

Other consequences Popularity of the mobile phone may be a hindrance to the spreading of the Internet because in most cases feasible Internet access requires a fixed line. The situation can be observed in telephone surveys. Figure 10 shows results obtained from nearly 10,000 telephone interviews at Statistics Finland. In the first half of 2002, respondents who were interviewed over mobile phones had fewer home computers and Internet connections than those who were interviewed over fixed phones. The difference was the most prominent when respondents had middle level education.

Computers and Internet access in households by respondents' educational level and phone type of interview Basic level education Fixed phone Mobile phone Middle level education Fixed phone Mobile phone High level education Fixed phone Mobile phone 0

10

20

30

40

50

60

70

80

90

% of interviews Computer with Internet access

Computer without Internet access

Figure 10: The frequency of computers and Internet access in households in Finland (2002) shown in relation to respondents’ educational level and the type of the phone by which the interview was conducted.

3. Conclusions and remarks Popularity of the mobile phone varies considerably from country to country. For example, according to an Oftel report (2002), 68% of households in the United Kingdom (UK) had a mobile phone and the proportion had remained about the same for some time. In addition, 70% of mobile phone users in the UK have prepaid packages. Nicolaas and Lynn (2002) reported that only 5% of households in the UK had mobile phones only and 93% had fixed phones. However, most countries do not have accurate figures concerning telephone coverage, which makes it difficult to estimate how widespread problems mobile phones introduce. Mobile phones are an inherent part of infrastructure in most developed countries and some underdeveloped countries may leap over the landline technology and assume mobile phones. In some countries the change has been very rapid while in some others, such as the United States, it has been slower mainly because of tariff policy. Mobile phones are becoming more popular in the USA, albeit not very rapidly, and they are usually not included in RDD samples as yet (see e.g. Jenkins, 2001). In most European countries, mobile phones are probably already so popular that they should be taken into account in telephone surveys. The structures of the telephone coverages and the installations and uses of mobile phone systems differ from each other in most countries and therefore a uniform methodology may is not yet be feasible. To gain adequate knowledge raises the need for special surveys in which these issues are elucidated. The change of the structure of the telephone coverage has some unavoidable consequences to telephone surveys. When mobile phones become more popular, the number of mobile phone interviews grows automatically. An immediate consequence is that the telephone costs of surveys go up. Although technical solutions may facilitate the situation, mobile phone calls are still more expensive than calls between fixed phones. Survey practices need redefining if respondents are expected to have mobile phones. Apart from general guidelines for the whole interview survey unit, interviewers especially need 14

guidelines on how to proceed if the respondent answers a mobile phone in an unusual situation. For instance, what should the interviewer do if the respondent is driving a car? There are no indications that refusals would be more frequent in mobile phone contacts than in fixed phone contacts (see also Fuchs, 2000). In fact, mobile phones probably boost the response rates in telephone surveys because of the increased contact possibilities. However, sampling methods and local conditions, e.g. how calls are charged, may generate an opposite impact. As such, the type of phone the respondent is using while giving an interview probably have no impact on survey results or lengths of interviews. Further research on this topic is still needed, however. Indirect effects, on the other hand, are obvious because of the pronounced differences between the users of different phone types. However, some changes in survey results may emerge if the sampling frame is suddenly changed in a catalogue frame survey. Typically, the most deprived groups in society who previously had no phones now have mobile ones. Consequently, if mobile phones are included in frames there may emerge new groups of respondents who were previously excluded from telephone surveys, thus causing differences in comparisons to earlier results. The future of telecommunications might be very different from what we know now. In Finland, as well as in many other countries, it is possible that speech communication will move mainly via mobile phones in the future while fixed phone lines will be used mostly for other purposes, like Internet and email. There are many reasons why this scenario is plausible in general, and young people in Finland already consider the mobile phone as their principal means of voice communication. One of the most important factors is that mobile communication technology develops very fast, whereas there has been practically no development in fixed phones for many years. Mobile phones already have a lot of functions that will not be available for fixed phones, and many new – and imaginative – functions will become available in the future. Small, practical things, e.g. cheaper calls between two mobile phones than from a mobile to a fixed phone, also put pressure on this development. Finally, when the critical mass in mobile phone usage is reached, it will be more or less futile to use fixed phones. It may be that even greater changes are ahead when the next generation mobile phones become popular. They may have features that have a greater impact than today’s mobile phones on surveys. With third generation mobile phones, the line between a mobile phone and a palmtop computer will be very thin. It is possible that they will have instant Internet and email access in addition to enhanced telephone features. When they become popular, survey practices may be confronted with much bigger challenges than those presented by today’s’ mobile phones.

15

References Collins M (1999). Editorial: Sampling for UK Telephone Surveys. Journal of the Royal Statistical Society, A, 162, Part 1, pp. 1-4 Fuchs, M. (2000): Non-Response in a Cellular Phone Survey - Experiences from a Comparison with a regular CATI Study. International Sociological Association, Research Committee 33: Methodology (RC 33), Fifth International Conference, Cologne, Germany. Groves R. M., Biemer P. P., Lyberg L. E., Massey J. T., Nicholls W. L. Waksberg J. (Eds. 1988). Telephone Survey Methodology. New York: Wiley. Jenkins V. (2001): The Impact of Mobile Phones on Sampling. The Frame, September 2001. Survey Sampling Inc. (http://www.worldopinion.com/the_frame/2001/sept_1.htm) Khurshid A, Hardeo S (1995). A Bibliography on Telephone Survey Methodology. Journal of Official Statistics, Vol. 11, No. 3, pp. 325-367 Kuusela, V., Notkola, V. (1999): Survey Quality and Mobile Phones. International Conference on Survey Nonresponse, Portland, USA. Kuusela, V., Vikki, K. (1999): Change of Telephone Coverage due to Mobile Phones. International Conference on Survey Nonresponse, Portland, USA. De Leeuw, E. D. & Collins, M. (1997) 'Data Collection Method and Data Quality: An Overview' in L. E. Lyberg, P. Biemer, M. Collins, C. Dippo, N. Schwarz, & D. Trewin (In) Survey Measurement and Process Quality. New York: Wiley. Nathan, G. (2001): Telesurvey Methodologies for Household Surveys – A Review and Some Thoughts for the Future. Survey Methodology Nicolaas, G., Lynn, P. (2002): Random-digit dialing in the UK: viability revisited. J.R.Statist.Soc. A, 165, Part 2, 297-316. Oftel (2002): Consumers’ use of mobile telephony; Q9 (http://www.oftel.gov.uk/publications/research/2002/q9mob_r0702.htm) Piekarski L. (1996): A Brief History of (http://www.worldopinion.com/reference.taf?f=refi&id=1252)

Telephone

May

2002.

Sampling.

16

APPENDIX Table 1: Structure of telephone coverage of all households in Finland, September 2001 By the type of the households

Telephone coverage of households N All households

Mobile phone only %

Both phones

Fixed phone only

No phone

%

%

%

2359272

30.6

58.1

10.1

1.2

899802 731665

49.4 21.8

30.6 68.6

17.4 9.3

2.7 0.4

72305

34.8

61.4

3.2

0.6

Family with children

413507

17.6

81.0

1.3

0.0

Other

234639

7.9

89.0

3.0

0.1

Less than 18 years

695668

21.0

75.6

3.2

0.2

19-29 years

424100

44.2

46.8

8.1

0.9

30-49 years

549635

40.0

44.5

13.0

2.5

50-64 years

552296

25.5

59.5

13.7

1.3

65 + years

135726

20.2

52.7

25.9

1.1

Less than 35 years

587489

50.5

38.4

10.0

1.2

35-44 years

474907

32.2

59.4

6.5

1.9

45-54 years

613670

24.7

67.4

6.9

1.0

55- 64 years

438960

19.9

67.9

11.3

0.8

65 + years

241023

13.4

62.2

23.5

0.8

77719

77.3

11.4

5.3

5.9

370 - 669 €

184275

59.2

17.8

18.3

4.7

670 - 999 €

189691

33.9

31.5

32.7

1.9

1000 - 1669 €

360427

37.2

46.4

16.1

0.2

1700 - 2499 €

402327

35.5

58.4

5.5

0.6

2500 - 3360 €

316773

23.6

72.9

3.6

0.0

3361+ €

413060

12.5

85.7

1.6

0.2

Does not tell/know

415000

20.5

68.0

9.9

1.7

Household type Single adult Two adults One supporter

Age of youngest member in household

Age of oldest member in household

Total monthly income of household 0- 369 €

Table 2: Structure of telephone coverage of all households in Finland, September 2001 by region and type of dwelling

Telephone coverage of households N All households

Mobile phone only

Both Phones

Fixed phone only

No phone

%

%

%

%

2359272

30.6

58.1

10.1

1.2

1430769

36.0

53.9

9.2

1.0

Other urban area

476133

26.7

61.4

10.4

1.5

Sparsely populated area

450318

17.3

68.3

13.0

1.3

2052

75.0

0.0

0.0

25.0

Apartment building

986538

45.5

42.2

11.2

1.1

Row or semidetached house

394922

31.0

60.1

8.3

0.6

Single family house

941460

14.4

74.9

9.9

0.8

32846

40.0

35.0

7.0

17.9

3506

58.5

5.0

21.8

14.6

Owned apartment

1517863

14.5

73.2

11.8

0.4

Rented apartment

779044

60.2

31.0

6.9

1.8

Dormitory, etc.

14500

69.0

8.8

4.6

17.7

Other

28206

46.0

39.5

3.6

10.9

Do not know or tell

13765

27.8

33.8

27.2

11.2

Helsinki metropolitan area

433849

28.8

59.9

10.2

1.0

Other Southern Finland

191029

30.8

59.8

8.4

1.1

Western Finland

670785

31.9

56.0

11.0

1.1

Eastern Finland

455108

28.7

59.2

10.8

1.4

Middle Finland

325608

29.5

60.2

8.9

1.4

Northern Finland

261129

33.6

55.4

10.2

0.8

21587

43.8

49.5

3.2

3.5

Location of dwelling In city

Do not know or tell Housing type

Other Do not know or tell Type of dwelling

Major region in country

Åland

Table 3: Structure of telephone coverage in different households in Finland, 2001 Excluding single adult households

Telephone coverage of households (excluding single adult households) All households with more than one member

N

Mobile phone only

Both Phones

Fixed phone only

No phone

%

%

%

%

1459470

19.0

75.1

5.7

0.2

None

873440

19.6

71.6

8.4

0.3

One

269807

21.8

76.4

1.7

0.2

Two

214187

15.8

82.3

1.8

0.1

3-4 children

89842

12.9

86.6

0.5

0.0

5+ children

7644

17.3

82.7

0.0

0.0

Less than 18 year

628465

18.4

80.1

1.5

0.1

19-29 years

247115

42.0

56.9

1.1

0.1

30-49 years

196178

18.5

72.7

8.2

0.6

50-64 years

311186

6.4

82.6

10.7

0.3

65 + years

74679

2.4

68.4

28.6

0.7

Less than 35 years

270968

54.5

44.8

0.6

0.1

35-44 years

291766

18.4

79.7

1.7

0.3

45-54 years

415652

11.4

85.9

2.7

0.1

55- 64 years

297885

7.2

85.2

7.3

0.3

65 + years

179976

3.7

71.9

23.8

0.6

5899

88.5

11.5

0.0

0.0

370 - 669 €

23706

53.5

36.5

5.7

4.2

670 - 999 €

54772

33.1

44.6

21.3

0.9

1000 - 1669 €

163948

24.9

60.8

14.2

0.2

1700 - 2499 €

238680

25.0

69.9

5.0

0.1

2500 - 3360 €

258291

19.4

77.4

3.2

0.0

3361+ €

389975

11.3

87.3

1.4

0.1

Does not know or tell

324199

14.5

78.6

6.5

0.4

Number of children under 18 years

Age of youngest member in household

Age of oldest member in household

Total monthly income of household 0- 369 €

Table 4: Structure of telephone coverage by region and type of dwelling Excluding single adult households

Telephone coverage of households (excluding single adult households) All

N

Mobile phone only

Both Phones

Fixed phone only

No phone

%

%

%

%

1313801

19.0

75.1

5.6

0.2

In city

798240

22.3

72.5

5.0

0.2

Other urban area

321207

17.9

77.4

4.6

0.2

Sparsely populated area

340023

12.4

78.9

8.3

0.4

Apartment building

409413

32.7

61.0

5.9

0.4

Row or semidetached house

268211

23.4

73.1

3.4

0.0

Single family house

770631

10.2

83.3

6.3

0.2

10787

22.1

73.3

2.3

2.3

428

0.0

41.4

58.6

0.0

Owned apartment

1121314

10.4

83.5

5.9

0.1

Rented apartment

320422

48.9

45.9

4.5

0.7

649

38.7

38.7

22.7

0.0

10764

25.3

70.1

4.7

0.0

6070

12.4

68.1

19.5

0.0

Helsinki metropolitan area

275332

17.7

75.8

6.3

0.2

Other Southern Finland

119209

17.5

76.4

6.1

0.0

Western Finland

400947

19.5

74.0

6.2

0.3

Eastern Finland

289409

18.5

76.6

4.7

0.3

Middle Finland

202488

20.1

74.5

5.2

0.2

Northern Finland

160068

20.9

73.4

5.4

0.3

11840

19.2

77.2

1.4

2.1

Location of dwelling

Housing type

Other Do not know or tell Type of dwelling

Dormitory, etc. Other Does not know / tell Major region in country

Åland

Table 5: Structure of telephone coverage in single adult households by personal characteristics

Telephone coverage of single adult households N All

Mobile phone only

Both phones

Fixed phone only

No phone

%

%

%

%

899802

49.4

30.6

17.4

2.7

Male

504792

51.6

29.2

16.1

3.2

Female

395010

46.5

32.3

19.1

2.1

- 25 years

181602

46.3

33.3

18.9

1.4

25-35 years

155439

49.2

30.7

17.5

2.6

35-45 years

179550

52.9

28.9

13.7

4.6

45-55 years

201609

52.7

28.0

16.5

2.8

55-65 years

130302

46.5

32.7

18.9

2.0

51300

43.0

31.0

24.0

2.0

71820

76.4

11.4

5.7

6.4

370 - 669 €

160569

60.1

15.0

20.1

4.8

670 - 999 €

134919

34.2

26.2

37.3

2.3

1000 - 1669 €

196479

47.5

34.5

17.8

0.3

1700 - 2499 €

163647

50.8

41.7

6.3

1.3

2500 - 3360 €

58482

42.1

52.6

5.3

0.0

3361+ €

23085

33.3

57.8

6.7

2.2

Does not tell

90801

41.8

29.9

22.0

6.2

158517

48.2

32.4

16.8

2.6

71820

52.9

32.1

12.1

2.9

Western Finland

269838

50.4

29.3

18.1

2.3

Eastern Finland

165699

46.4

28.8

21.4

3.4

Middle Finland

123120

45.0

36.7

15.0

3.3

Northern Finland

101061

53.8

26.9

17.8

1.5

9747

73.7

15.8

5.3

5.3

Gender

Age

65+ years Total monthly income of household 0- 369 €

Major region in country Helsinki metropolitan area Other Southern Finland

Åland

Table 6: Structure of telephone coverage in single adult households by region and type of dwelling

Telephone coverage of single adult households All Single adult households

N

Mobile

Both Phones

Fixed

No phone

%

%

%

%

899802

49.4

30.6

17.4

2.7

In city

632529

53.3

30.3

14.4

1.9

Other urban area

154926

45.0

28.1

22.5

4.3

Sparsely populated area

110295

32.6

35.8

27.4

4.2

2052

75.0

0.0

0.0

25.0

Apartment house

516375

54.6

28.9

14.9

1.6

Row or semidetached house

113373

47.0

32.4

18.6

2.0

Single family house

152847

33.3

37.2

25.8

3.6

19737

48.8

16.3

9.3

25.6

2754

66.7

0.0

16.7

16.7

Owned apartment

396549

26.1

44.1

28.6

1.2

Rented apartment

458622

68.1

20.7

8.6

2.6

Dormitory, etc.

13851

70.4

7.4

3.7

18.5

Other

17442

58.8

20.6

2.9

17.6

EOS

7695

40.0

6.7

33.3

20.0

158517

48.2

32.4

16.8

2.6

71820

52.9

32.1

12.1

2.9

Western Finland

269838

50.4

29.3

18.1

2.3

Eastern Finland

165699

46.4

28.8

21.4

3.4

Middle Finland

123120

45.0

36.7

15.0

3.3

Northern Finland

101061

53.8

26.9

17.8

1.5

9747

73.7

15.8

5.3

5.3

Location of dwelling

Do not know Housing type

Other Does not know / tell Type of dwelling

Major region in country Helsinki metropolitan area Other Southern Finland

Åland