Interaction effects occur when a treatment’s impact depends on another factor, such as device or channel. This quiz checks if you can spot and model those cases correctly.
An interaction effect is present when the combined effect of factors differs from the ______ of their individual effects.
page weight
sum (or simple addition)
cache state
median
In a two-way analysis, non-parallel lines on an interaction plot suggest ______.
a logging outage only
a treatment-by-segment interaction
perfect randomization
no variance
A regression model tests interaction by including a ______ term (e.g., Treatment × Device).
product (multiplicative)
encryption
hash
square root
When interactions are expected, analysts should ______ them in the analysis plan to avoid data dredging.
ignore
only inspect after rollout
randomly switch
pre-specify
If power is limited, a ______ model can stabilize noisy segment lifts.
static CSS
deterministic cache
raw pixel
hierarchical (partial pooling)
Simpson’s paradox is a reminder that overall lift can differ from segment lifts due to ______.
font rendering
image sprites
composition differences
gzip settings
To report interactions responsibly, include point estimates with ______ to show uncertainty.
folder names
RGB swatches
HTTP verbs
confidence or credible intervals
If treatment works on mobile but not desktop, this is best described as ______.
effect modification by device
pure confounding
UAT failure
instrumentation loss only
Adding many exploratory interactions raises false positives; a common control method is ______ adjustment.
image dithering
lossless ZIP
TTL tuning
false discovery rate (FDR)
Visualizing marginal means by segment helps interpret interactions because it shows ______.
how treatment effects differ across levels
CSS cascade depth
DOM paint time
server CPU load
Starter
New to interpreting interaction effects? Review the basics and try a few small, low-risk tests.
Solid
Good grasp of interpreting interaction effects. Tighten design choices and deepen your analysis habits.
Expert!
Excellent command of interpreting interaction effects. You are ready for scaled rollouts and mentoring others.
