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
Testing Hypotheses of No Meaningful Effect
Search
Carlisle Rainey
January 04, 2013
Research
1
400
Testing Hypotheses of No Meaningful Effect
Carlisle Rainey
January 04, 2013
Tweet
Share
More Decks by Carlisle Rainey
See All by Carlisle Rainey
Topic Sampling @ SPSA 2024
carlislerainey
0
210
Topic Sampling
carlislerainey
1
430
ggcarly:theme_carly_presents()
carlislerainey
0
79
Unnecessary
carlislerainey
0
560
When BLUE Is Not Best: Non-Normal Errors and the Linear Model
carlislerainey
0
210
Dealing with Separation in Logistic Regression Models
carlislerainey
0
140
Logit Models in Small Samples
carlislerainey
0
120
When BLUE is Not Best
carlislerainey
0
460
Dealing with Separation in Logistic Regression Models
carlislerainey
0
510
Other Decks in Research
See All in Research
2025/7/5 応用音響研究会招待講演@北海道大学
takuma_okamoto
1
150
2021年度-基盤研究B-研究計画調書
trycycle
PRO
0
170
Adaptive fusion of multi-modal remote sensing data for optimal sub-field crop yield prediction
satai
3
230
Vision and LanguageからのEmbodied AIとAI for Science
yushiku
PRO
1
430
データサイエンティストの採用に関するアンケート
datascientistsociety
PRO
0
1.1k
EarthSynth: Generating Informative Earth Observation with Diffusion Models
satai
3
150
経済学と機械学習:因果推論と密度比推定を中心に
masakat0
0
110
大規模な2値整数計画問題に対する 効率的な重み付き局所探索法
mickey_kubo
1
310
Agentic AIとMCPを利用したサービス作成入門
mickey_kubo
0
370
診断前の病歴テキストを対象としたLLMによるエンティティリンキング精度検証
hagino3000
1
120
電力システム最適化入門
mickey_kubo
1
780
集合間Bregmanダイバージェンスと置換不変NNによるその学習
wasyro
0
120
Featured
See All Featured
Performance Is Good for Brains [We Love Speed 2024]
tammyeverts
10
1k
Templates, Plugins, & Blocks: Oh My! Creating the theme that thinks of everything
marktimemedia
31
2.5k
GraphQLの誤解/rethinking-graphql
sonatard
71
11k
Reflections from 52 weeks, 52 projects
jeffersonlam
351
21k
XXLCSS - How to scale CSS and keep your sanity
sugarenia
248
1.3M
Art, The Web, and Tiny UX
lynnandtonic
301
21k
The Myth of the Modular Monolith - Day 2 Keynote - Rails World 2024
eileencodes
26
3k
StorybookのUI Testing Handbookを読んだ
zakiyama
30
6k
Fight the Zombie Pattern Library - RWD Summit 2016
marcelosomers
234
17k
Practical Orchestrator
shlominoach
190
11k
The Success of Rails: Ensuring Growth for the Next 100 Years
eileencodes
46
7.5k
RailsConf & Balkan Ruby 2019: The Past, Present, and Future of Rails at GitHub
eileencodes
139
34k
Transcript
Testing Hypotheses of No Meaningful Effect Carlisle Rainey
I'd like to convince you of three things
Important 1
Our arguments are not compelling 2
Our arguments can be compelling 3
Important Step 1:
Hypotheses of no meaningful effect are crucial to complete evaluation
of theories
Interaction
Social heterogeneity increases the number of parties, but only when
electoral institutions are sufficiently permissive. “ ” Clark and Golder (2006)
Adjudication
None
None
How often do these examples occur?
30%
Our arguments are not compelling Step 2:
No American president since FDR has won a second term
when the unemployment rate topped 7.2 percent... Obama must defy that trend to keep his job. “ ” New York Times June 1, 2011 “ ”
None
None
The nation's unemployment status by itself is not going to
affect Obama's. “ ” Seth Masket June 2, 2011
My critique
Rule out implausible relationships
None
None
None
None
None
None
Insignificance can't be used to argue for “no effect.”
But doesn't everyone already know that? “ ” A skeptic
Political scientists draw strong conclusions from insignificance.
Recessions have no effect on whether a democracy is consolidated.
“ ” Svolik (2008)
Our arguments can be compelling Step 3:
just like any other hypothesis
Argue against relationships inconsistent with the hypothesis
Define substantively meaningful Step 1:
Check the 90% CI Step 2:
Only negativity about the respondent’s preferred candidate should have a
significant demobilizing effect on voter turnout. “ ” Krupnikov (2011)
What's a meaningful effect? 1% 3% 5% 7% 9%
The demobilizing effect could be as large as 9%
Those are the three things
Important 1
Our arguments are not compelling 2
Our arguments can be compelling 3
But what should I do about this? “ ” You
Go read the details crain.co/nme 1
Think about your own work. 2
Keep in mind when reviewing others' work 3
None