Speaker Deck
Sign in
Sign up
for free
experiment sandbox distributional design thinking
cjlortie
September 16, 2019
Education
3
58
experiment sandbox distributional design thinking
Variation, replication, and the general categories of experimental design are proposed.
cjlortie
September 16, 2019
Tweet
Share
More Decks by cjlortie
See All by cjlortie
cjlortie
0
44
cjlortie
3
200
cjlortie
2
30
cjlortie
0
53
cjlortie
0
33
cjlortie
0
69
cjlortie
0
67
cjlortie
0
62
cjlortie
0
44
Other Decks in Education
See All in Education
nyshumanrights
0
190
shu223
2
110
masakiaota
0
330
sfaasms
0
260
event2020
0
110
hiranofumio
1
110
iqiglobalslides
0
140
ivy_waffle
0
900
yoshizaki_kkgk
0
480
mapcmetroboston
0
1.2k
yasushihara
0
180
yasushihara
3
390
Featured
See All Featured
eitanlees
105
9.4k
keathley
12
420
malarkey
189
8.2k
chriscoyier
681
180k
moore
123
21k
ammeep
652
54k
shpigford
163
18k
stephaniewalter
256
10k
3n
159
22k
keavy
101
14k
cassininazir
342
20k
cromwellryan
99
5.7k
Transcript
@cjlortie experiment sandbox resource: experimental design 4 the life sciences
4e
variation, replication, and designations
goal is to minimize variation (or at least comprehend to
some extent)
response = dependent variable
factor = independent variable
random variation is inherent variation in a system that cannot
be explained by the independent variables or factors
between versus within variation patterns
one solution is replication
replication is repeated sampling typically on different subjects
replication can be wide or narrow (deﬁned as extent that
the goal of replication is to sample inherent variation)
what is a p-value?
The p-value is the level of marginal signiﬁcance within a
statistical hypothesis test representing the probability of the occurrence of a given event. The p-value is used as an alternative to rejection points to provide the smallest level of signiﬁcance at which the null hypothesis would be rejected.
distributional (versus point) design thinking
None
None
None
distributional design thinking seeks to use replication to facilitate detection
of ‘true’ differences between groups of subjects/samples
design options a. simple random b. random stratiﬁed c. cluster
d. convenience
random versus haphazard & selection bias with external validity are
key criteria to consider for better design thinking
never neglect independence i.e. a measure on one subject should
not predict/inﬂuence the measurement of another individual