[Forum SIS] Correction DEC - Statistics Seminar - October 27th

[Isadora Antoniano]

Dear all, there was an error in the previous email. The correct date for
the seminar is *Thursday, October 27*.

Apologies!

We are glad to announce the next:

*DEC - Statistics Seminar*


*Thursday, October 27th*

Bocconi University,
Room 3-E4-SR03
Via Rontgen 1 - 3rd floor

Time: *12:30 pm*




*Vanessa Didelez*(Leibniz Institute for Prevention Research and
Epidemiology - BIPS GmbH)


*Causal inference under case-control and other outcome dependent sampling*

*Abstract:*

In this presentation I will review when and how it is possible to draw
causal conclusions from case–control designs; some of the results are valid
more generally for other situations where by design or accident the
sampling depends on the outcome. The main focus is on the question of
identifiability: does the available data, at least in principle (for ‘very
large’ samples sizes), allow us to consistently estimate the desired causal
quantity? If the answer is ‘no’ then this is typically due to structural
bias, i.e. to fundamental problems of design and available information,
which is obviously strongly influenced by the sampling design. In
case–control studies, we face the following potential sources of structural
bias regarding causal inference: (1) Case–control studies are necessarily
observational, so confounding is likely to be present. (2) Case–control
studies are retrospective with sampling being conditional on disease status
which means there is also a threat of selection bias. (3) A consequence of
the retrospective sampling is that methods which depend on, or are
sensitive to, the marginal distribution of the outcome cannot be used
without some modification, since the required information is not generally
available. This is potentially relevant to certain methods of adjusting for
confounding as well as to the identifiability of typical causal effect
measures, such as the average causal effect.
While confounding is a problem of any observational study and has been
widely addressed in the causal inference literature, points (2) and (3) are
more specific to case–control studies and will be the focus here. I will
specifically consider: identification of the null-hypothesis of no causal
effect which is closely related to the non-parametric identification of
causal odds ratios; further I will address how certain structural knowledge
can enable us to reconstruct the full joint distribution; finally I will
briefly compare these approaches with those that rely on additional
knowledge of the population prevalence, such as standardisation, propensity
scores, and instrumental variables.


The DEC statistics seminars schedule is available at
http://www.unibocconi.eu/statseminar

Kind regards,
*Raffaella Piccarreta and Isadora Antoniano*



*ATTENTION:*

If you are a guest and you do not have a Bocconi ID Card to access to the
Bocconi Buildings, please confirm your participation by sending an email to
simona.gagino a unibocconi.it
In addition, we suggest to print, fill out and send the attached form (to
give your consent to personal data handling) via email to the same address.
In this way, the check-in procedure at the entrance will be faster and
easier.


***************************************************
Isadora Antoniano Villalobos

Assistant Professor
DEC - Department of Decision Sciences,
Bocconi University, via Guglielmo Röntgen 1 (3rd floor, room D1-16). 20136,
Milano.
email: isadora.antoniano a unibocconi.it

Web page: http://faculty.unibocconi.eu/isadoraantonianovillalobos
***************************************************
-------------- parte successiva --------------
Un allegato HTML è stato rimosso...
URL: <http://www.stat.unipg.it/pipermail/sis/attachments/20161021/6f703fe0/attachment.html>