Two failed assumptions, skewed minutes and one comparison center's early shift change, are recorded alongside four that held in this composite assumption check summary from GB892 Unit 4. Searches like "gb 892 unit 4 assignment example", "gb892 unit 4 sample" and "gb892 unit 4 example" land here.
What a finished GB892 Unit 4 assumption check summary looks like
Four pages built around one table, a row per assumption and five columns: assumption, check, result, verdict and consequence for the analysis. Parallel pre-period trends come first. Pooled across the nine months before adoption, the difference in slopes between groups is 0.003 log points a month, interval -0.005 to 0.011, yet DC-09's own series drops 0.112 log points against the other comparison centers from month five. Stable composition follows: floor-loaded share moved 0.1 points and night share -0.2 points. Next is the outcome's shape, skewness 1.77 in minutes and 0.26 after logging. Unload time serves as a negative control, since appointments should not change how long a trailer takes to empty, and it moved 0.5 percent. Twelve clusters, six treated, close the table. A figure plots the monthly gap between groups.
How a GB892 Unit 4 example is structured
Each row answers what would go wrong if the assumption were false, and the consequence column says what the analysis does about it. Verdicts take one word, so a reader scanning the table cannot miss a failure. The skew failure changes the planned outcome from minutes to log minutes, a change carried into the Unit 9 log. DC-09 is handled differently: a pooled test that passes is not allowed to hide a single site that plainly does not, so the summary commits to a sensitivity model without that center rather than removing it from the primary one. The cluster row explains why twelve centers make conventional clustered errors too optimistic and names the wild bootstrap to be used instead. One assumption, that carriers did not divert freight between groups, is marked untested, with comparison volume up 1.0 percent offered only as weak evidence.
One word per verdict
Held, failed or untested sits in its own column. Nuance lives in the result and consequence columns, while the verdict stays blunt, so a committee member reading only the table still sees both failures.
The center that changed early
DC-09 added a second receiving shift in month five, and its median wait for a door fell from 39 to 33 minutes. The pooled slope test does not detect it; the site's own line does, and the summary trusts the line.
Minutes are not symmetric
A handful of four-hour visits pull the mean well above the median, skewness 1.77. Logging brings it to 0.26 and turns effects into percentages, which suits a comparison between centers of very different size.
Unload time, a control that should not move
Appointments govern arrival, not the speed of emptying a floor-loaded trailer. A shift in unload time would signal that something besides the portal changed at the six centers; the estimate, 0.5 percent, sits close to zero.
Twelve clusters, six treated
Clustered standard errors assume many clusters. With twelve, conventional intervals run too narrow, so the summary names a wild cluster bootstrap with six-point weights and 4,999 replications as the inference method.
Diversion, untested
Carriers shifting freight away from centers that required appointments would raise comparison waits and widen the gap. No field records carrier choice, so the assumption stays untested and is listed that way.
Where marks go in GB892 Unit 4
Tests listed without consequences are the commonest weakness in a GB892 assumption summary, since a reader learns that a check ran and nothing about what it changed. Parallel trends asserted from a single pooled test, with no look at individual centers, miss exactly the kind of site DC-09 represents. Declaring a failed assumption harmless without a sensitivity analysis asks the reader to trust the analyst. The opposite error costs marks too: abandoning the design because one assumption failed, when a narrower claim would survive. Summaries leaving untestable assumptions off the table imply they were checked. Conventional clustered errors used with twelve clusters and no comment suggest unfamiliarity with the few-cluster problem. Instructors credit a negative control chosen for a stated reason, and discount one that appears as decoration.
Get a GB892 Unit 4 example written to your instructions
Share what your analysis plan assumed, the checks you have run or intend to run, and any results that worried you, along with the Unit 4 prompt and rubric. A summary comes back table-first, a verdict and consequence per assumption, with failures named rather than buried. Your first custom sample carries no cost; 24-48h.
GB892 Unit 4 questions, answered
What should happen when a key assumption fails?
State the failure, then change what the analysis claims or how it estimates, in proportion to the problem. A skewed outcome may call for a transformation; a comparison unit on a different trend may call for a sensitivity model without it. Abandoning the design is rarely required. What committees resist is a failure noticed and then quietly left out of the write-up.
Is a pooled pre-trend test enough to support parallel trends?
Usually not on its own. A pooled test averages across units and can pass while one unit diverges sharply, which is what DC-09 does in this sample. Plotting each unit's series against the comparison average, and naming any unit that departs, gives a reader far more than a single p value. Your section may specify which evidence it expects.
Why use a negative control outcome?
Because it tests the design rather than the hypothesis. An outcome the intervention should not affect, unload duration here, ought to show no change. If it moves, something else changed at the treated sites at the same time, and the main estimate would absorb that change. A near-zero result does not prove the design sound, but it removes one competing explanation.