Time line for Bristol by luckboy

VIEWS: 31 PAGES: 44

More Info
									RSS & External Relations
(history; present/future; News examples)

Randomisation
Biology, 1926: Sir Ronald Fisher

Medicine, 1947: Sir Austin Bradford Hill Randomised Controlled Trial 4% random sample: RSS external relations

Council

Titled RSS Presidents
Sample Year

FRS/MP/LLD/MD; Prof/Dr etc
Titled Council
Female Fellows in Sampled Years 1st 1857 + 1st 1858

Quarter century 1834-58 1859-83 1883-1908 1909-33 1934-58 1959-83 1984-2008

1839 1865 1886 1925 1941 1961 1985

13/13 14/29 17/29 3/33

15/31 25/59 26/622F 29F/618F

FN+EN+HB-C

1st 1975 2nd 1999

RSS at random: 1839
18 March 1839: Committee to observe & record operation of new Poor Law in Belfast Union – to enable the Society to judge of the effect of the great experiment now to be tried . . .
List of Statistical Papers printed by Houses of Parliament. Papers on: sickness & mortality among Troops in UK; railways in Belgium (cost per mile - RWR); education in Pendleton (only half school-age children attend); crime in E&W (Rawson W. Rawson); the numerical method applied to Science, esp. to Physiology & Medicine (William A. Guy: mean, range, risk-adjustment; not variance) in favour of Systematic Collection of the Statistics of Agriculture – “What should we think of a General who should make no inquiries into the means that existed for the daily feeding of the army committed to his charge?”

Criminal Statistics of Preston: Rev John Clay
who intends to adopt form of registry suggested by a Committee of the Statistical Society

Tarriff Transportation: 14 years; or life Transportation: 7 years; or 10 Imprisoned for 2 years Imprisoned for 1 year Imprisoned for [6 months, 12 months) Imprisoned for [3 months, 6 months) Imprisoned for [1 month, 3 months) Imprisoned for under 1 month Acquitted at bar Other disposals

Frequency 16; 2 32; 2 2 26 56 71 46 9 25 11/302

RSS at random: 1865
From the Economist: It seems probable that, on re-assembling Parliament, some effort will be made to deal with the crime of child murder. Not a new crime . . . but has recently become very obvious . . . The case of Charlotte Winsor, the terrible statements of Dr Lancaster . . . have shocked the public conscience. From Saturday Review: once a thief, not always a thief! Prior to 1861, all persons ever convicted were reckoned by the police as belonging to the criminal classes; but now such persons, if known to have been living honestly for a year, are excluded. This is only one of many examples of the careless compilation of statistics . . . Papers on: infant mortality, & on alleged inaccuracies of census: Part 1 (William Farr); mortality at different stages of life: maturity= 55-75; ripeness= 75-85; old age = 85+ when ‘hazard’ was 30%.

On the original and acquired meaning of “Statistics”, and the proper functions of a Statistical Society (William Augustus Guy) Aim: to vindicate the claim of Statistics to an honourable place among the sciences. Common mistake: to overlook alike i) the judgment and scientific insight which planned the inquiry; and ii) the critical and analytical talent employed in discovering and displaying the truth.

“As the statist thinks, the bell clinks”

RSS at random: 1886
Regret at decease of valued member Dr Guy {1839-85}: Had done so much to support and advance interests of the Society for nearly half a century, and then at his death bequesting a sum of £7,000+ for its support. FORM of BEQUEST: Persons who are inclined to benefit the Society by legacies are recommended to adopt the following! JUBILEE year follow-ups: International Statistical Institute; & RSS incorporated by Royal Charter 1887 Papers: Suicides in E&W in relation to age, sex, season and occupation (William Ogle MD FRCP); Note on the progress of New Zealand for 20 years, 1864-84 (Sir Robert Stout, Premier of Colony) - pullout colour maps & graphs; ‘81 Census: 490K, 121K born E&W, 53K Scotland, of whom 70% male.
Miscellanea: The Mathematical Method of Statistics (Edgeworth)

(FY Edgeworth: in London Philosophical Magazine)

The Mathematical Method of Statistics

Motivating example: Total abstainers have longer mean life than general population. What is the probability that this discrepancy is not accidental, but significant of a real connection between longevity & total abstinence from liquor? Modulus & Theory of Errors: 2 standard errors! “The modulus is sufficiently small when it is less than half the observed difference upon which the argument is based.” Undecomposed average mortality + “I cannot pronounce without an extensive palpation of the statistics”

RSS at random: 1925
Presidential Address: G. Udny Yule FRS Growth of population and factors which control it Illustrations of logistic law . . . notes on mathematics of logistic curve and methods of fitting. Experimental data . . .
100-bird v. 150-bird pens: 10% reduction in mean annual egg-production via 1/3rd reduction in space per Bird! Papers: Bearing of labour unrest upon the path to be taken to sound currency (Alfred Hoare): “I make an apology for the political character of this paper . . . But, studies of the economic tendencies and probable results of Government action may, in the time through which we are now passing, be useful to the country . . . “

RSS: 1941

history of Society & Dinner Club

(1,000+ Fellows)

Inaugural Address: Henry W. Macrosty Economic statistics – retrospect and prospect. From 1834-69 our Presidents . . . distinctly belonged to the ‘Best People’. Newmarch definitely broke down this tradition, and for a long time our Presidents were generally men who had distinguished themselves in statistical work. About the beginning of this century, the custom grew of alternating an “internal” with an “external” President, one who would keep us in touch with the great world outside our walls. This year that has not been possible, so here I am . . . My war work taught me very thoroughly the risks of forecasting Suspension of Guy medals for the present . . . Contrary to national interest to continue to publish the monthly Trade Accounts . . . (House of Commons announcement) Papers: Secular changes in death rates (EC Rhodes; discussed by Kermack & Lt-Col McKendrick); Cost of living of the working classes - 3% on smoking; 1% on drink; 2% on doctor/medicine; 8% on fuel/light.

(1961 Presidential Address by Maurice G. Kendall)

Natural law in the social sciences

Motivation: Systematic regularities which we can detect in the social sphere have an immediate application as instruments of government even if we do not fully understand the reasons for their existence. Common mistake: Pathway of knowledge is littered with the wreckage of premature generalization . . . Strikes per 4-weeks: non-Poisson for coal-mining (DJS)

RSS: 1961

(3,000+ Fellows)

Presidential Address: MG Kendall Natural law in the social sciences. Chair in Mathematical Statistics at Cambridge: RSS Council to sponsor appeal for funds to raise £4K in each of 10 years, then University’s cost. Supplement to the statistical journal began in 1934: known as Series B from 1948. Papers: Diets of bank clerks (JA Heady); Design of experiments: the control of error (DR Cox) “Various ways in which error can enter into an experiment are discussed, and the techniques of experimental design for controlling errors are reviewed.”

(1985 Presidential Address: Walter F. Bodmer FRS)

Understanding Statistics

Motivation: Simple statistical and experimental principles underlie the understanding and interpretation of many everyday phenomena. RA Fisher’s Design of Experiments: still worth reading for its clear exposition of the basic principles of experimental design, esp. randomisation. If all scientists, whether medical, social, biological or physical, understood no more than this, then much would already have been achieved. Levels of Understanding: 1. everyone in a democracy; 2. professions wrt data interpretation/analysis; 3. professional – from data gatherer to mathematician.

RSS at random: 1985

Sir David Cox

Presidential Address by WF Bodmer FRS: Understanding Statistics “an issue that the Society should take up seriously . . . “ Opinion polls are thrust at us daily . . . ; What might be the effect of lowering speed limit on . . . ; Excess cost in millions of some Government project – should the millions be calculated in 1985 or 1955 pounds?

Royal Society committee: Public Understanding of Science. RSS Working Party Report: Projection of Student Numbers in Higher Education Papers on: Multivariate generalizations of proportional hazards model (Clayton & Cuzick) Initial Examination of Data (Chatfield) Potential contribution of enhanced electoral registers (Black) Fluoridation of Strathclyde’s water supply: opinion of Lord Jauncey – a review (Oldham)

Adrian FM Smith FRS

Mad Cows and Ecstasy!

“What works” in UK criminal justice?
Large RCTs essentially untried . . . (bar restorative justice)

Judges prescribe sentence on lesser evidence than doctors prescribe medicines

Is public aware?

21st Century External Relations “Significance”

RSS External Relations: Andrew Garratt Consultation responses/RSS position: rss.org.uk Media contacts for Fellows: myths & memoranda Press coverage of (local) RSS meetings: toolkit National Honours for FellowsHonorary: recognition Statistical Excellence in Journalism: awards RSS NICE Seminars for Journalists: Qs to ask Public Understanding: soon “Straight Statistics” Parliamentary Liaison: Statistics Act, briefings, seminars & PQs: work in progress.

RSS Working Parties
Counting with Confidence: Statistics Commission Statistics Act UK Statistics Authority; Holt Statistics and Statisticians in UK Drug Regulation: statisticians appointed to MHRA debate @ EMEA Measurement of Unemployment in UK: definitions administrative data professional integrity Performance Monitoring in the Public Services: see later Statistical Issues in First-in-Man Studies: experimental design early-evidence safety synthesis by regulators NB scientist peers Data Capture & Society: RSS Edinburgh; Coroners Bill.

Outstanding Matters M’learned friends: Sally Clark
court . . . injustice biased evidence in balance of probabilities.

Statistical education for lawyers/judges:

“forensic statistics” detection rate via DNA database Colin Aitken. time for RSS Working Party to count the costs? figures lie when liars figure . . . see Michael Blastland

Policy Trials versus Ministerial kite-flying:

PQs & Ministers “lying to the House”/Today:

Knifed by Press Release: “errors” of omission and
commission.

The Tiger That Isn’t Michael Blastland & Andrew Dilnot
Defence of tagging was brisk, assertive & spurious!

Today interviews Home Office Minister on Home Detention Curfews (HDC): only 4%

of 130,000 people on HDCs had committed offences on the scheme compared to 67% of newly-released untagged prisoners . . . Who? HDC-selected because at low risk of re-offending When? HDC lasts < 4.5 months; 67% refers to 2 years What? HDC compared to prison no re-offending. “Open and shut statistical felony, and even prompted unusually forthright attack from the normally reserved Royal Statistical Society.”

Posse of Peers: “statistical felony” Investigate & Influence Policy Trials

Performance Monitoring in the Public Services

Change in teenage pregnancy rates: 1998 to 2001
1.6 99.8 % limits 95 % limits Ratio of Standardised rates (2001/1998) 1.4

1.2

1.0

0.8

0.6

0

200

400

600 Average observed

800

1000

1200

Electronic tag as worn by offenders

Scottish Information Commissioner
Projected Escorts; Costs & Public outcomes
Financial Year Payouts 2005/06 2006/07 2007/08 2008/09
projected

Escorts Cost: cost
per escort

167,603 £20.7m: £123 185,739 £22.3m: £120 183,258 £23.1m: £126 190,896 £24.2m: £127

% at court on-time 93% 90% 85% 84%

# Reliance prisoners Custody at large Deaths Not 20
reported Not reported

13 16 3 in 8m.

6 6 in 8m

Military matters . . . !

Clive B. Fairweather CBE The late Lord Tim Garden: RAF Patrick Mercer OBE MP

Headlines, November 2004 [1]: 100,000 civilian casualties in Iraq
Study design: cluster-sample of households
(pre-2003 census: Jonty Bloom’s RSS award)

Lancet Commentary: public health & military sciences need to ally . . . Significance: story continued . . .

High noon for statistical science: “you can’t extrapolate from a sample . . . “

20-week Reporting Periods [3] . . . UK military fatalities in Afghanistan v. Iraq: rates, major causes, clustering, projections (statistical methods explained in IJE) Clustered fatalities: air-related & due to Improvised Explosive Devices (IEDs)

Period 1: from 1 May to 17 Sept. 2006 UK military fatalities in Afghanistan!
Ex-military intelligence & ex-SAS . . . “restricted” MOD clearance for report go-ahead: Lancet/BMJ declined . . .

NEWSCIENTIST written-down to 300 words by their reporter ~ world-wide News ~ RSS Belfast Guardian’s Ian Sample: guardian online gave link to report on BSU website . . .

Compulsory Drugs Testing in the British Army [4]: awash with data
The late Lord Tim Garden; Patrick Mercer MP.

Increased Efficiency at Detection masked trend in soldiers’ cocaine use

British Army, 2003 - 2007 1. Accentuated Monday testing 2. Differential testing by rank: privates! 3. Lowered threshold for cocaine

Posse of Peers: “statistical felony” Investigate & Influence Policy Trials

Special Thanks
Peers, MPs, policy-leads, media specialists & Fellows for judging
RSS Awards for Statistical Excellence in Journalism plus Andrew Garratt.

Fellows & other-science-peers who contribute to
RSS Seminars for Journalists: Abrams, Ashby, Barnett, Bird, Claxton, Diggle, Donnelly, Ghani, Grieve, Hand, Holt, Spiegelhalter; plus Ben Goldacre & Sir Michael Rawlins.

Guardian Society: 17 Nov. 2004
“Some statisticians are so severe that they would stop social policy making in its tracks. For example, Bird would forbid the government to introduce any policy that had not been assessed through controlled trials. . . ”

Fatal IED incidents & IED fatalities
Period 3=5 Feb to 24 June 2007; period 7=18 Aug 08 to 4 Jan 2009

THEATRE IRAQ AFGHANISTAN 20-week period 3 period 7 period 3 period 7 Surge-in period 10 12 42 Fatal IED 155 incidents 9 8 29 1 fatality 97 2 fatalities 22 1 1 7 3 fatalities 20 2 5 4 fatalities 10 1 5+ fatalities 6 1 IED Fatalities 280 11 22 62

20-week periods 1 + 2 + 3 + 4 (half): 1 May 2006 to 2 Sept 2007

THEATRE Fatalities US:
pys

IRAQ 280 416 463 178
56K 56K < 63K ~ 31K

AFGHANISTAN 40 96 63 31 13
1.3K

299 436 490 194 117 54A

18 50A 6
2.0K

UK:
pys

14
2.8K

12
2.7K

23
2.6K

15 33N
1.1K 1.7K

15
2.2K

Estimated fatality rates per 1,000 pys (95% CI) US, surge UK, exits Basra City 5.0 7.5
(7– 8)

7.3 8.8

5.7
(5 - 7)

US troop numbers not ascertained

4.8
(3.1 to 7.0)

14

19

3

7

10

(6-13) (8-23) Nimrd

(4-11) (5-17)

Series B, topics such as:

1961, volume 23
Delays on a 2-lane road Queuing model: road traffic

1985, volume 47

(3 parts)

Spline smoothing approach to nonparametric regression curve fitting Balance, orthogonality and efficiency factors in factorial designs

Average run length in CUSUM chart Changeover designs and factorial with V-mask structure Prediction by exponentially weighted moving averages & related methods Inaccuracy & inference Optimal programmes for varietal selection Mixtures of geometric distributions Use of uncategorized data to improve nonparametric estimation of mixture density Symposium on stochastic networks Distributional model for repeated failure time measurements Bayesian estimation of finite population parameters in categorical data models incorporating order restrictions.

Randomised controlled trials
to

police Policy
by

Home Office Prisons
&

Criminal Justice
et alia.

Royal Statistical Society called for
PM protocols
(& explained how not to set targets)

Independent scrutiny of disputed PIs Reporting of measures of uncertainty Research: into strategies other than “name & shame” + better designs for evaluating policy initiatives Wider consideration of PM ethics & costeffectiveness


								
To top