Testing a vendor's claim that its scorecard lowers twelve-month defaults, this GF582 Unit 7 hypothesis test report runs a two-proportion z test and sets five assumptions beside it. Searches like "gf 582 unit 7 assignment example", "gf582 unit 7 sample" and "gf582 unit 7 example" land here.
What a finished GF582 Unit 7 hypothesis test report looks like
Four pages in report form: a claim box, a test box, an assumptions table and a two-paragraph conclusion. The claim box states the hypotheses in symbols and words, the null that the two default rates are equal and the alternative that the new rate is lower, fixed as one-sided before the data were opened because the claim itself is directional. The test box gives 5.29 percent against 4.19, a difference of 1.11 points, a pooled rate of 4.77, a standard error of 0.633 and z of 1.75, for a one-sided p-value of 0.040. Beside it sits the 95 percent interval for the difference, minus 0.13 to 2.34 points. The assumptions table lists five conditions, each marked met, checked or open. The conclusion rejects the null at 0.05 and states what that does not establish.
How a GF582 Unit 7 example is structured
Assumptions sit beside the result rather than after it, because a rejection is only as good as the conditions under it. Independence between cohorts is met, since no borrower appears in both. The success-failure condition is checked with counts of 127, 2,273, 90 and 2,060, all well above ten. Equal follow-up is met, both cohorts observed for twelve full months. Comparable conditions are open: the cohorts were booked in consecutive years, and a softer or stronger economy in the second could account for part of the gap. Representativeness is open too, since the new scorecard also changed which applicants were approved. The conclusion then holds two findings together. The one-sided test rejects at 0.05, yet the two-sided interval includes zero, and the report explains why both are true and why the test's power was only about 54 percent.
The claim, written as hypotheses
Null: the twelve-month default rate under the new scorecard equals the old. Alternative: it is lower. The one-sided direction is justified from the vendor's wording, not from the data.
The statistic, step by step
Pooled proportion 217 over 4,550, or 4.77 percent; standard error 0.633 points; z equal to 1.106 divided by 0.633, or 1.75. The one-sided p-value is 0.040.
Five conditions, three statuses
Independence met, counts checked, follow-up met, economic conditions open, applicant mix open. The table gives each a one-line reason so that a reader can dispute a status directly.
An interval that crosses zero
The 95 percent interval for the difference, minus 0.13 to 2.34 points, is two-sided and so looser than the one-sided test. The report adds the one-sided 95 percent lower bound, 0.07 points, to reconcile them.
Limits of the rejection
It does not show the scorecard caused the drop, nor that 1.11 points is the size of the effect. It shows the data would be unusual under equal rates, with two conditions still open.
Where marks go in GF582 Unit 7
Hypothesis test reports in GF582 lose most when the conclusion is stated in the test's vocabulary and never in the business's: reject the null, with no sentence on defaults, dollars or the vendor's claim. Many sections deduct for hypotheses written after seeing the data, especially a one-sided test chosen because the two-sided result missed. Assumptions listed generically, independence and normality, without checking them against these cohorts, earn little of the credit the table is meant to collect. Reporting only the p-value, with no difference or interval, leaves a reader unable to judge size. The subtler loss is causal language: a lower default rate among new-model loans does not by itself show the model caused it, and the confound is expected to be named.
Get a GF582 Unit 7 example written to your instructions
A claim to test, the counts behind it, and the instructions for Unit 7 with their rubric: send those and the custom report states the hypotheses, runs the test the data call for and sets each assumption beside the result. Expect the report within 24-48h, with no fee attached to a first request.
GF582 Unit 7 questions, answered
When is a one-sided test justified?
When the question is directional before the data are seen, and a result in the other direction would lead to the same decision as no difference. A vendor claiming its model lowers defaults is a fair example. Choosing one-sided after a two-sided test narrowly misses is not, and most rubrics penalize it. State the direction and the reason in the hypotheses section.
How can a test reject the null while the interval includes zero?
Because they allocate error differently. A one-sided test at 0.05 puts all of its error in one tail; a two-sided 95 percent interval splits it between two. The matching interval for a one-sided test is a one-sided bound. Reporting both, and explaining the difference in a sentence, usually earns more credit than hiding either one from the reader.
Do I need to check normality for a test of proportions?
Not of the data themselves, since each outcome is a yes or no. What needs checking is whether the counts are large enough for the normal approximation to the sampling distribution, commonly at least ten successes and ten failures in each group. With small counts, an exact test is the safer choice and worth naming.