You asked
1. The report
1a. For all school worker categories except "teacher of unknown type,"
the error bars are larger than the positivity values (by more than a factor of 2 in most instances), indicating that these data were too under-powered to draw useful conclusions, but strong conclusions were nevertheless stated.
1b.The only school-related datum with reasonable error bars------that for "teacher of unknown type," had the highest positivity of all categories------a 28% higher positivity than that for key workers and a 16% higher positivity than that for "other professions"------but the ONS nevertheless captioned Figure 12: "Since the start of the school year there is no evidence of difference in the positivity rate between teachers and other key workers," a claim which has been almost universally reported------and which the caption writers almost certainly knew would be reported------as evidence that there is no difference in the positivity rate.
1c. This caption further uses the phrase "since the start of the school year" despite the Figure only having used data acquired up through 16 October, for a bulletin published on 6 November which otherwise reported data up to 30 October, even though the strongest evidence for difference in infection rate between school and other workers would have come approximately from the weeks from 9 October onward. This caption further uses the phrase "since the start of the school year" despite the Figure only having used data acquired up through 16 October, for a bulletin published on 6 November which otherwise reported data up to 30 October, even though the strongest evidence for difference in infection rate between school and other workers would have come approximately from the weeks from 9 October onward.
2. Methodology
2a. How was this data collected? (For instance, was this occupation information collected during Pillar 1, Pillar 2, random sampling, or other testing, or was it collected in some other manner such as surveying groups to ask if they had been tested and what the results were? Was this data collected in all of the UK or only in certain regions/venues or with select groups? Were people given a multiple-choice form for their occupation, or did they write it in?)
2b. Was such occupation-differentiated test data also acquired later than 16 October? If not, when was the decision made to stop acquiring this data beyond 16 October?
3. Data
3a. What are the absolute numbers from which these percentages were calculated; i.e. for each of the 8 worker categories in Figure 12, what were the number of people tested and the number of positive tests? (Since these were unweighted percentages, this data should be easy to provide.)
3b. Please could you also provide a time-dependent version of this absolute data, in week-sized buckets? (This is crucial information, since the currently-reported data is diluted with several weeks' worth of positivity data that preceded the opportunity for significant school infection spread, and is cut off prematurely during the rise of this spread.) If you are only able to provide positivity instead of absolute numbers for this time-dependent version, please could you create a 9th "teacher" category which combines Figure 12's categories 1, 2, 3, and 5, for its positivity calculation?
3c. If occupation-differentiated test data continued to be acquired beyond 16 October, please could you provide the data requested in point for these latter weeks as well, up through at least 30 October, even if the data is incomplete?
We said
Thank you for your comments in section 1 of your correspondence. As always we aim to be fully transparent about decisions made and the analysis underpinning these decisions. We have published a statement which should hopefully cover any additional questions you may have further to those comments. The statement can be found here and covers the grouped version of the analysis we published.
With regards to the specific questions in your request:
The COVID-19 Infection Survey is a random household survey designed to be representative of the whole population. The survey is part of pillar 4 of the COVID-19 testing strategy. Individuals complete a questionnaire when a fieldworker attends their home to provide a swab kit for them to self-swab. You can find information on the study protocol here and the questionnaire here. The analysis is for England only.
This data continues to be collected but at the time of publication the latest coded data was until the 16 October.
On tabs 1k and 1l in the data tables accompanying our bulletin on 6 November we published the absolute numbers for this analysis.
We do not hold a version of this analysis by week.