Comment Risto Lehtonen University of Helsinki Quality 2010 Conference, 4-6 May 2010, Helsinki.

Comment

Risto LehtonenUniversity of Helsinki

Quality 2010 Conference, 4-6 May 2010, Helsinki

First message (to me) from Carl-Erik Särndal’s talk

The method of classical probability sampling does not work anymore in governmental social surveys because of increasing nonresponse rates (and diverging patterns of nonresponse)

2

Probability sampling tradition…

3

Even a stronger message: Paradigm shift

The paradigm of classical probability sampling is shifting / has shifted in the context of Official statistics

4

Paradigm shiftMore complex strategies have

gradually superseded the strategies relying solely on probability sampling

BUT: Probability sampling still constitutes a component in the new strategies

5

Looking beyond Official statistics… serious questions have arisen…

6

…in online survey methods…

No data are better than bad data?

7

…also referring to Official stats

8

Signs of paradigm shift

Mainstream statistics and survey sampling are more and more interacting and learning from each other

9

Signs of paradigm shift…

10

Signs of paradigm shift…

11

Signs of paradigm shift…

12

Dominating paradigm in Official statistics

Design-based model-assisted survey sampling

13

…proposal for new paradigm

14

...a possible new paradigm…

Register-assisted survey sampling

Forthcoming paradigm shift?

15

High policy relevance of paradigm shifts

16