Iain Bell to Ed Humpherson: Covid-19 infection survey publication time

Dear Ed,

We continue to review many aspects of statistical production and publication with regards to informing measures to respond COVID-19.

The latest results of the COVID-19 Infection Survey were published at 11am today, brought forward from the originally scheduled 12pm publication time. This was discussed and agreed between ONS and the OSR before the publication. The decision was made as part of understanding emerging operational requirements for the latest evidence on COVID-19 infection rates and incidence.

We anticipate that forthcoming weekly publications of results from the COVID-19 Infection Survey will be at 12pm.

Yours sincerely

 

Iain Bell

Deputy National Statistician and Director General, Population and Public Policy

Office for National Statistics

 

Related Links

Ed Humpherson response to Iain Bell: Covid-19 infection survey publication time

Mark Pont to Siobhan Carey: Statistics from the Northern Ireland Safe Community Survey

Dear Siobhan

STATISTICS FROM THE NORTHERN IRELAND SAFE COMMUNITY SURVEY (NISCS)

We have completed our review of the compliance of statistics from the Northern Ireland Safe Community Survey, produced by the Department of Justice (DoJ), with the Code of Practice for Statistics. This was the final part of our three-part review, which began in May 2018. This letter sets out our key findings from the review. I am pleased to confirm the statistics from the survey should continue to be designated as National Statistics: in this letter we have highlighted several areas for improvement, which will ensure the statistics continue to meet the highest standards of the Code.

As we completed our review, the COVID-19 pandemic meant the Safe Community Survey, which was a face-to-face survey, had to pause while arrangements were made to collect data for the 2020/21 statistics by phone instead. We support this approach, which will ensure the safety of those collecting data for, or contributing data to, the statistics. We look forward to seeing how your plans develop for the 2020/21 statistics and encourage the team to keep users of the statistics informed about the changes.

The rest of this letter briefly summarises our approach to the recent review and our findings, which will apply to the 2019/20 statistics. We illustrate areas where progress has been made since the first part of the review, as well as areas where we consider improvements should still be made to enhance the value of the statistics.

Our review of these statistics has been in three parts; this part focusses on the presentation and reporting of the statistics. As part of our research, we spoke to a range of users of the statistics to understand the impact of recent changes to the statistical bulletins and the extent to which users feel that their needs are being met.

In each part of our review, we considered the trustworthiness, quality and value – all of which are fundamental to supporting public confidence in statistics and, together, form the basis of the Code – of the survey and the statistics. Taken together, our findings across the review are as follows.

 

Trustworthiness

Ensuring sufficient resource

The team has taken on two new statisticians during the course of our review, allowing improvements to the current statistics and to address the improvements suggested in the earlier phases of our work. A new member of staff has now joined the statistics team and the team has published a Future Programme of Work, which provides a high-level outline of developments that the team aims to address in relation to the survey and statistics. These steps demonstrate DoJ’s prioritisation of, and commitment to, the statistics, while offering greater transparency about upcoming changes to the statistics for users. Users we spoke to said that engagement on the Future Work Programme had been good and they were happy with the information around the proposed statistical priorities. We note that dates for different pieces work on the Future Programme are still to be confirmed. In order to enhance trustworthiness, the team should include more-explicit timescales in the Programme and we encourage DoJ and NISRA to continue to ensure enough resources in the team to fulfil this work. Including this information will also allow users of the statistics to more easily engage with DoJ on the order and prioritisation of developments.

 

Quality

Improving data sources

Throughout this review, the statistics team has demonstrated ongoing commitment to improving the quality of the statistics from the Safe Community Survey. In general, the information about quality provided alongside the survey and statistics in the Quality Report and User Guide is extensive, and provides helpful information that allows users to understand the strengths and limitations of the data and statistics.

The survey sample size has increased from 2,000 to 3,500 from April 2018. This was achieved through additional financial investment in the survey, in response to feedback that statistics from the survey were no longer adequate to track progress against outcomes in the Northern Ireland Executive’s Programme for Government Framework. Given the change in the sample size, it was good to see that the team had done some work to ensure the integrity of the long-term crime trend estimates, and included confidence intervals around published survey estimates, so that users can more easily understand the quality of the statistics.

Regional Data

Despite the increase in survey sample size since 2018, some users of the statistics we spoke to still feel that estimates below country and population level need to be more robust. For example, when looking at disaggregated data for different counties in NI, users feel they are unable to make evidenced-based decisions because the confidence intervals around the estimates are so large. This issue arises because the sample sizes for geographic areas and demographic groups below country and population level respectively are relatively small. The team is aware of the user demand for robust statistics at a more disaggregated level, especially for analysis at Local Government District level, and is currently considering how this might be met. In addition to considering how the user need at low geographic levels can be met, the statistics team should consider how it can most effectively explain uncertainty around the survey estimates, so that users understand why the confidence intervals around disaggregated data are variable and can be large. The team should also be clear about how the statistics can and cannot be used.

Impact of COVID-19

It is good that the statistics team was able to move from face-to-face survey data collection to phone data collection. The team has been working well with policy colleagues to discuss changes to the questions and informing users of the changes happening to the survey on its website. We understand that there are some areas that will not be able to be covered in a phone survey. However, it is good know that important topics have been retained, such as:

  • experience of crime
  • perceptions of crime and anti-social behaviour
  • confidence in the police and police accountability arrangements

It is also good that the team has contacted both ONS and Scottish Government to discuss their experiences of changing to a phone survey design, this type of cross-nation engagement will further the improvement of the statistics, particularly during the current challenges. The team now has quarterly meetings with statisticians from the ONS and the Scottish Government, which will enable shared learning and provide an opportunity to discuss current issues and future development.

The change of data collection method will have an impact on the type of data that can be collected and the questions that can be asked. As this work develops, we encourage the team to be clear to users in advance about the impact this may have on future statistics, for example on trends in domestic violence, which could be difficult to collect by phone.

Value

Continued relevance to users

We have been impressed by the statistics team’s collaborative and user-focused approach to developing these statistics. This approach continues to help the statistics team through the current difficulties; the communication between the team, the Scottish Government and ONS will ensure the continued value of the statistics despite a more reduced output in 2020/21.

In 2018, the statistics team completed a consultation process and had extensive dialogue with users and potential users in government about their needs from, and proposed changes to, the survey and statistics. The team took account of these needs, by accommodating demand from policy teams to collect data that reflect important issues not previously covered by the survey. Several of the users we spoke to had been able to input into the questions asked in the survey. To maintain an understanding of the use and potential use of the statistics and data, the team now holds a biannual working group meeting with key policy and statistical stakeholders. In addition, the team has recently completed a branch-wide customer survey which sought comments on individual branch publications including NISCS reports. We look forward to seeing the results of this survey. Having this dialogue will help ensure that the maximum value is achieved from the survey and the statistics.

One of the key messages from the DoJ consultation in 2018 was that the presentation and reporting of the statistics needed to be improved. In response, the statistics team has introduced a new layout for its statistical bulletins, with changes to the content and presentation of the data and supporting narrative. Many of the users we spoke to now find the statistics to be useful and the narrative easy to follow: in particular, the use of annotations on charts has proved effective, as this helps users to better understand trends in the data.

Greater insight for users

As part of the improvements to the statistical bulletins, the inclusion of tables of police recorded crime data alongside the survey data in the Experience of crime publication gives users a fuller picture of crime in Northern Ireland. Our user engagement highlighted that, while the addition of these data is welcomed, there is still more that could be done to maximise the insight users get from them: the statistics team should include commentary that highlights which source of data is best for answering the different questions users might have about different crime types, for example. The crime statistics produced by ONS do this well and provide a helpful example. We anticipate the positive relationship the team already has with statisticians in ONS will facilitate greater insights.

Among the planned developments to the statistics, an issue worth highlighting is the clear demand to understand child victimisation. It has been positive to see questions being included in the Young Persons’ Behaviour and Attitudes Survey as part of an effort to meet this need, it also good that the Northern Ireland Safe Community Survey: Assessment of the Feasibility to Include Children Aged 10 to 15 and Communal Establishments was published on 17 April 2020. To maintain transparency, we encourage the team to continue to communicate its plans effectively to users in this area. This is an important issue and it is vital that the team is transparent with users about whether and when the data and statistics could be improved, remaining clear about the possible effects of moving to a phone survey.

I would like to thank your team for working closely with us throughout the review process and for its commitment to improving these statistics. Our Crime and Security team will continue to liaise with you and your team and, if you wish to discuss any aspects of this letter, we are happy to do so. I look forward to seeing the trustworthiness, quality and value of the statistics continue to improve. I am copying this letter to Ruth Fulton, Head of NISRA Statistical Support Branch; and to Joan Ritchie, the crime statistician at the Department of Justice.

Yours sincerely

 

Mark Pont

Assessment Programme Lead

 

Related Links

Compliance check of Northern Ireland Safe Community Survey

Siobhan Carey reply to NISCS compliance check

Northern Ireland Safe Communities Survey

Ed Humpherson to Roger Halliday: Use of COVID-19 prevalence rates by Scottish Government

Dear Roger

Use of COVID-19 prevalence rates by Scottish Government

On 3 July First Minister Nicola Sturgeon in her COVID-19 speech, claimed that the prevalence of the virus in Scotland is five times lower than it is in England. The sources used to underpin this claim have been difficult to identify. The explanation provided to my team at the Office for Statistics Regulation was not clear. You have now explained to us that the Scotland prevalence figure was sourced from the Scotland’s COVID-19: modelling the epidemic (issue no.6) 25 June and the England prevalence figure was sourced from modelling work done by the London School of Hygiene and Tropical Medicine, using a UK estimate as a proxy for England. The UK estimate for the dates in question was not published and was provided to you to allow for this comparison.  A UK prevalence figure is available on the London School of Hygiene and Tropical Medicine website  however this is a real-time report with an unclear update schedule.

As the UK/England prevalence rate was not available publicly, we understand that you then compared the upper prevalence rates published in Scotland’s COVID-19: modelling the epidemic (issue no.6) 25 June and the Office for National Statistics’ COVID-19 Infection Survey pilot: 25 June. This was done to corroborate the figures from the London School of Hygiene and Tropical Medicine and referred to externally as the other data were not publicly available.

When unpublished figures are quoted in the public domain, we expect that this information is shared with the media and the public in a way that promotes transparency and clarity. There are lessons to be learnt in this case, with different data sources being quoted to the media and to us. We expect that any figures used are appropriately sourced, explained and available in the public domain.

Furthermore, it is important to recognise that a comparison of COVID-19 prevalence rates is not straightforward. If it is to be undertaken, the results and the uncertainties should be communicated transparently. We do not think that the sources above allow for a quantified and uncaveated comparison of the kind that was made. In future if such comparisons are made, we would expect to see sources made publicly available and a clear explanation of the limitations and associated uncertainty.

The Office for Statistics Regulation will continue to monitor Scottish Government’s use of statistics and data.

Yours sincerely

 

Ed Humpherson

Director General for Regulation

 

Related Links

Miles Briggs MSP to Sir David Norgrove – Use of COVID-19 prevalence statistics by Scottish Government

Sir David Norgrove to Miles Briggs MSP – Use of COVID-19 prevalence statistics by Scottish Government

 

Ed Humpherson response to Lewis Macdonald MSP: Press Statement from Public Health Scotland on Minimum Unit Pricing for Alcohol

Dear Mr Macdonald 

 Thank you for your letter of 3 July about the use of figures in a press release of 10 June by Public Health Scotland (PHS) concerning the impact of the first year of the Scottish Government’s policy on minimum unit pricing for alcohol. You were concerned that the data used in the press release misdirected subsequent media coverage of the research.  

Having looked at the concerns in detail, we concluded that the main findings of the research (a net decrease in Scotland of between 4 and 5% in sales per adult over the first year of the policy) are dependent on the difference in movements in alcohol offsales in Scotland (a decline in sales of 2.6%) compared with England and Wales (an increase in sales of 2.3%). The researchers have controlled for different effects including using England and Wales as a geographical control. The results of using the geographical control and other controls such as household income gives slightly different results depending on which controls are applied, hence the range of the estimated decline in offsales of between 4 and 5% over the period. The figures in the press release are consistent with these findings.  

The research has effectively reported an estimated change but has not reported the observed year-on-year change. There are areas where there could have been a clearer presentation in the press statement, particularly that the research reports an estimated effect of the policy change. The PHS statisticians we were told had no intention to mislead in the press statement, they were trying to keep the messages straight-forward.  

It would have been helpful in the press statement to refer to the observed year-on-year decline in offsales as was apparently referred to in the Good Morning Scotland radio interview with the PHS investigator that you refer to. Also, there could have been a better explanation about the limitations in the statistics, and PHS might have included some indication of the range of the plausible reductions in offsales (-5.3% to -3.0%). PHS might also have provided some useful lines in the press release to help the media to interpret and quote the figures appropriately. Such explanations protect the integrity of the findings and support users of these numbers in drawing the correct conclusions to inform the decisions they make. 

As you point out PHS is a new agency and through our engagement we have seen that its staff understand the importance of maintaining the standards required of the Code of Practice for Statistics. PHS is building on the strong pedigree of ISD Scotland, one of its component parts, and its statistics officials are working with staff across PHS to instil a culture which looks to draw heavily on the principles of the Code of Practice for Statistics and associated guidance. 

The Office for Statistics Regulation will continue to work with PHS to ensure that support is provided as it develops its practices and will support them in applying the principles of the Code of Practice. 

I am copying this letter to Scott Heald, Head of Profession for Statistics at Public Health Scotland, and Roger Halliday,Chief Statistician and Joint Head of Covid Modelling and Analysis Team at Scottish Government. 

 Yours sincerely 

 

Ed Humpherson 

Director General  

 

Related Links

Lewis Macdonald MSP to Ed Humpherson

Lewis Macdonald MSP to Ed Humpherson: Press Statement from Public Health Scotland on Minimum Unit Pricing for Alcohol

Dear Mr Humpherson,

I write to raise concerns about a press release from a new Scottish Government agency, Public Health Scotland, published on 10 June 2020. It was issued with a report containing a statistical analysis of the impact of Minimum Unit Pricing (MUP) on off-sales of alcohol in the first year after the policy was introduced on 1 May 2018.

The news release states that, in the twelve months after the introduction of MUP, off-licence and retail sales of alcohol in Scotland decreased by between 4 and 5% when compared with England and Wales. The news release does not say how the level of such sales changed when compared with off-sales in Scotland itself in the 12 months before MUP was introduced – not even in a footnote. Even more surprisingly, the actual change in levels of sales in Scotland is also omitted from the report published at the same time.

This presentation quite predictably led journalists to report that there had been a fall in off-licence sales in Scotland of between 4 and 5%, thanks to the introduction of MUP. That in turn led other reputable individuals to comment on the story as if there had indeed been such a large fall, because the actual figures were not made available.

The actual figures seem to have emerged only in an interview given by Lucie Giles, Public Health Intelligence Principal at Public Health Scotland, to BBC Radio Scotland’s Good Morning Scotland on the morning of 10 June. She told the interviewer that the actual reduction in off-licence and retail alcohol sales in Scotland was 2.6% during the first year after MUP was introduced, and that there was an increase in sales over the same period in England and Wales of 2.3%.  “Incorporating the figures into a single model, using England and Wales as a geographical control,” she said, “that’s how we have come to those figures of a net reduction of four to five percent.”

There is nothing intrinsically wrong with comparing a 2.6% reduction in Scotland with a 2.3% increase in England and Wales, and suggesting possible explanations for the difference. There may be issues about adding these two numbers together, given that the market for alcohol sales in England and Wales is quite different in scale from the market in Scotland: but a combined figure of this kind could have been presented with the appropriate caveats.

That did not happen. Nobody, as far as I know, other than listeners to Good Morning Scotland at the relevant time on the day of publication, will know how far sales in Scotland actually changed from one year to the next, far less how the published figure of “between 4 and 5%” was reached. Most people are bound to assume that the figure presented actually represents the fall in off-sales in Scotland.

The Glasgow Times for example reported: “According to the report by Public Health Scotland, minimum unit pricing (MUP) could be attributed to a reduction in alcohol consumption per head in Scotland of between four and five per cent overall in the 12 months from May 2018.”

STV, on the other hand, mentioned that the figure was in comparison with England and Wales, but did not make clear what this actually meant. Their story quoted Alison Douglas, chief executive of Alcohol Focus Scotland, who said: “A reduction of between four and five per cent in off-sales in the 12 months following the introduction of MUP is really significant.”

None of the outlets I checked or authorities I saw quoted gave the actual figures for the fall in consumption in Scotland, because that information was not presented to them. It is disappointing that a public agency chose to present statistics in such a misleading way, so that even expert witnesses concluded that the fall in off-sales of alcohol following introduction of MUP was nearly twice as large as it actually was. Given that Public Health Scotland is a new agency, with many important responsibilities, I would encourage you to draw the principles of best practice in the use of statistics to their attention.

I attach below links to the PHS news release, and to the news reporting I have mentioned.

Yours sincerely,

 

Lewis Macdonald MSP

North East Scotland

 

https://publichealthscotland.scot/news/statistical-analysis-of-off-trade-alcohol-sales-in-the-year-following-mup-published/

https://www.glasgowtimes.co.uk/news/18507554.minimum-pricing-sees-scots-alcohol-consumption-fall—sales-fortified-wines-increase/

https://news.stv.tv/west-central/drop-in-alcohol-consumption-down-to-minimum-pricing?top

 

Related Links

Ed Humpherson response to Lewis Macdonald MSP

Jess McGregor and Julie Billett to Ed Humpherson: Public Health England Care Home Management Information

Dear UK Statistics Authority

RE: Public Health England (PHE) Care Home Management Information

We are writing on behalf of Camden Council and in relation to the PHE management information on care homes.

Camden Council have had serious concerns about how the data on outbreaks within care homes for Camden have been represented. The data first came to our attention a few weeks ago when the percentage of outbreaks was over 100% and subsequently rose to 136% when the local evidence shows the percentage affected is actually much lower. The discrepancy has caused a lot of work locally in explaining to political leaders and the media why the figure is incorrect and it is of particular concern to us that it was being reported in this way from a national agency. It has caused substantial issues around credibility and trust of statistics. On further review of the outputs from the data, we are equally concerned that the presentation of this management information is generally misleading and below the standards we normally expect to see.

While we appreciate that PHE has now corrected the Camden data (by removing homeless hostels from figures and a care home which on testing had negative results) and has made an amendment in the meta data to make it clearer that the numerator and denominator are not equivalent, we still think more should be done to ensure that the data that is being published by a national agency is not misleading. We also understand that PHE are looking into making further amendments to this data, but we are unclear about what specific action they will be taking to rectify issues.

Our specific concerns are:
1. Discrepancy between the numerator and the denominator

The reason that Camden has been reported >100% is that the numerator and denominator are not equivalent, with non CQC registered care settings included in the numerator but not the denominator. We have requested that PHE reviews whether it is appropriate to publish data with this type of discrepancy when it should be possible to limit the numerator to CQC registered care homes and make the numerator and denominator data equivalent.

We have also requested that PHE considers whether combining settings into a ‘care home’ category like this is meaningful to social care. Within social care, the term care home does not incorporate these wider settings.

2. Presentation within the infographic

There is no explanation of the major limitations of these statistics on the infographic provided by PHE and we think that some of the wording and presentation of the data needs to be amended so it is not misleading. Given the way it has been calculated, it will become more misleading as the pandemic progresses and less useful.

It would be easy for someone looking at these statistics (bottom graph) to come to the conclusion that there were 111 outbreaks in care homes during the week of the 25th May 2020. When in fact, the actual figure is somewhere between 111 and 6,225 if there are some ongoing in some care settings and there are new outbreaks in care homes that have already had one.

We think that the wording about the ‘weekly suspected or confirmed COVID-19 outbreaks in care homes’ needs to be amended to show they are new outbreaks and in homes that have previously not had a reported outbreak (or may still be having one) and may include other care settings too.

We have discussed with PHE that they are looking at ways to start calculating the beginning and the end of an outbreak using the data, and this would clearly be a better way of describing the impact of COVID-19 on care homes or settings, and enabling areas to identify whether they have care homes or  settings with repeated outbreaks which may require further interventions and support. We would be supportive of this but only if PHE can provide accurate and meaningful data.

3. Meta data and data tables

In line with comments above on the infographic, we think the meta data and data tables also need to be clearer on all of these points.

As we have said above, the production of this management information has caused us a lot of work locally because in our view, there has been insufficient thought about how the data are presented, its limitations, and understanding of social care settings. We have been in dialogue with PHE about this for three weeks, and while we are pleased that they have now corrected the Camden figures, we are not clear whether they accept our points above and will be modifying their outputs in line with this. Given the amount of mistrust this has all caused in statistics locally, we would be grateful if the regulator could look into this and take a view to prevent this happening again in the future.

Yours faithfully

 

Jess McGregor, Director of Adult Social Care

Julie Billett, Director of Public Health

 

Related Links

Ed Humpherson response to Jess McGregor and Julie Billett

Ed Humpherson response to Jess McGregor and Julie Billett: COVID-19: number of outbreaks in care homes – management information published by Public Health England

Dear Ms Billett and Ms McGregor 

COVID-19: number of outbreaks in care homes – management information published by Public Health England 1 

Thank you very much for your letter of 24th June to the UK Statistics Authority. My team at the Office for Statistics Regulation have considered the matter in detail and talked extensively with the team responsible for the statistics at Public Health England (PHE). 

We agree with your concerns and, whilst the information presented has been compiled at speed with the intention of transparency, there are a lot of important details about the types of setting that have been lost.  

The dataset is released with granular information at the local authority level, which would be of great use to stakeholders; however, the misleading use of the term “care home” in this context gives the data less utility than it should have. The data is sourced by PHE from all supported living facilities and it should be described as such, to make most sense for the experienced users. We have suggested that PHE amend the title of the release to better reflect this. 

We agree there is an erroneous calculation of proportions, using mismatching numerator and denominator data. The numerator and denominator are not comparable, because non-Care Quality Commission (CQC) registered care settings are included in the numerator, but not the denominator. This fact is noted in the metadata on the landing page and in the data tables. The use of a denominator that underestimates the total number of facilities at risk of suspected or confirmed outbreaks, leads to over-inflation of the proportion of facilities with a suspected or confirmed outbreak. My team has explored with PHE the use of the small denominator and they are considering how to improve the calculation with the data they currently collect. 

There is no explanation of the major limitations of these statistics on the infographic provided by PHE. The infographic includes the proportion of outbreaks and the number of care homes for each PHE centre but, as noted above, this denominator does not include all settings at risk captured in the numerator. PHE is considering removing the proportion numerical and replacing it with an indicator calculated using comparable numerator and denominator. 

We agree with your concern that the presentation as it stands makes it difficult to identify whether a region has supported living facilities with repeated outbreaks which may require further interventions and support. My team has explored with PHE what improvements are possible to the data and presentation to make the information more useful for a range of stakeholders. PHE assures us that it has identified more data that it can use to present a more comprehensive publication and that this will be released later in 2020. In the interim, we have asked PHE to engage publicly with stakeholders to let them know about proposals to change the current publication and obtain stakeholders views on the content, timing and presentation of the new material. 

We agree with your third concern and have asked PHE to improve the information about the publication itself (the metadata), by publishing details of the data sources, the methodology, the quality assurance processes and the strengths and limitations of the data. This will help a range of stakeholders to decide if the publication is the right material for their needs. 

Thank you very much for drawing this to our attention and will closely follow PHE’s progress with this work. 

I am copying this letter to Clare Griffiths, Head of Profession for Statistics at PHE and Julia Verne, Head of Clinical Epidemiology at PHE. 

Yours sincerely 

 

 Ed Humpherson 

Director General for Regulation 

 

1 COVID-19: number of outbreaks in care homes – management information

 

Related Links

Jess McGregor and Julie Billett to Ed Humpherson: Public Health England Care Home Management Information

Ed Humpherson response to Peter Benton: Temporary suspension of National Statistics designation for overseas travel and tourism statistics

Dear Pete 

 TEMPORARY SUSPENSION OF NATIONAL STATISTICS DESIGNATION: OVERSEAS TRAVEL AND TOURISM STATISTICS 

Thank you for your letter of 22 July, regarding statistics on overseas travel and tourismGiven the suspension of the International Passenger Survey in mid-March which provides the underlying data for these statistics, I support your decision to temporarily remove the National Statistics designation. 

I welcome your plans to publish data for the first quarter of 2020 later this weekwhich will be highly valued by users. I encourage you to ensure that the impact of the missing March data on the quality and interpretation of the statistics is communicated clearly. am pleased to hear that you have been engaging with users about the impact of COVID-19 on these statistics and encourage you to continue this over the coming months as you plan for the safe return of the survey. 

We will continue to engage with your statisticians, via our Travel, Transport and Tourism team, and look forward to considering the restoration of National Statistics status once data collection has resumed and you are assured the statistics can meet the standards expected by the Code of Practice for Statistics. 

 Yours sincerely 

 

Ed Humpherson 

Director General for Regulation 

 

Related Links

Peter Benton to Ed Humpherson: Temporary suspension of National Statistics designation for overseas travel and tourism statistics

Peter Benton to Ed Humpherson: Temporary suspension of National Statistics designation for overseas travel and tourism statistics

Dear Ed,

I am writing to ask you to consider a temporary suspension of the National Statistics status for the international travel and tourism estimates based on the International Passenger Survey (IPS).

The IPS stopped interviewing on 16th March 2020 due to safety concerns arising from the Covid-19 situation. It has been possible to produce estimates both for March and for the first quarter of 2020, but we are unable to confirm that the estimates fully meet the required quality standards demanded of National Statistics.

This is partly because the volume of data is greatly reduced for March. Also, we cannot be certain that the data obtained for March are representative: that the characteristics of passengers interviewed in the first half of the month represent those travelling in the second half of the month. Traveller numbers were decreasing at this time but had not yet fallen to the extremely low numbers seen from April onwards.

We have investigated the impact that the lack of data for the last two weeks of March will have on the Q1 2020 and March 2020 estimates. The data is weighted to the total traffic numbers provided at a monthly and quarterly level by the Civil Aviation Authority (CAA) and the Department for Transport (DfT). Therefore, we are weighting an incomplete dataset to the complete traffic figures which are not suitable for weighting at a weekly level. Travel started to be significantly affected towards the end of March as COVID-19 restrictions started to come into force. Hence, our investigation suggests that the IPS estimates for March and Q1 2020 have a degree of uncertainty that makes it preferable to suspend National Statistics status for this period. We have nonetheless made every effort to ensure that the estimates are of the highest possible quality in the circumstances.

Given the constraints on data collection experienced by the IPS, I am requesting temporary de-designation of travel and tourism statistics based on the survey on the basis set out by the published COVID-19 guidance on the suspension of National Statistics status.

Subject to your agreement, we propose to remove National Statistics labelling for March and Quarter 1 of 2020 outputs in the first instance. The survey’s output is suspended for the period from April 2020 until interviewing resumes. It may be advisable to continue the suspension of National Statistics status for a further limited period after interviewing resumes,depending on our assessment of the quality of data obtained.

Yours sincerely,

 

Peter Benton

Director of Population and Public Policy Operations, Office for National Statistics

 

Related Links

Ed Humpherson response to Peter Benton: Temporary suspension of National Statistics designation for overseas travel and tourism statistics