Estimating the likelihood of the digitally excluded undertaking on-line activities

We analysed data from OxIS relating to the online activities which Internet users in each cluster undertake to estimate the likelihood of non-Internet users undertaking these activities if they were to gain Internet access (as shown in Table 3). So the analysis takes account of the fact that certain groups do less of certain activities digitally because they do less of them off-line as well; those with low incomes, for example, shop less than people on higher incomes.

Table 3. Specific clusters and activities: Average frequency and percentage of Internet users in each group that undertake specific activities

 

Comparing
products & prices

Distance
learning

Looking for jobs
/benefits

Health or
medical care

 

Average
frequency

%age

Average
frequency

%age

Average
frequency

%age

Average
frequency

%age

Internet activity of low income unemployed group compared with other groups

Not unemployed,
High Income

2.7

76

1.4

22

1.8

48

2.2

69

Either unemployed
or low income

2.4

66

1.6

24

1.9

44

2.0

62

Unemployed
and Low Income

1.9

41

1.2

11

2.6

61

1.9

57

Internet activity of the elderly, disabled and retired group compared with other groups

Younger than 65,
not disabled

2.6

75

1.5

22

1.9

51

2.1

68

Either elderly
or disabled

2.5

68

1.5

21

1.6

34

2.1

62

Elderly, disabled

2.3

56

1.0

  0

1.2

  4

2.4

81

Internet activity of educationally disadvantaged, employed group compared with other groups

Higher education,
not unemployed

2.7

77

1.6

27

1.9

51

2.2

71

Either basic education
or unemployed

2.5

70

1.2

13

1.7

41

2.0

62

Basic education,
employed

2.0

49

1.2

   8

2.2

51

2.0

60

Total

2.6

74

1.5

22

1.9

48

2.1

68

Source: OxIS, 2009

Note: other groups are shown for comparison.

Probe. Are the assumptions reasonable - that non-Internet users will, when using they have access, carry out these activities with similar frequency to internet users in the same demographic group? Please comment below.

8 comments on "Estimating the likelihood of the digitally excluded undertaking on-line activities"

Hugh Tollyfield (visitor) said on 15 March 2010 - 12:56pm:

I think it is an assumption to be tested, but I'd like to know the extent to which all groups, when they have access, undertake the listed activities in preference to other on-line activities, such as socioal networking, gaming and entertainment.

Guy Ker (visitor) said on 16 March 2010 - 2:09pm:

The paradigm could be more usefully framed within the context of whether the digitally excluded be persuaded to take up online services. That persuasion should involve a mixture of carrot and stick. Super-serving clients via telephony or High Street offices does not incentivise people to swap modes of take-up. Contrast the private sector where there is heavy incentivisation with deals, offers, freeebies because the organisations have a clear view of the value of getting their clients to switch to online take-up..

Ben Anderson (visitor) said on 23 March 2010 - 10:19pm:

There are a number of well-known methods in econometrics specifically designed to analyse this sort of thing. You question is, to what extent can we predict the behaviour of group A (e.g. low income non internet users) given the behaviour of group B (low income internet users) if something about group A was changed to make them like group B (i.e. give them 'internet' - whatever that means)?

It looks to me like you are assuming that the behaviour observed in group B will occur in group A when they become internet users. But this ignores the fact that the internet users who make up group B might be a particular kind of low income people. They may, for example be less educationally disadvantaged. Or older. So you need to account for this in your prediction of group A's 'new' behaviour. It is not clear that you have done this but it is widely accepted as required in (e.g.) econometric microsimulation analysis of policy intervention scenarios. See http://books.google.co.uk/books?id=jmR6PAAACAAJ&dq=microsimulation+new+f... amongst others.

Since these predicted behaviour rates are crucial to your net benefits calculations it is equally crucial that they are robust. I do NOT think (without testing it) that assuming the rates for 'converted' non-users will be the same as current users is robust. It is more likely that they will be substantially lower - if there were strong incentives for current non-users to engage in these behaviours then they probably would already be doing so.... 'Giving' them access will not change these incentives.

James Stewart (visitor) said on 24 March 2010 - 12:01pm:

It is difficult to estimate these things from this type of statistics. We need a much clearer idea of why specific groups of non-users are not using the internet, and developing the confidence in it. An important factor in those with low education is literacy. When we get down to a group of people who cannot read, then there is nothing about the Internet that will get them to use it, they need reading classes, or they need someone to sit down with to help them. The OXIS does have some of this information - it would be good to see it presented and analysed here.

Anonymous (visitor) said on 12 April 2010 - 12:21pm:

I agree that we need to know why these people don't access the internet. Knowing this may totally change the groups and therefore any predictions as people in different groups may be more similar to each other than to people in their own group.

Knowing the reasons for exclusion will help with any reliability of your prediction. If the reason why some of the educationally disadvantaged don't access the internet is because their type of disadvantage is different from those that do then you clearly can't make this sort of prediction. If the reason why some access it and some don't is down to giving people some help and basic training then you might be able to conclude that the excluded group will access it given the appropriate help.

Anonymous (visitor) said on 16 April 2010 - 9:59am:

This assumption is key to the whole methodology and at the moment feels a bit too much like a leap of faith. In common with others it strikes me that there needs to be a good deal more understanding of specific barriers informing the assumptions in this section.

Carol Moonlight (visitor) said on 19 April 2010 - 8:46pm:

More needs to be known and understood about the barriers for these people. This links back to my other comment about people's skills - lack of them, and of confidence, are a big barrier for some people. And there's combinations. Even if people have the education skills, they still may not have the confidence. Motivation is also really key - and that links in with giving them a reason to access web-based services, through whatever channel. And that in turn means linking it in with what matters in their lives. What is in it for them? What can they get of it that would benefit them and their lives/famailies? Their perspective, their agenda - not anyone else's.

Helen Margetts said on 15 June 2010 - 4:53pm:

Team Response. With respect to how we calculated the likelihood of our digitally excluded groups undertaking these activities, the Oxford Internet Survey (OxIS) data allows us to know what percentage of people in each group who are online undertake these activities on the internet and then assume that similar percentages of those of the group who are offline would do so if they did have internet access. However, we note the several comments which point out that there may be something distinctive about the internet users in each group in comparison with the non-internet users and accept that for this reason, the percentages will not be the same. However, when it comes to the overall financial benefits of each activity being carried out online, you will see that in general we have not assumed that 100 per cent of non-users in each group will 'become' users, but some more modest proportion.