The statistical information contained in additional observations
Let EnEn\mathscr{E}^n be a statistical experiment based on nnn i.i.d. observations. We compare EnEn\mathscr{E}^n with En+rnEn+rn\mathscr{E}^{n+r_n}. The gain of information due to the rnrnr_n additional observations is measured by the deficiency distance Δ(En,En+rn)Δ(En,En+rn)\Delta (\mathscr{E}^n,...
Saved in:
| Main Author: | |
|---|---|
| Format: | Article (Journal) |
| Language: | English |
| Published: |
1986
|
| In: |
The annals of statistics
Year: 1986, Volume: 14, Issue: 2, Pages: 665-678 |
| ISSN: | 2168-8966 |
| DOI: | 10.1214/aos/1176349945 |
| Online Access: | Volltext Volltext Volltext 
|
| Author Notes: | Enno Mammen |
| Summary: | Let EnEn\mathscr{E}^n be a statistical experiment based on nnn i.i.d. observations. We compare EnEn\mathscr{E}^n with En+rnEn+rn\mathscr{E}^{n+r_n}. The gain of information due to the rnrnr_n additional observations is measured by the deficiency distance Δ(En,En+rn)Δ(En,En+rn)\Delta (\mathscr{E}^n, \mathscr{E}^{n+r_n}), i.e., the maximum diminution of the risk functions. We show that under general dimensionality conditions Δ(En,En+rn)Δ(En,En+rn)\Delta(\mathscr{E}^n, \mathscr{E}^{n+r_n}) is of order rn/nrn/nr_n/n. Further the behavior of ΔΔ\Delta is studied and compared for asymptotically Gaussian experiments. We show that the information gain increases logarithmically. The Gaussian and the binomial family turn out to be--in some sense--opposite extreme cases, with the increase of information asymptotically minimal in the Gaussian case and maximal in the binomial. |
|---|---|
| Item Description: | First available in Project Euclid: 12 April 2007 Gesehen am 01.03.2018 |
| Physical Description: | Online Resource |
| ISSN: | 2168-8966 |
| DOI: | 10.1214/aos/1176349945 |