Stats entertainment

Category Archives: Quantitative

Working in a low-paid dead-end job in a call centre, living with the parents and approaching 30 an old friend of mine made a decision – to pack their bags and head for New Zealand on a 12 month work visa. This was a few years back and they haven’t looked back, moving after their first year to Australia.

And they’re not the only one. Recently I’ve noticed more and more people have been following in this intrepid friends footsteps, in fact rarely a day goes by without hearing about some other person moving to Australia. But what do the stats have to say?

My first stop is the official statistics produced by the Australian Department of Immigration and Citizenship. What I’m interested in particularly is the numbers of working holiday visas – these are the ones granted to young people aged 18-30 and which last 12 months.

graphAus

And here it is:

Though the figure has fallen slightly for 2009/10 the trend over the period shows a gradual increase suggesting that in the past few years an increasing number of young people are taking the opportunity to leave the UK for Australia using a Working Holiday visa.

Rather interestingly the graph shows the peak period for the number of Working Holiday visas granted was in 2008/09 coinciding with the recession. During this period some 40 182 Working Holiday visas were granted to UK citizens.

Part of the explanation may be the rising popularity of a ‘gap-year’, but as the visas are open to all people aged for 18-30 there may be another group taking increasing advantage of the Working Holiday visa programme;  Twentysomethings who lack ties such as a family or mortgage (due to their unaffordability) and who find career aspirations unfulfilled, either through unemployment or the growth of the dead-end job and who therefore find both possible and attractive the prospect of leaving the UK.


It’s easy to knock the project of to measure the nations well-being – indeed it’s been variously dubbed pointless, a waste of money or an exercise in stating-the-bleeding-obvious whilst there have also been questions about how reliable a measure of well-being asking a person to gave a rating between 0 – 10 is.

This graph however, puts the whole project into context:

Source: ONS

Source: ONS

The graph shows what appears to be a strong correlation (R = -0.88) between the suicide rate and reported levels of life-satisfaction when these are organised in age categories.

Source ONS:Life Satisfaction ratings 0-10 where 0 in not at all satisfied and 10 is completely satisfied. Suicide rate = suicides per 100 000 population

Source ONS:
Life Satisfaction ratings 0-10 where 0 in not at all satisfied and 10 is completely satisfied. Suicide rate = suicides per 100 000 population

What this suggests is a spectrum of well-being; In terms of age for groups where the overall average level of life satisfaction is lower rates of extreme emotional distress, as indicated by suicide, are correspondingly higher.

The importance of well-being therefore cannot be overstated.


Just over a month ago I went for a job at a market research agency. Talking over my CV the interviewer noted that I’d been working as a telephone interviewer and asked if the job simply involved reading the words which appeared on screen. If only….

A common conception, even among some interviewers, is that in a CATI system the role of the interviewer is no more than to run, machine like, through a series of questions exactly as they appear on screen. Indeed as May (1997) neatly summarises:

The theory behind this method is that each person is asked the same question in the same way so that any differences between answers are held to be real ones and not the result of the interview situation itself

But this does this necessarily mean that the interviewer is a wholly passive actor? Carrying out my first interview, just over a year ago, I quickly learnt that the role of the interviewer is far more complex than I’d ever imagined, in fact it’s not too much of a stretch to say that it’s really an art form…

1.) The interviewer must persuade:

The first job of the telephone interviewer is to persuade a member of the sample to take-part in the survey. This is something which is becoming more challenging for a whole multitude of factors including household composition and the use of mobile phones with a caller display, but two key factors are survey fatigue – there are now simply far too many surveys competing for peoples attention and secondly a suspicion of telephone surveys resulting from unsolicited sales calls purporting to be a survey.

This is made especially challenging when there is no incentive on offer for participation and the interviewer here must use all their powers of persuasion, assuaging any concerns over details such as confidentiality whilst emphasising the benefits of the survey. The interviewer must strike up an instant connection with whoever it is who happens to have picked up the phone and Listening to experienced interviewers I am often in awe of their skill in this area.

2.) The interviewer must be engaging

Once on the hook it is up the the interviewer to keep the respondent engaged, not necessarily an easy task over 20, 30, or even 40 minutes of interviewing, which may at times be repetitive and tough-going.  An interviewer must interpret and use various cues to judge the mood of the respondent; are they sighing, glazing-over, becoming agitated, or distressed? The interviewer can then make minor adjustments to the tempo of the interview, or to try to re-engage the respondent. Satisficing is a real danger in the telephone interview and whilst online surveys have turned to graphics and more recently ‘gamification’ to keep respondents engaged the telephone interviewer must rely solely on their charisma and mastery of communication.

My experience has also shown me that I get more from a respondent if I offer a little bit of myself. For example a respondent may have a crying baby and apologise for the noise and I’ll respond that it’s not a problem at all and mention that I have a baby too. This must be kept professional of course, and it’s important not to get over-familiar as this can then affect the data, as McNeill (1990) states

Interviewers have to strike a careful balance between establishing the kind of relationship with respondents that will encourage them to be frank and truthful, and avoiding becoming too friendly so that respondents try hard to please.

The skill of the interviewer is to manage the relationship with each respondent to maintain this balance, though I can say that I have turned-round some very difficult interviewers by displaying that I am a human, rather than a robot.

3.) Knowledge of the interview

The telephone interviewer must have a good knowledge of the survey and the various rules governing it for example how to record a , or respondents may ask for clarification and the interviewer needs to provide this without biasing the data. Respondents may also ask about aspects of the survey such as technical questions about sampling, or most commonly the question ‘so what is this all for?’  Giving a good explanation of the purpose of the survey, and emphasising the importance of their contributions  makes respondents feel more valued and more likely to take part in subsequent waves, or even other research. For me this is a basic ethical requirement and one in which interviewers make an essential contribution as the ambassadors of social research.

4.) The interviewer as Louis Theroux

Finally, in many surveys there will be open questions which can call on the interviewer to probe, being careful not to lead the respondent  When probing I often feel like Louis Theroux as, when needed, I use the tactic of pretending to know less than I do in order to elicit a more detailed response.  When probing a persons occupation  for example I’ll say “oh, I’m terrible at finance, my brother works in that field and I still can’t understand what he does after 10 years.” I will then find that even a previously cagey respondent will go out of their way to explain their job in simple terms.

Interviewers – More important than you think

When it comes to data, interviewers are the unsung heroes. They are to statistics what swashbuckling fossil hunters were to the science of natural history – getting out there and collecting the raw material for the scientists to test and interpret. It is primarily the skills of the interviewer determine the quality of data gathered, or indeed if any data are gathered at all! I hope that I have shown through this discussion that there is more to the art of interviewing than simply following a script.

References:

May, T. (1997) Social Research; Issues, Methods and Process (2nd edition); Buckingham; OU Press

McNeill, P. (1990) Research Methods 2nd edition London; Routledge


Can statistics make you happy? Maybe if you’re a stats geek like me, but seriously can the path to happiness, or at least some insights into the path to follow be found through the use of statistics? The Office for National Statistics (ONS) have been engaged in work to create a quantitative measure of our national well-being and despite some criticisms in the mainstream media, along the lines that in many cases it was either pointless, or a case of stating the obvious, the work, carried out on a large scale, has produced some interesting results and points to a number of areas for future research. As the well-being data has been collected through the Labour Force Survey the analysis so far seems particularly revealing about the work we do, or indeed don’t do  (though this is perhaps also a weakness as the data reveals less about other non-work factors) and in this post I will be dealing primarily with these issues of work status, occupation type and their relationship with life-satisfaction.

Is working less the key to happiness? The ONS well-being statistics suggest this may well be the case. Image http://www.moderntoss.com

The most immediately striking thing about the well-being stats is the affect of age. According to the ONS figures well-being follows a ‘u’ shaped curve according to age as shown by this graph referring to responses to the question ‘Overall, how satisfied are you with your life nowadays? in which respondents were asked to give a rating on a scale of 0 -10 with 0 being not at all and 10 being completely.

Source: Annual Population Survey (APS) – Office for National Statistics. Data Collected between April 2011 and March 2012

What could be the reasons for this similarly high ratings given by members of the older and younger age group? One thing both groups also have in common is their level of satisfaction of the amount of free time they have. When asked  ‘if they were satisfied with the amount of time to do the things that they like doing’ both younger and older people reported the highest levels of satisfaction.

Source: Opinions Survey, Office for National Statistics. Data collected in October 2011, February and June 2012.

It is quite likely that these higher levels are chiefly because both groups are less likely to have family commitments, or to be engaged in paid-work; For younger people this is because they are still in education, whilst for the older age group many of those aged 65+ will be retired from paid-work.

As a slight aside an interesting observation is that whilst in general life satisfaction appears to increase with age, this actually declines for the 80+ age group. There may be a number of possible explanations for this; one being poor health, which is in the ONS analysis also implicated in lower levels of well-being, or an alternative may be that social capital declines at this age leading to isolation.

So in terms of work is it that not working the answer to happiness? Well, not quite. The well-being statistics also showed that unemployment has a significantly detrimental impact on the level of well-being reported by an individual with 45% of unemployed people rating their life satisfaction as low (5-6), or very low (0-4). This compares with a much lower figure of 20% for people in employment whilst for economically inactive people this figure is 27.1% However, this latter category includes  people who are caring for a relative, looking after children, or who are unable to work due to illness/disability along with students, the idle rich and retired people, who we may possibly expect to have higher ratings – indeed more people in the economically inactive group, some 30.5% also rated their life satisfaction as high (9-10) compared with 24.4% of employed people, and 16.3% of unemployed people suggesting this is a rather bifurcated group.

Because of this wide variation among the ‘Economically Inactive’  group it is particularly hard to reach any conclusion without breaking down the category and though we can clearly see that being employed is clearly far more preferable from a well-being perspective to being unemployed – something which may well be due to the stigma and low status of being in the unemployed group – an interesting comparison to make would be between the unemployed and the idle rich.

Annual Population Survey (APS) – Office for National Statistics. Data from April 2011 to March 2012.

But can the type of work we do have an effect on well-being – yes, according to the ONS analysis which states that in terms of the ‘life satisfaction’ ratings, and sthe ‘how happy did you feel yesterday’ question.

higher scores appear to be given by the occupational groups who tend to have more responsibility and control over their work, as well as higher incomes (which were not controlled for in this analysis)

This seems to align the findings of the Whitehall II study,  based on a sample of 10 308 civil servants, which found that low levels of control at work had a particularly detrimental  effect on health. A (2004) pamphlet explaining the results of the research, which was carried out between 1985 and 1998, states:

While it is common for demands to increase as the occupational hierarchy is ascended, degree of control over work decreases with lower position. Whitehall II provides ample documentation of this: the lower the grade of employment, the less control over work. This combination of imbalance between demands and control predicted a range of illnesses. The evidence from Whitehall II suggested that low control was especially important. People in jobs characterised by low control had higher rates of sickness absence, of mental illness, of heart disease and pain in the lower back.

Whilst there is likely to be a correlation between level of control and remuneration  as this was not controlled in the analysis it is not possible to tell form the data if higher levels of pay alone can increase well-being. One interesting finding however, is that those in the occupational group ‘caring, leisure and other service occupations’ as well as those in ‘professional occupations’ achieved the highest average response to the question ‘overall, to what extent do you feel the things you do in your life are worthwhile’ at 8 out of ten. the lowest averages were reported by ‘sales and customer service occupations’, ‘process plant and machine operatives’, and ‘elementary occupations’ who all averaged 7.5.

This would seem to suggest that work which involves a level of altruistic motivation, or else purports to serve a higher purpose, is beneficial for overall well-being, compared to work in which financial reward is the primary, or sole motivation. This finding is particularly relevant to questions into the role of social purpose, or public service in work and in organisations in general. Could it be that individual well-being is enhanced within organisations such as social enterprises, or other voluntary organisations where there is a strong sense of social mission?

Overall:

In terms of the ONS well-being data it would appear that both younger and older persons, groups who due to their respective ages tend not to be engaged in full-time paid work, report both higher levels of life satisfaction, and higher levels of satisfaction with the time to do things. Though it is not necessarily the case that it is simply full-time work  that results in lower levels of satisfaction in both measures for the other age groups, and it may certainly be the case that other commitments such as looking after children also play a role, it does seem possible that full-time work, can result in dissatisfaction with the time a person has to do things and this in turn may well be linked to the corresponding lower levels of reported well-being for these age groups. Certainly this is an area where future research may be directed as the ONS analysis does not include any details on the relationship between well-being and hours worked, or with a persons satisfaction, or not, with the hours they currently work.

Finally It is also important to note that the LFS does not collect data on voluntary work. It may be the case that, particularly among the retired group, voluntary work may play a key role in well-being. A 2005 Joseph Rowntree Foundation paper Volunteering in Retirement, suggests that voluntary work can be particularly beneficial for the health of retired people. As mentioned the ONS analysis seems to suggest that work which performs a higher purpose is beneficial to well being therefore it may be interesting in particular to analyse the role of voluntary work, or more widely make a comparison of well-being levels between workers in the voluntary, or public sector and the private sector.

References:

Ferrie, J. (2004) Work, Stress and Health: the Whitehall II study CCSU/Cabinet Office

Office for National Statistics (2012) First ONS Annual Experimental Subjective Well-being Results 24th July 2012

Office for National Statistics (2012) Measuring National Well-being –
Measuring young people’s well-being 10th October 2012

Smith & Gay (2005) Active Ageing in Active Communities; Volunteering and the Transition to Retirement Bristol: The Policy Press


This graph was produced to analyse the relationship between a teams form and attendance at football matches as part of an investigation into the factors influencing attendance at football matches.

The key question was; Did a winning streak attract higher crowds, or conversely did a losing streak repel them? I obtained some data for Havant and Waterlooville FC which included attendance figures and results for a number of past seasons. Choosing 2009/10 and 2010/11,  the last two complete years at the time, I then excluded data from cup competitions as attendance for these games varied greatly depending on the prestige of the tournament and the level the opponents played at.

The match attendance for each league game, some 42 in total, is represented by the red line whilst the blue line represents a measure of the teams form at that point in time. I created a measure of the teams form by compiling the results of the previous five league games; Awarding 1 point for a won, 0.5 for a draw and 0 for a defeat I calculated a ratio so that say if the team won all their last five games the ratio would be 1.0, but if the team won two, drew one and lost two the ratio would be 0.5.

I then plotted both sets of data on a single graph.

Whilst not a perfect fit the graph does seem to suggest that there is at least some kind of link between form over the previous five games and matchday attendances. Other factors almost certainly come into play in determining attendances. It may also be interesting to see what  the measure of form can be tweaked to be a shorter, or longer time period, or maybe also made to include goals scored or conceded.


This line-graph came about as part of an investigation into the impact of ground-sharing on the attendance figures of the football clubs involved following an accusation made by a fan of Kingstonian FC that sharing a ground with AFC Wimbledon was having  a detrimental effect on their clubs attendances.

As part of the analysis, which can be seen in full on my Row Z blog, I decided to look at what is perhaps one of the longest ground-sharing arrangements in the upper-reaches of English football; Between 1985 and 2003 Selhurst Park, home of Crystal Palace FC, also hosted Charlton Athletic (1985-1991) and Wimbledon FC (1991-2003).

On the graph I have plotted the average attendance figures for Palace and their respective tenants. Rather than seeing a relationship where  rises in one clubs attendances corresponded with falls in the other clubs attendances (which we would expect if ground-sharing was having a negative impact on one club) it seemed that over the long-term both sets of attendance figures were rising – the nosedive of the ‘tenants’ attendance at the end of the graph is the result of the clubs 2001 announcement that it would be seeking to relocate to Milton Keynes.

One explanation for the long-term upward trend is the general rise in attendance at football games which took place at the same time. 1985, the year the graph begins, was a low-point for English football attendances; the net result of years of hooliganism and poor facilities.

Plotting the line of average football league attendances on the same graph we can see how despite fluctuations, over the long-term the attendances of Palace, and whoever there tenants are follow a very similar trajectory. If we calculate the correlation coefficient for Palace’s attendances and the league average we get  a moderate-to-strong 0.64 which suggests that the variations in overall league attendances explain 41% of the variation in Palace’s attendances over the period shown.

Ultimately the lesson from the graph is that when it comes to football attendances over a long-term period a rising tide really does lift all boats.