A night shift by age product term is added, read and plotted in PHLT 8500's interaction week, with the scale question, multiplicative or additive, settled before any conclusion is drawn. Searches like "phlt 8500 week 8 assignment example", "phlt8500 week 8 sample" and "phlt 8500 week 8 example" land here.
What a finished PHLT 8500 Week 8 interaction analysis looks like
Three pages with one model table, one figure and a short stratified table. The logistic model adds a product of night shift and centered age to the week five specification, and the table reports it with its interval and a likelihood ratio test for the added term. A paragraph explains that the night shift coefficient now describes workers of average age only, because age is centered, and that the product term gives how the log odds ratio changes per decade. The figure plots predicted probability of hypertension against age separately for night and day workers, so the changing gap is visible. A stratified table gives odds ratios for the younger and older halves. The final section shows the interaction is modest on the odds scale but larger on the risk difference scale, because baseline risk rises with age.
How a PHLT 8500 Week 8 example is structured
The analysis begins with a hypothesis, not a fishing trip: prior evidence suggests that older workers may tolerate disrupted sleep less well, so the association might strengthen with age. That stated reason is what licenses testing a product term at all. The specification section explains centering and what each coefficient now describes. The test of the product term comes next, then the move most drafts skip: translating it into conditional effects at meaningful ages. The figure follows, since predicted risks show the pattern directly. The stratified table offers a check readers can follow without software. Last comes the scale discussion, which explains why an interaction can be absent on one scale and present on the other, and argues that the additive scale is the one a transit agency planning shift policy would care about.
A reason to look
The product term enters because prior evidence suggests older workers may be more affected by shift disruption. Stating that reason first separates a planned test from a search through every possible pairing.
Main effects, redefined
Once the product term is in, the night shift coefficient describes workers of average age only, because age is centered. The analysis says so, since reading it as an overall effect would misstate what it now measures.
Conditional effects at real ages
Odds ratios for night shift are computed at younger, middle and older ages. Three numbers at named ages communicate the interaction far better than a product coefficient on the log odds scale.
Risks, plotted
Predicted probability of hypertension is drawn against age for night and day workers. The gap between the lines, and how it changes, is the interaction made visible.
Two scales, two answers
On the odds scale the interaction is small; on the risk difference scale it is larger, because baseline risk climbs with age. The analysis explains why, and argues which scale suits a decision about shift assignment.
Where marks go in PHLT 8500 Week 8
Interpretation of the model once the product term is present carries most of this rubric's weight. An analysis that adds an interaction and then reads the night shift coefficient as the overall effect of shift work has misread its own specification, and that costs the central criterion. Graders look for a stated reason to test the interaction, and products added to every pair of variables draw criticism for inflating false findings. Conditional effects at meaningful values are expected; a product coefficient reported alone leaves the reader unable to use it. The figure earns credit when it shows the pattern clearly. The scale discussion separates the strongest analyses, since recognizing that interaction depends on the measure is an advanced point. Reporting format and the stratified table take the rest.
Get a PHLT 8500 Week 8 example written to your instructions
Send the Week 8 prompt and rubric, the model being extended, and the variable you expect to modify the relationship. An interaction analysis with conditional effects, a predicted-risk figure and the scale question addressed comes back in 24-48h, free when it is your first. The cohort, its product term and every plotted risk are illustrative.
PHLT 8500 Week 8 questions, answered
Should the main effects stay in a model with an interaction term?
Yes, in almost every case. Removing a main effect while keeping its product term forces an unusual constraint on the model and makes the remaining coefficients hard to interpret. The main effects stay, but their meaning changes: each now describes the effect of one variable when the other equals zero, which is why centering continuous terms helps.
What does it mean that interaction depends on the scale?
An effect can be constant in relative terms, such as the same odds ratio at every age, while the absolute difference in risk changes because baseline risk differs. Logistic models test for departures from multiplicativity; public health decisions often concern absolute risk. Reporting both, or explaining which the question needs, prevents a finding of no interaction on one scale from being read as no interaction at all.
How should a nonsignificant interaction be reported?
With its estimate and interval, plus a note on what the data can and cannot rule out. Tests for interaction usually have low power, so a nonsignificant result often means the study could not detect a moderate difference, not that none exists. The analysis states whether the interval includes differences that would matter for practice before dropping the term from later models.