Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Features
Speaker Deck
PRO
Sign in
Sign up for free
Search
Search
p値を巡る最近の論争 / ~ Moving to a World Beyond “p < 0....
Search
Tomoshige Nakamura
May 17, 2019
Science
0
1.3k
p値を巡る最近の論争 / ~ Moving to a World Beyond “p < 0.05” ~
近年、The American Statisticianで特集が組まれるほど、問題視されている「p値の誤用問題」について、第3回ヘルスデータアナリティクス・マネジメント研究会で発表したスライドです。
Tomoshige Nakamura
May 17, 2019
Tweet
Share
More Decks by Tomoshige Nakamura
See All by Tomoshige Nakamura
一般化ランダムフォレストの理論と統計的因果推論への応用
tomoshige_n
11
2.9k
ランダムフォレストによる因果推論と最近の展開
tomoshige_n
12
9.5k
傾向スコアのモデルに含める共変量選択のアプローチ
tomoshige_n
2
2k
統計的因果推論とデータ解析 / causal-inference-and-data-analysis
tomoshige_n
32
73k
Other Decks in Science
See All in Science
Direct Preference Optimization
zchenry
0
280
Snowflake上でRを使う: RStudioセットアップとShinyアプリケーションのデプロイ
ktatsuya
0
420
ultraArmをモニター提供してもらった話
miura55
0
190
『データ可視化学入門』を PythonからRに翻訳した話
bob3bob3
1
500
はじめてのバックドア基準:あるいは、重回帰分析の偏回帰係数を因果効果の推定値として解釈してよいのか問題
takehikoihayashi
2
750
Machine Learning for Materials (Lecture 8)
aronwalsh
0
410
Improving Search @scale with efficient query experimentation @BerlinBuzzwords 2024
searchhub
0
240
Transformers are Universal in Context Learners
gpeyre
0
550
拡散モデルの概要 −§2. スコアベースモデルについて−
nearme_tech
PRO
0
580
Cross-Media Information Spaces and Architectures (CISA)
signer
PRO
3
29k
(論文読み)贈り物の交換による地位の競争と社会構造の変化 - 文化人類学への統計物理学的アプローチ -
__ymgc__
1
110
教師なしテンソル分解に基づく、有糸分裂後の転写再活性化におけるヒストン修飾ブックマークとしての転写因子候補の抽出法
tagtag
0
120
Featured
See All Featured
Evolution of real-time – Irina Nazarova, EuRuKo, 2024
irinanazarova
4
380
Fight the Zombie Pattern Library - RWD Summit 2016
marcelosomers
232
17k
How to Ace a Technical Interview
jacobian
276
23k
"I'm Feeling Lucky" - Building Great Search Experiences for Today's Users (#IAC19)
danielanewman
226
22k
No one is an island. Learnings from fostering a developers community.
thoeni
19
3k
Being A Developer After 40
akosma
87
590k
What’s in a name? Adding method to the madness
productmarketing
PRO
22
3.1k
Docker and Python
trallard
40
3.1k
5 minutes of I Can Smell Your CMS
philhawksworth
202
19k
GitHub's CSS Performance
jonrohan
1030
460k
Raft: Consensus for Rubyists
vanstee
136
6.6k
The Cult of Friendly URLs
andyhume
78
6k
Transcript
1 ౷ܭత༗ҙࠩ Q Λ८Δ࠷ۙͷ૪ɿ .PWJOHUPB8PSME#FZPOElQ z தଜൟ ܚጯٛक़େֶେֶӃ UPNPTIJHFOBLBNVSB!HNBJMDPN 6QEBUFEPO.BZUI
ܚጯٛक़େֶࡾాΩϟϯύε
࣍ ‣ ΠϯτϩμΫγϣϯ ‣ ͜ͷൃදͰѻ͏༰ͱɼൃදͷαϚϦʔ ‣ Qͷ᠘ɻ ‣ QͱԿ͔ʁ ‣
QΛ͏᠘ʮݩͷఆʯͰ͋Δʢ͕͜͜ϙΠϯτʂʣ ‣ Qʹର͢Δ"4"ͷ1SJODJQMF ‣ QͷΛड͚ͯɼԿΛ͖͔͢ʢ.PWJOHUPBXPSMECFZPOElQzʣ ‣ ʮ౷ܭత༗ҙʯ͔Βͷ٫ ‣ "50.ͱ͍͏ݪଇɽ ‣ ݱ࣮తͳΨΠυϥΠϯͱͯ͠ ‣ ࢲ͕ͨͪɺʮQͷੈքʯ͔Β٫͢ΔͨΊʹ !2
࣍ ‣ ΠϯτϩμΫγϣϯ ‣ ͜ͷൃදͰѻ͏༰ͱɼൃදͷαϚϦʔ ‣ Qͷ᠘ɻ ‣ QͱԿ͔ʁ ‣
QΛ͏᠘ʮݩͷఆʯͰ͋Δʢ͕͜͜ϙΠϯτʂʣ ‣ Qʹର͢Δ"4"ͷ1SJODJQMF ‣ QͷΛड͚ͯɼԿΛ͖͔͢ʢ.PWJOHUPBXPSMECFZPOElQzʣ ‣ ʮ౷ܭత༗ҙʯ͔Βͷ٫ ‣ "50.ͱ͍͏ݪଇɽ ‣ ݱ࣮తͳΨΠυϥΠϯͱͯ͠ ‣ ࢲ͕ͨͪɺʮQͷੈքʯ͔Β٫͢ΔͨΊʹ !3
͜ͷϓϨθϯςʔγϣϯͰѻ͏༰ ‣ ͜ͷϓϨθϯςʔγϣϯͰɼҎԼͷจࢽͷͷ ಛूʹ͋Δɹ࣍ͷจΛѻ͍·͢ ‣ จࢽɿ5IF"NFSJDBO4UBUJTUJDJBO ‣ 7PMVNF ‣ ಛूɿ4UBUJTUJDBM*OGFSFODFJOUIFTU$FOUVSZ"
8PSME#FZPOEQ ‣ จɿ8BTTFSTUFJO3 4DIJSN" BOE-B[BS/ .PWJOHUPB8PSME#FZPOElQ z ‣ จͷ༰ΠϯλʔωοτͰΞΫηε͢ΕӾཡՄೳͳ ͷͰɼৄࡉͳ༰ʹ͍ͭͯɼࣗ͝Ͱ͓͔֬Ίͩ͘͞ ͍ɽ ‣ ͜ͷൃදͰɼۙߦΘΕ͍ͯΔʮQʹ͍ͭͯͷٞʯΛ ͰίϯύΫτʹઆ໌͢Δ͜ͱΛඪͱ͠·͢ɽ !4
͜ͷϓϨθϯςʔγϣϯͰѻ͏༰ ‣ ·ͨɺຊൃදͷ༰݄ʹ/BUVSFͰൃද͞Ε্ͨهͷهࣄͷ༰Λ ؚΈ·͢ɻ ‣ ͜ͷهࣄɺֶज़తͳݚڀʹ͓͍ͯ1ͷෆదͳ༻ʢओʹɺ1ΛԾઆͷ ཱূͷࠜڌͱ͢Δ͜ͱʣʹରͯ͠ͷܯΛ໐Β͢ͷͰ͢ɻ !5
ˎ ‣ εϥΠυͰɺQʹ͍ͭͯͷਖ਼͍͠ཧղͷͨΊʹɺվΊͯQͱͳʹ͔ʹ ͍ͭͯهࡌͨ͠εϥΠυ͕͋Γ·͕͢ɺ۩ମతʹઆ໌͢Δ༧ఆ͋Γ· ͤΜɻ ‣ ࠓճհ͢ΔจͰͷQޡ༻ͷఆֶज़తͳݚڀͰ͕͢ɺࠓճͷΠϕ ϯτ͓ӽ͠ͷօ༷ͷதʹɺϏδωεͰ༻͢Δํʑ͍Βͬ͠ΌΔͱࢥ ͍·͢ɻΑͬͯɺจͷ༰Λͦͷ··͢ͷͰͳ͘ɺඞཁͳՕॴͷΈந ग़ͯ͠ɺൃදऀͷҙݟΛՃ͑ͯهड़͍ͯ͠·͢ɻ
‣ Ҏ্ͷʹɺྃ͝ঝ·͢Α͏͓ئ͍͍ͨ͠·͢ɻ !6
ൃදͷαϚϦʔ* ‣ ʹ৺ཧֶܥͷจࢽ#BTJDBOE"QQMJFE4PDJBM1TZDIPMPHZͰɼQ ͷ༻͕ېࢭ͞Εͨɽ ‣ ͦͷ͋ͱɼ/BUVSFɼ4DJFODF/FXT 4UBUJTUJDJBOͳͲͰɼQʹجͮ͘Պֶత ͳจɼ࠶ݱੑͷอূͳͲͷ͔Βܯ͕໐Β͞Ε·ͨ͠ɽ ‣ ʹɼ"4"ʢΞϝϦΧ౷ܭֶձʣ͕Qʹؔ͢Δ໌ΛൃදɽQͷ
༻ʹؔ͢ΔͭͷݪଇΛఏࣔɽ ‣ ݄ͷ"NFSJDBO4UBUJTUJBOͰʮQ͔Β٫͠ɼσʔλղੳ ࣍ͷεςʔδʯͱ͍͏ಛू͕·ΕΔɽ ‣ QΛ༻͍ͨೋݩతͳൃ͔Β٫͠Α͏ʂ ‣ ʮ౷ܭతʹ༗ҙʯͱ͍͏ݴ༿ΛࣺͯΑ͏ʂ ‣ QΛཧ༝ʹͨ͠ʮ݁ʯΛΊΑ͏ʂ ‣ Q͕খ͘͞Ͱؔ࿈͕ͳ͍߹͋ΕɼQ͕େ͖ͯؔ͘࿈͕͋Δ ߹͋Δɽ ‣ Qɼֶज़తʹҙຯ͕͋Δ͜ͱͱৗʹಉٛͰͳ͍ !7 ͜ ͜ ͷ ྲྀ Ε จ ͷ ֓ ུ
ൃදͷαϚϦʔ** ‣ ʮ౷ܭతʹ༗ҙʯͱ͍͏ݴ༿͕Ռׂ͖ͨͯͨ͠ʹมΘΔͷଘࡏ͠ͳ͍ ͕ɼࢲͨͪ࣍ͷͭΛҙࣝͯ͠ߦಈ͍ͯ͘͠ඞཁ͕͋Δɽ ‣ "DDFQU6ODFSUBJOUZʢσʔλղੳͷෆ࣮֬ੑΛड͚ೖΕΔ͜ͱʣ ‣ #F5IPVHIUGVMʢσʔλͷղੳʹରͯ͠৻ॏͰɼࢥྀਂ͋͘Δ͜ͱʣ ‣ #F0QFOʢσʔλͷղੳͷϓϩηεఆͳͲΛެ։͢Δ͜ͱʣ
‣ #F.PEFTUʢσʔλͷղੳͷ݁Ռʹ͍ͭͯݠڏͰ͋Δ͜ͱʣ ‣ σʔλͷղੳΛධՁ͢ΔࡍͷϙΠϯτม͍͑ͯ͘ඞཁ͕͋Δɽ ‣ ಘΒΕͨσʔλղੳͷ݁Ռɼ࠶ݱՄೳͳͷ͔ʢݚڀతʹରͯ͠ɼσ βΠϯ͕దʹͳ͞Ε͍ͯΔ͔ɼ·ͨख๏ͷબͷϩδοΫద͔ʣ ‣ σʔλղੳͷՁʮ݁Ռʯʹ͋ΔͷͰͳ͍ɽղੳΛߦͬͨʮલఏʯɼ ͦͷաఔͰஔ͔ΕͨʮԾઆʯेʹٞ͞Εͨͷ͔ɽ ‣ ྫ͑ɼϞσϧબͰతʹ4UFQXJTF3FHSFTTJPOΛ͍ͯͨ͠Βɼ ͦΕϞσϧʹରͯ͠ͷྀ͕໌Β͔ʹΓͳ͍ͱࢥͬͯྑ͍ɽ !8 จ ͷ ֓ ུ
ൃදͷαϚϦʔ*** ‣ ͜ͷಛू߸ʹدߘ͍ͯ͠Δͷɼ΄ͱΜͲ͕ݚڀऀͳͷͰɼ༰ͷεϙοτ ಛʹݚڀͱ͚ΒΕ͍ͯΔɽ ‣ ࣮ࡍɼ࣏ݧͳͲʹ͓͍ͯQͷ༗༻ੑʹ͍ͭͯɼ༄ ͳͲͰड़ ΒΕ͍ͯΔ௨Γɼঢ়گ࣍ୈͰ͋Δɽ ‣
͔͠͠ͳ͕ΒɼʮQͷޡ༻ʯͱʮ౷ܭత༗ҙͷཚ༻ʯɼσʔλղੳΛߦͬ ͨࡍͷϨϙʔτͰʑʹ͢Δɽ ‣ ྫ͑ɺճؼϞσϧʹ͓͚ΔʮQʯʮ"*$ʯͷཚ༻͕ͦͷҰྫͰ͋ Δɻ ‣ 4UFQXJTF๏-േଇਖ਼ଇԽͰਖ਼͍͠Ϟσϧ͕બͰ͖͍ͯΔͱ͍͏Α͏ ͳʮա৴ʯ͕ຮԆ͍ͯ͠Δɽ ‣ ࠓͦ͜ɺσʔλͷऔಘ͔ΒɼϞσϧͷߏஙɼϨϙʔτͷ࡞·Ͱͷɼσʔλ ղੳͷϑϩʔΛݟ͠ɼࣗͨͪͷσʔλղੳʹదͨ͠ղੳϑϩʔΛ࡞Γ ͢ඞཁ͕͋Δɽ !9 ࢲ ͨ ͪ Ͳ ͏ ͢ Δ ͔ Moving to a World Beyond “p < 0.05”
࣍ ‣ ΠϯτϩμΫγϣϯ ‣ ͜ͷൃදͰѻ͏༰ͱɼൃදͷαϚϦʔ ‣ Qͷ᠘ɻ ‣ QͱԿ͔ʁ ‣
QΛ͏᠘ʮݩͷఆʯͰ͋Δʢ͕͜͜ϙΠϯτʂʣ ‣ Qʹର͢Δ"4"ͷ1SJODJQMF ‣ QͷΛड͚ͯɼԿΛ͖͔͢ʢ.PWJOHUPBXPSMECFZPOElQzʣ ‣ ʮ౷ܭత༗ҙʯ͔Βͷ٫ ‣ "50.ͱ͍͏ݪଇɽ ‣ ݱ࣮తͳΨΠυϥΠϯͱͯ͠ʢൃදऀͷҙݟʣ ‣ ࢲ͕ͨͪɺʮQͷੈքʯ͔Β٫͢ΔͨΊʹ !10
QͱԿ͔*ʢԾઆݕఆ֓આʣ ‣ ౷ܭతԾઆݕఆͱʁ ‣ ؼແԾઆΛغ٫͠ରཱԾઆΛࢧ࣋͢Δ͔ɼຢؼແԾઆΛغ٫͠ͳ͍͔Λ ؍ଌʹج͍ܾͮͯΊΔͨΊͷ౷ܭతखଓ͖ɻ ‣ ͦͷखଓ͖ɼؼແԾઆཱ͕͍ͯ͠Δʹ͔͔ΘΒͣغ٫͢Δ͕֬Ћ ҎԼʹͳΔΑ͏ʹܾΊΒΕΔɻ͜ͷЋΛ༗ҙਫ४ͱ͍͏ɻ ‣
1ͱʁ ‣ ༩͑ΒΕͨσʔλͷʹରͯ͠ɺؼແԾઆΛغ٫Ͱ͖Δ࠷খͷ༗ҙਫ४ ‣ 1ͱʁʢטΈࡅ͘ͱʣ ‣ 1ͱɺಛఆͷ౷ܭϞσϧͷͱͰɺσʔλͷ౷ܭతͳཁʢྫ͑ ͭͷ܈ͷฏۉͷࠩɺճؼͷਪఆʣ͕ໃ६͢ΔఔΛࣔ͢ࢦඪʂ ‣ ࢿྉʹɺཧతͳ1ͷఆ͕ٛॻ͔Ε͍ͯΔɻ !11
QͱԿ͔**ʢԾઆݕఆͷྫɿճؼʣ ‣ &YBNQMFઢܗճؼϞσϧ ‣ ྫ͑ɺσʔλʹઢܗճؼϞσϧΛͯΊͨ߹ͷɺճؼЌʹର͢Δ ݕఆͰɺɹɹɹɹɹɹ͓ΑͼɹɹɹɹɹɹͰ͋Δɻ ‣ ͜ͷͱ͖ɺ༗ҙਫ४ͷݕఆʢQͰ༗ҙͱ͢ΔݕఆʣɺؼແԾઆ ͕ਖ਼͍͠ʢЌʣͳͷʹɺЌͰͳ͍ͱͯ͠͠·͏֬ΛҎԼʹ͑ ΔΑ͏ͳݕఆΛߦ͍ͬͯΔ͜ͱʹରԠ͢Δɻ
‣ ҙ͖͢͜ͱ ‣ 5ZQF**&SSPSʹ͍ͭͯԿݴٴ͍ͯ͠ͳ͍ɻ ‣ σʔλऔಘɺϞσϧͷਖ਼͠͞ʹ͍ͭͯҰݴٴ͞Ε͍ͯͳ͍ɻ ‣ Өڹͷେ͖͞ʹ͍ͭͯҰݴٴ͠ͳ͍ɻ !12 H0 : β = 0 H1 : β ≠ 0
QΛ͏᠘ʮݩͷఆʯͰ͋Δ* ‣ Ծઆݕఆͷཧͷ݁ՌɺҰൠతʹ ‣ ɹɹɹɹɹ͕౷ܭϞσϧɹɹɹɹɹɹɹɹɹɹɹɹ͔ΒϥϯμϜͳඪຊ͕ ಘΒΕͨͱԾఆ͢Δɻ ‣ ͱ͍͏ຐ๏ͷ͜ͱ͕࠷ॳʹ͍͍ͭͯΔɻ ‣ σʔλղੳͰɺ࣍ͷ͕ͭΘ͔Βͳ͍ɻ
‣ σʔλ͕ɺຊʹΓ͍ͨूஂ͔ΒϥϯμϜʹऔΒΕ͍ͯΔ͔Ͳ͏͔ ‣ σʔλ͕ɺͲΜͳϞσϧ͔Βੜ͞Ε͔ͨ ‣ ࣮ࡍͷղੳͰɺຐ๏ͷ͜ͱͷେલఏ͔Βٙ͏ඞཁ͕͋Δɻ ‣ ਪఆྔɺͦͷݕఆɺ͜ΕΒʹ͍ͭͯेͳۛຯͱߟͷ্ʹ͔͠ҙຯΛ ࣋ͨͳ͍ɻ !13 X1 , . . . , XN {f(x; θ) ; θ ∈ Θ ⊂ R} ݱ࣮Ͳ͏ͩʂʁ
QΛ͏᠘ʮݩͷఆʯͰ͋Δ** ‣ ʲྑ͘ͳ͍ղੳϨϙʔτͷྫʳ ‣ σʔλͷऔಘํ๏ɾղੳϞσϧͷଥੑʹ͍ͭͯͷใࠂ͕Ͱ͖͍ͯͳ͍ ‣ 1ɺಛఆͷ౷ܭతͳج४͕ߴ͍͜ͱΛɺஅͷཧ༝ͱ͍ͯ͠Δɻ ‣ ͜ͷΑ͏ͳจҰఆଘࡏ͍ͯͯ͠ɺ·ͨۀʹ͓͚ΔσʔλղੳͷϨ ϙʔτͰࢄݟ͞ΕΔɻ
!14 ʲݕূʳ 1ͦΜͳʹ৴༻Ͱ͖Δͷ͔ʂʁ
QΛ͏᠘ʮݩͷఆʯͰ͋Δ*** &YBNQMFͦͷ̍ ‣ ҎԼͷϞσϧ͔ΒɺσʔλΛൃੜͤ͞Δɻ ‣ ݁Ռมʹର͢ΔϞσϧɺ ‣ ͜ͷͱ͖ɺαϯϓϧΛͱͯ͠ɺճ܁Γฦ͠σʔλΛൃੜͤͨ͞ɻ ‣ ൃੜͤͨ͞σʔλʹɺͯ͢ͷ9Λઆ໌มͱͨ͠ઢܗճؼϞσϧΛͯΊ
ͯɺճؼΛਪఆ͠ɺQΛܭࢉͯ͠ɺΛԼճͬͨճΛɺҎԼͷද ʹ·ͱΊͨɻ !15 X1 ∼ N(0,1) X2 = X1 + N(0,1) X3 ∼ Bernoulli(expit(X1 + X2 )) X4 = |X1 + X3 X2 | Y = 0.2X2 + 0.2X3 + N(0,1) X2 X1 X3 X4 ย ༗ҙʹͳͬͨ ճ ܁Γฦ͠ճճதɺQͰ༗ҙʹͳͬͨճ
QΛ͏᠘ʮݩͷఆʯͰ͋Δ*** &YBNQMFͦͷ ‣ ղੳΛ͍ͯͯ͠ɺɹ͕ޮՌ͕͋ΔΑ͏ͳ݁Ռ͕΄͍͠ͱࢥͬͨʂ ‣ มΛൈ͍ͯΈΑ͏ɻɻɻ ‣ ݁Ռมʹର͢ΔϞσϧʹؚ·Εͳ͍ม͕༗ҙʹͳͬͨɾɾɾ ‣ αϯϓϧΛߋʹ૿ͯ͠ɺ/ʹͨ͠Β
‣ ؆୯ʹQΛԼճΔճΛ૿͢͜ͱ͕Ͱ͖ͨɻͭ·ΓɺQͳΜͯϞσ ϧͷԾఆͱαϯϓϧͰ͍͔Α͏ʹίϯτϩʔϧՄೳɻ/ͻͲ ͍ɻ !16 X4 / X1 X3 X4 ย ༗ҙʹͳͬͨ ճ / X1 X3 X4 ย ༗ҙʹͳͬͨ ճ / X1 X3 X4 ย ༗ҙʹͳͬͨ ճ
17 ༗ҙʹ͍ͨ͠มɺ༗ҙʹͰ͖Δʂ Ϗοάσʔλ࠷ߴͰ͢Ͷὑʢൽʣ
Qʹର͢Δ"4"ͷQSJODJQMF 8BTTFSTUFJO3- -B[BS/" 1ɺσʔλͱ͕ࣗԾఆͨ͠౷ܭϞσϧͷໃ६ͷఔΛࣔ͢Ͱ͋Δɻ 1ɺௐ͍ͯΔԾઆ͕ਖ਼͍֬͠ɺσʔλ͕ۮવͷΈͰ͑ΒΕͨ֬ ΛଌΔͷͰͳ͍ɻ
Պֶతͳ݁ɺϏδωεࡦʹ͓͚Δܾఆɺ1͕͋ΔᮢΛ͔͑ͨ Ͳ͏͔Λࠜڌʹ͢Δ͖Ͱͳ͍ɻ దͳ౷ܭతਪͷͨΊʹɺݚڀͷதͰௐΔԾઆͷɺσʔλऩूͷࡍ ʹߦͬͨͯ͢ͷܾఆɺ࣮ߦͨͯ͢͠ͷ౷ܭղੳɺͦͯ͠ܭࢉͨͯ͢͠ ͷ1Λݚڀऀ։͖ࣔ͢Ͱ͋Δɻ 1౷ܭత༗ҙੑɺޮՌͷେ͖݁͞ՌͷॏཁੑΛҙຯ͠ͳ͍ɻ 1ɺͦΕ͚ͩͰ౷ܭϞσϧԾઆʹؔ͢ΔΤϏσϯεͷɺΑ͍ࢦඪͱ ͳΒͳ͍ɻ !18
࣍ ‣ ΠϯτϩμΫγϣϯ ‣ ͜ͷൃදͰѻ͏༰ͱɼൃදͷαϚϦʔ ‣ Qͷ᠘ɻ ‣ QͱԿ͔ʁ ‣
QΛ͏᠘ʮݩͷఆʯͰ͋Δʢ͕͜͜ϙΠϯτʂʣ ‣ Qʹର͢Δ"4"ͷ1SJODJQMF ‣ QͷΛड͚ͯɼԿΛ͖͔͢ʢ.PWJOHUPBXPSMECFZPOElQzʣ ‣ ʮ౷ܭత༗ҙʯ͔Βͷ٫ ‣ "50.ͱ͍͏ݪଇɽ ‣ ݱ࣮తͳΨΠυϥΠϯͱͯ͠ʢൃදऀͷҙݟʣ ‣ ࢲ͕ͨͪɺʮQͷੈքʯ͔Β٫͢ΔͨΊʹ !19
౷ܭత༗ҙ͔Βͷ٫* ‣ &EHFXPSUI 'JTIFS Ͱ1ɺ݁ՌͷߟͷͨΊͷಓ۩ʢ1 ͷͦͷͷΛؾʹ͍ͯͨ͠ʣ ‣ /FZNBO1FBSTPOͷ౷ܭతԾઆݕఆ͕·Γɺ1͕ੜ·Εʮ༗ҙͰ͋
Δ͜ͱʯʮ4UBUJTUJDBMMZ4JHOJpDBOUʯͱ͍͏ݴ༿ੜ·ΕΔɻ ‣ ͋Δਫ४ΛԼճͬͨʢ༗ҙʣʮҙຯ͕͋Δʯͱ͍͏ಾͷ͕ؔੜ·ΕΔɻ ‣ ͍·ɺ༗ҙ͡Όͳ͚ΕɺQVCMJTI͠ͳ͍ͱ͍͏ѱ͍෩ைʹͳΓɺՊֶతͳ จͰग़൛͞ΕΔͷ༗ҙͳͷ͔Γɾɾɾʢग़൛όΠΞεʣ ‣ ͜ͷΑ͏ͳޡΓΛͳͨ͘͢Ίʹɺʮ4UBUJTUJDBMMZ4JHOJpDBOUʢ౷ܭతʹ༗ ҙʣΘͳ͍ʯΑ͏ʹ͠Α͏ʂʢ8BTTFSTUBJO ɻ ‣ ࣍ͷ̐ͭͷݪଇʢ"50.ʣʹج͍ͮͨղੳͱղੳͷධՁΛਪ͢Δɻ !20 1 ޡ ༻ ͷ ྲྀ Ε ౷ܭతʹ༗ҙ ≠ ॏཁͳ݁Ռ จ ͷ ओ ு
"50.ͱ͍͏ݪଇ* ‣ "DDFQU6ODFSUBJOUZʢෆ࣮֬ੑΛड͚ೖΕΔʣ ‣ σʔλͷऔಘํ๏ϞσϧͷԾఆ࣍ୈͰɺղੳͷ݁Ռมಈ͢Δɻ ‣ ղੳͷ݁ՌʹɺΒ͖͕ͭ͋Δʢ͖ͪΜͱهࡌʣɻ ‣ Ґਪఆͷࢄਪఆͷ৴པ۠ؒΛඞͣॻ͘ ‣
#F5IPVHIUGVMʢࢥྀਂ͘ʣ ‣ ʲσʔλͷղੳऀ͕ҙ͖ࣝ͢͜ͱʳ ‣ ௐࠪղੳͷҙਤ ‣ ҙຯͷ͋ΔޮՌͷେ͖͞ ‣ తʹର͢ΔɺσʔλΛऔಘํ๏ͷద͞ɻ ‣ σʔλʹͯΊΔख๏ͱɺख๏ͷ౷ܭతੑ࣭ͷཧղɻ ‣ ྫ͑ɺઢܗճؼϞσϧͰ͍͑ɺࢄੑ !21
"50.ͱ͍͏ݪଇ** ‣ #F5IPVHIUGVMʢࢥྀਂ͘ʣଓ͖ ‣ ʲղੳϨϙʔτΛݟΔଆ͕ҙࣝ͢Δ͖͜ͱʳ ‣ ಘΒΕͨਪఆͷɺ࣮ࡍతͳ࣮༻తͳҙຯ ‣ ਪఆͷਖ਼֬͞ʢΒ͖ͭʣ ‣
༻ͨ͠ϞσϧͷԾఆͷదੑ ‣ ղੳऀͷϞσϧʹର͢Δཧղ ‣ ෳϞσϧͷൺֱͷ༗ແൺֱͨ͠߹ͷ݁ՌͷมԽͱߟ ‣ Ҏ্Λɺ࠶ݱͰ͖ΔϨϕϧͰɺϨϙʔτʹ·ͱΊΒΕ͍ͯΔ͔ !22 ʲϙΠϯτʳ σʔλղੳͷධՁɺQͦͷଞͷ౷ܭతईͰߦΘΕΔ͖Ͱͳ͍ɻ ઌߦݚڀௐࠪͷਂ͞ݚڀσβΠϯͱσʔλͷ࣭Ծఆͨ͠ϝΧχζϜͷ ଥੑݱ࣮తͳՁൃݟͷ৽نੑΛ૯߹ͯ͠அ͖͢ ʲΞΠσΞʳ ݁ՌΛCMJOEͯ͠ϨϙʔτΛಡΜͰՁΛஅ͢Δͷͭͷํ๏Ͱ͋Δ
"50.ͱ͍͏ݪଇ*** ‣ #FUIPVHIUGVMʢࢥྀਂ͘ʣଓ͖ ‣ ϞσϧͷଥੑධՁʹ͍ͭͯɺQҎ֎ʹఏҊ͞Ε͍Δ ‣ ϕΠζҼࢠୈ̎ੈQʢ4FDPOE(FOFSBUJPOQWBMVFʣͳͲ ‣ ͨͩ͠ɺͦΕୈ̎ͷQΛੜΈग़͢ͷͰ͋ͬͯͳΒͳ͍ ‣
#F0QFOʢެ։͢Δʣ ‣ σʔλղੳͷՁɺσʔλͦͷͷͷ٬؍ੑɻ ‣ ͔͠͠ɺσʔλͷऔಘ͔ΒɺղੳʹࢸΔ·ͰɺղੳऀઐՈͷʮओ؍ੑʯ ʮஅʯʹґଘ͢Δ෦͕େ͖͍ɻ ‣ ݁ՌΛಡΈղͨ͘Ίʹɺഎޙʹ͋Δʮߟ͑ํʯ͕ඞཁͰɺͦΕͳ͠ʹୈ ऀతͳϨϏϡʔ͕͘͠ɺ٬؍ੑ͕ଛΘΕΔɻ ‣ ղੳͷ٬؍ੑΛอ࣋͢ΔͨΊʹɺσʔλղੳͷϓϩηεΛެ։͠ɺஔ ͔Ε͍ͯΔఆͳͲʹ͍ͭͯৄࡉͳϨϙʔτΛ࡞͠ใࠂ͢Δ͖ !23
"50.ͱ͍͏ݪଇ*7 ‣ #F.PEFTUʢݠڏͰ͋Ζ͏ʣ ‣ ౷ܭతͳख๏ʹɺͦͦݶք͕͋Δ͜ͱΛཧղ͢Δɻ ‣ ౷ܭϞσϧෳࡶͳݱ࣮Λ࣮ʹ࠶ݱ͢Δख๏Ͱͳ͘ɺΉ͠Ζݱͷ ʮ؆қܕʯɻ ‣ ༻͞ΕͨϞσϧ͕ʮਅͷϞσϧʯͰͳ͍͜ͱΛཧղ͓ͯ͘͠
‣ ಘΒΕͨ݁ՌɺʮϞσϧ͕ਖ਼͍͠ʂʯͱ͍͏ԾఆͷͱͰ͔͠ҙຯͷͳ ͍ͷ ‣ ࣗͷग़ͨ͠ղੳ݁Ռɺ݁ʹରͯ͠ʮͲΜͳؒҧ͍ͷՄೳੑ͕͋Δ ͔ʯΛߟ͓͑ͯ͘ඞཁ͕͋Δɻ ‣ ݚڀʹ͓͍ͯɺಉ͡ςʔϚʹରͯ͠ɺෳͷಉ༷ͷݚڀ͕ߦΘΕͯɺಉ ͡Α͏ͳ݁Ռ͕ಘΒΕͯɺ͡ΊܾͯఆతͳͷͱͳΔɻ ‣ ݚڀऀɺ࠶ݱੑΛอূ͢ΔΑ͏ͳݚڀΛྭ͖͢Ͱ͋ΓɺͦͷҙຯͰ ݚڀͷखॱɺ༻ͨ͠σʔλʹ͍ͭͯɺެදΛߦ͏͖Ͱ͋Δɻ !24
࣍ ‣ ΠϯτϩμΫγϣϯ ‣ ͜ͷൃදͰѻ͏༰ͱɼൃදͷαϚϦʔ ‣ Qͷ᠘ɻ ‣ QͱԿ͔ʁ ‣
QΛ͏᠘ʮݩͷఆʯͰ͋Δʢ͕͜͜ϙΠϯτʂʣ ‣ Qʹର͢Δ"4"ͷ1SJODJQMF ‣ QͷΛड͚ͯɼԿΛ͖͔͢ʢ.PWJOHUPBXPSMECFZPOElQzʣ ‣ ʮ౷ܭత༗ҙʯ͔Βͷ٫ ‣ "50.ͱ͍͏ݪଇɽ ‣ ݱ࣮తͳΨΠυϥΠϯͱͯ͠ʢൃදऀͷҙݟʣ ‣ ࢲ͕ͨͪɺʮQͷੈքʯ͔Β٫͢ΔͨΊʹ !25
ࢲͨͪͲ͏͢Ε͍͍ͩΖ͏ʁ ‣ ͕ࣗσʔλղੳ͍ͯ͠ΔτϐοΫʹ͓͍ͯɺʮ1ʹՁ͕͋Δ͔Ͳ͏ ͔ʁʯͱ͍͏࣭Λߟ͑Δ͜ͱͰ͋Δɻ ‣ ͦͷͨΊʹͭͷ࣭ʹ͢Δɻ ‣ σʔλऔಘɺ͖ͪΜͱσβΠϯ͞Ε͓ͯΓɺ3$5ʢϥϯμϜԽൺֱ࣮ ݧʣͰ͋Δ͔ʁ ‣
ղੳʹ༻͍ΒΕΔϞσϧͷԾఆଥ͋Δ͜ͱ͕ɺઌߦจݙͳͲͷ݁Ռ͔ Β໌Β͔Ͱ͋Δ͔ʁ ‣ ͍ͣΕ͔Ұํ͕/0Ͱ͋ΔͳΒɺ1Λར༻ͨ͠Ծઆݕఆͷ݁Ռ͔Βɺ҆қʹ ݁Λಋ͘ͷదͱݴ͑ͳ͍ɻ ‣ ͳͥͳΒɺ1͕ʮҙຯΛ࣋ͭʯͷɺ࣮ݧܭըͱϞσϧͷԾఆ͕ଥͳ ߹Ͱ͋ΓɺͦΕҎ֎1Λܭࢉ͚ͨͩ͠Ͱ͋Δɻ !26
ଟ͘ͷղੳʮతʯͰͳ͘ʮσʔλʯ͔Β࢝·Δ* ‣ ύλʔϯͭɻ ‣ ʢ"ʣత͕͋ͬͯɺσʔλͷूΊํ͔ΒσβΠϯ͢Δ ‣ ʢ#ʣख࣋ͪͷσʔλ͔Βɺతʹରͯ͠Ξϓϩʔν͢Δ ‣ ͪΖΜɺʢ"ʣͷ΄͏͕σʔλղੳͱͯ͠దͳΞϓϩʔνͰ͋Δ͕ɺ ࣮ࡍʢ#ʣͷΑ͏ʹͳͬͯ͠·͏ͷɺํͷͳ͍͜ͱʂ
‣ ӡಈྔ͕ଟ͍ਓ΄Ͳɺ࣬ප͕Լ͕Δ͔ʁͱ͍͏ٙʹ͑Δ߹ʹɺ 3$5·ͰΔͱ͍͏ͷͳ͔ͳ͔͍͠ɻ ‣ ࣮ࡍɺۙͳਓଌఆػثΛͯ͠ɺาߦྔͱੜମࢦඪͳͲΛൺֱ͢ Δ͔͠ͳ͍ɻ ‣ ͜ͷ߹ͷҙ ‣ ۙͳਓʹຊશࠃ͔ΒͷϥϯμϜαϯϓϧͰͳ͍ͷͰɺ·ͣ݁Ռ ҰൠԽͰ͖ͳ͍ɻ ‣ ͦͦଌఆػثΛਅ໘ʹ͏ͷɺ݈߁ҙࣝͷߴ͍ਓͳͷͰɺ σʔλऔಘͷόΠΞε͕ੜ͍ͯ͡Δɻ !27
ଟ͘ͷղੳʮతʯͰͳ͘ʮσʔλʯ͔Β࢝·Δ** ‣ αϯϓϧͷऔಘʹόΠΞε͕͋Δ͜ͱΛ౿·্͑ͨͰɺؔੑΛݟΔͨΊʹ ɺઢܗճؼϞσϧΛͯΊͯɺճؼΛݟΕ͍͍ʂ ‣ ͱ͍͚ͯ͠ͳͯ͘ɺʮϞσϧͷଥੑʯΛ͖ͪΜͱઆ໌Ͱ͖ΔΑ͏ʹͯ͠ ͍͔ͳ͍͚ͯ͘ͳ͍ɻ ‣ ͦͦ9ͱ:ͲΜͳؔͳΜ͚ͩͬʁ ‣
Ϟσϧͷଥੑʁมຊʹઢܗʹޮ͍ͯΔͷʁͳͲͳͲ !28 ༗ҙͩʂʂʂʂ ٩ ๑ÒТÓ๑ ۶
ଟ͘ͷղੳʮతʯͰͳ͘ʮσʔλʯ͔Β࢝·Δ*** ‣ ઌ΄ͲͷઢܗճؼϞσϧͷ݁Ռɺӈ ͷσʔλΛ༻͍ͨͷɻ ‣ ͔֬ʹɺԿ͔ͷ͕ؔ͋Γͦ͏͚ͩ Ͳɺ҆қʹઢܗʹ͍͍ͯ͠ͷʁ ‣ ͜͏͍͏ͱ͖ɺʮઢܗճؼϞσϧΛ ͯΊͨͱ͖ͷԾఆʯΛ͍ͬͯΔ
͔Ͳ͏͔Ͱɺஅ͕Ͱ͖Δɻ ‣ ԼͷਤʮͯΊWTࠩʯͷϓ ϩοτɻઢܗճؼϞσϧͷ߹ʮ ࢄ͕ҰఆʯͳͷͰɺΒͳ͍ ͣɻ ‣ ͔͠͠ɺ໌Β͔ʹ࣍ͷ͕ͬͯ ͍ͯɺ͜ΕͰઢܗճؼϞσϧͷԾఆ ຬͨ͞Ε͍ͯͳ͍ɻ !29
ଟ͘ͷղੳʮతʯͰͳ͘ʮσʔλʯ͔Β࢝·Δ*7 ‣ ࣍ͷ߲·ͰϞσϧʹؚΊΔͱɺ ࠩʹ͕Βͣɺ͓ΑͦࢄҰ ఆʹͳͬͨɻ ‣ ઢܗճؼϞσϧͷԾఆɺཱͯ͠ ͍Δͱߟ͑ͯྑͦ͞͏ɻ ‣ ࣍ʹɺ͜͜Ͱਪఆ͞Εͨʮճؼ
ʯ͕ɺʮ9͔Β:ͷӨڹʯͱߟ͑ ΒΕΔͱͯ͠ɺ͜ͷճؼͲͷ ఔΒͭ͘ͷ͔ʁ ‣ ͜ͷ࣭ʹ࣍ͷͭΛ࣋ͬͯ͑ Δɻ ‣ ̍ʣۙࢄͱ৴པ۠ؒͷهࡌ ‣ ̎ʣ#PPUTUSBQਪఆྔͷώετά ϥϜͱɺTVNNBSZΛهࡌ !30
ଟ͘ͷղੳʮతʯͰͳ͘ʮσʔλʯ͔Β࢝·Δ7 ‣ σʔλղੳͷऴΘΓʹɺ࣍ͷͭඞͣνΣοΫ͢Δɻ ‣ ʮ౷ܭతͳԾఆʯʹର͢Δໃ६࠷খݶʹ͑ΒΕ͍ͯΔ͔ʁ ‣ σʔλղੳͷ݁Ռͱɺݱ࣮తͳࢲͨͪͷײ֮ʹେ͖͗͢Δᴥᴪͳ͍͔ʁ ‣ Ϩϙʔτͷ࡞ ‣
·ͣɺσʔλͷऔಘͱɺࠓճͷղੳ݁Ռͷద༻ՄೳൣғΛ໌ه͢Δɻ ‣ ·ͨɺͲͷΑ͏ͳղੳΛݕ౼͠ɺ్தͰͲΜͳ݁ՌΛಘͯɺ࠷ऴ݁Ռʹ ࢸ͔ͬͨΛ໌ه͢Δɻ ‣ ݱ࣮తͳײ֮ͱͷᴥᴪ͕ͳ͍͔ɺᴥᴪ͕͋ΔͷͰ͋ΕͲΜͳݪҼ͔Λॻ ͘ɻ ‣ ྫ͑ɺʮʓʓͷΑ͏ͳม͕Γͯͳ͍ʯͳͲ !31 σʔλͷղੳΛߦͬͨϨϙʔτͰɺਪఆྔͷΒ͖ͭͷهࡌ͕ͳ͍߹ɺ ͦΕσʔλղੳͰͳ͍ɻ
·ͱΊ ‣ ͜͜·ͰɺσʔλղੳͷҰྫΛ͖͕ࣔͯͨ͠ɺಛʹʢ#ʣख࣋ͪͷσʔλ͔ Βɺతʹରͯ͠Ξϓϩʔν͢ΔσʔλղੳͰɺ ‣ ղੳରͱͳΔݱʹରͯ͠ɺਂ͍͕ࣝඞཁ ‣ ౷ܭֶʹର͢Δਂ͍ཧղͱɺͦΕͷԠ༻ೳྗ͕ඞཁɻ ‣ ͜ΕΒͭͰɺσʔλղੳͰ͋Δɻ·ͣɺࣗͨͪͷߦ͍ͬͯΔʮσʔλղ
ੳʯͷϑϩʔΛݟͯ͠΄͍͠ɻ ‣ 1ଞͷ౷ܭతͳج४Λɺओுͷࠜڌͱ͍ͯ͠ͳ͍͔ɻ ‣ ༻͍ͯ͠Δղੳπʔϧेͳཧղ͕͋Γɺਖ਼͍͠ӡ༻Λߦ͍ͬͯΔ͔ ʢ·ͨɺղੳޙࠩϓϩοτͳͲΛνΣοΫ͠ɺϨϙʔτܝࡌ͍ͯ͠Δ ͔ʁʣɻ ‣ औಘ͞ΕͨσʔλʹόΠΞεଘࡏ͍ͯ͠ͳ͍͔ʁ ‣ ݁ՌͷաͳҰൠԽɺߦΘΕ͍ͯͳ͍͔ʁ ‣ ղੳ݁Ռʹɺۙࢄ৴པ۠ؒ#PPUTUSBQਪఆྔͷΒ͖ͭͷϓϩοτͷ ͖ͭͪΜͱܝࡌ͍ͯ͠Δ͔ʁ !32 ν Σ ο Ϋ ߲
33 -FU`TNPWFUPBXPSMECFZPOElQ zUPHFUIFS தଜɹൟ ܚጯٛक़େֶେֶӃ UPNPTIJHFOBLBNVSB!HNBJMDPN ͋Γ͕ͱ͏͍͟͝·ͨ͠ʂ
ࢀߟจݙ ‣ "NSIFJO 7 (SFFOMBOE 4 BOE.D4IBOF # 4DJFOUJTUTSJTFVQ
BHBJOTUTUBUJTUJDBMTJHOJpDBODF/BUVSF ‣ 'JTIFS " 4UBUJTUJDBM5FTU /BUVSF ‣ )FME - BOE0UU . 0OQ7BMVFTBOE#BZFT'BDUPST"OOV3FW 4UBU"QQM ‣ )VOH + 0/FJMM 3 #BVFS 1 BOE,PIOF , 5IF#FIBWJPSPGUIF 17BMVF8IFOUIF"MUFSOBUJWF)ZQPUIFTJTJT5SVF#JPNFUSJDT ‣ 4FMMLF 5 #BZBSSJ 4 BOE#FSHFS + $BMJCSBUJPOPGQ7BMVFTGPS 5FTUJOH1SFDJTF/VMM)ZQPUIFTJT5IF"NFSJDBO4UBUJTUJDJBO r ‣ 8BTTFSTUFJO 3 BOE-B[BS / 5IF"4"`T4UBUFNFOUPOQ7BMVFT $POUFYU 1SPDFTT BOE1VSQPTF 5IF"NFSJDBO4UBUJTUJDJBO r ‣ 8BTTFSTUFJO 3 4DIJSN " BOE-B[BS / .PWJOHUPB8PSME #FZPOEQ 5IF"NFSJDBO4UBUJTUJDJBO ‣ ༄ᴲ Qྟচݚڀσʔλղੳ݁Ռใࠂʹ༗༻ͳ༏ΕͨϞϊαγͰ ͋Δܭྔੜֶ !34