| Course | GB701 Business Statistics |
|---|---|
| Unit | Unit 7 |
| Paper type | Hypothesis test writeup |
| Length | About 860 words, 6 pages double-spaced |
| Format | APA 7 student paper |
| School | Purdue University Global |
| Program | MBA |
| Updated | October 2026 |
Free sample paper for GB701 Unit 7
No Difference, and Why That Is Not the Answer: A Unit 7 Hypothesis Test Writeup on Portal Users
[Student Name]
Purdue University Global
GB701: Business Statistics
Unit 7 Hypothesis Test Writeup
[Instructor Name]
[Date]
The distributor, its accounts and every figure are composite and illustrative; the methods and sources cited are real.
Research Question and Hypotheses
Research question. Among the distributor's active commercial accounts, does mean quote-to-order time differ between accounts that request most quotes through the customer portal and accounts that do not?
Null hypothesis. The population mean quote-to-order time of portal users equals that of other accounts: mu-portal minus mu-other equals zero.
Alternative hypothesis. The population means differ: mu-portal minus mu-other is not zero.
The test is two-sided at alpha of .05. These hypotheses, the alpha level and the choice of test were recorded in the Unit 4 memo before the sample was drawn.
Descriptive Statistics
Portal users: 74 accounts, mean 8.45 days, standard deviation 4.58. Other accounts: 142, mean 8.43 days, standard deviation 4.67.
Why Welch and Not the Pooled Test
The pooled two-sample t test assumes the two populations have equal variances. Welch's test does not, and it adjusts the degrees of freedom to reflect the variances and sample sizes actually observed. Here the sample standard deviations are similar, but the groups are unequal in size, and the pooled test's error rate can be distorted when unequal variances coincide with unequal group sizes. Delacre et al. (2017) recommend Welch's test as the default for exactly that reason: it performs almost as well as the pooled test when variances are equal and much better when they are not. Because the choice was made before the data were seen, it cannot be accused of being selected to produce a result.
Results
The difference in means, portal users minus others, is 0.01 days. The 95 percent confidence interval runs from -1.29 to 1.32 days. Welch's t is 0.02 on 150.7 degrees of freedom, p = .986. The effect size, Cohen's d computed with the pooled standard deviation, is 0.00.
The null hypothesis is not rejected. More usefully, the interval rules out an unadjusted difference of two days in either direction, the smallest difference the vice president of sales said would matter.
A Rank-Based Check
Because quote-to-order time is right-skewed, the comparison was repeated with the Mann-Whitney test, which works on ranks and so needs no assumption of normal data. The result agrees: U = 5,305, p = .908. Ruxton (2006) notes that Welch's test is often a better choice than the Mann-Whitney test when the question concerns means, so the rank-based result is reported as a check rather than as the primary test.
The interval does more than fail to find a difference; it rules out the size of difference the business said would matter.
Reading a Null Result
A result with p of .986 tempts two opposite mistakes. One is to report that the portal makes no difference, treating a failure to reject the null hypothesis as proof that it is true. The other is to dismiss the result as uninformative because it is not significant. Both are wrong here. The interval shows what the data are consistent with: an unadjusted difference somewhere between about 1.3 days slower and 1.3 days faster. That is informative, because it excludes the two-day difference the business cares about. But it is informative only about the unadjusted comparison. Wasserstein and Lazar (2016) warn against concluding that no effect exists because a p value is large, and the next section explains why that warning applies with special force to this comparison.
What the Comparison Cannot Show
An unadjusted comparison cannot show whether portal use is associated with faster conversion among similar accounts. Unit 3 showed that special-order accounts convert much more slowly than stock accounts, and in this sample 61 percent of special-order accounts use the portal against 24 percent of stock accounts. Portal users are therefore disproportionately the slower kind of account. If portal use speeds conversion within each kind of account, the unadjusted comparison could still show no difference, because the faster conversion would be offset by the slower mix. This is a classic case of confounding, and the regression in Unit 8 is designed to address it. Until then, the result should be read as "portal users and others convert at similar average speeds," not as "the portal makes no difference."
What the Next Unit Must Do
The regression in Unit 8 must therefore compare portal users with other accounts of the same order type, while also accounting for revision counts, tenure and purchases. If the portal association appears only after adjustment, the unadjusted result here will have been misleading on its own, and the results section will need to report both and explain the difference.
Power, Revisited Honestly
The Unit 4 memo planned 76 portal users and 144 other accounts for 80 percent power to detect a two-day difference. After exclusions, the sample has 74 and 142, which gives power of about 79 percent for that difference. The shortfall is small and does not change the interpretation, but it is reported because power planned is not power achieved.
References
Delacre, M., Lakens, D., & Leys, C. (2017). Why psychologists should by default use Welch's t-test instead of Student's t-test. International Review of Social Psychology, 30(1), 92-101. https://doi.org/10.5334/irsp.82
Ruxton, G. D. (2006). The unequal variance t-test is an underused alternative to Student's t-test and the Mann-Whitney U test. Behavioral Ecology, 17(4), 688-690. https://doi.org/10.1093/beheco/ark016
Wasserstein, R. L., & Lazar, N. A. (2016). The ASA statement on p-values: Context, process, and purpose. The American Statistician, 70(2), 129-133. https://doi.org/10.1080/00031305.2016.1154108
GB701 Unit 7 instructions, in plain terms
The seventh GB701 assignment asks you to write up a hypothesis test the way a doctoral committee expects. State the research question and both hypotheses as claims about all accounts in the frame, in words and symbols, with the alpha level and direction, and note where they were fixed in advance. Report descriptive statistics for each group. Justify the test procedure before giving results. Give the difference, its interval, Welch's statistic and its degrees of freedom, the exact p and an effect size in standard deviation units. Add a robustness check suited to the data. Explain what the result shows and what it cannot show, especially for observational groups. Revisit power using the achieved sample. About four pages with APA 7 citations is common.
How the GB 701 Unit 7 example is put together
The writeup states the question and both hypotheses about all active accounts, two-sided at .05, citing the Unit 4 memo. Descriptive statistics follow for each group. A section defends Welch's test, citing Delacre, Lakens and Leys. Results report a 0.01-day difference with its interval, t on 150.7 degrees of freedom and d of 0.00, and the highlighted sentence says the interval rules out the business threshold. The Mann-Whitney test agrees, with Ruxton on its role. A section on reading null results cites Wasserstein and Lazar. Confounding by order type is explained with the 61 percent and 24 percent portal shares, and achieved power of about 79 percent closes the writeup.
GB701 Unit 7 rubric: what earns full points
Points in the GB701 hypothesis writeup depend on doing things in the right order. Hypotheses about the population, fixed in advance, earn credit because they show the test was not chosen after seeing data. A justified procedure, with the reason for Welch over the pooled test, demonstrates understanding. A result reported in full, from the mean difference and its interval to d, carries heavy weight. A robustness check suited to skewed data shows care. Interpreting a null result correctly, as neither proof of no effect nor uninformative, shows maturity. Naming confounding in observational groups and revisiting power honestly complete the rubric, along with APA 7 citations.
Common GB701 Unit 7 mistakes, and how to avoid them
Many GB701 writeups state hypotheses about the sample rather than the population. Others report only a p value, omitting the interval and effect size. A common error is treating a large p value as proof that groups are the same. Writers often choose the pooled t test by default without checking whether its assumption is reasonable. Observational comparisons are frequently interpreted as if groups were randomly formed. Power is rarely revisited after exclusions reduce the sample. Write hypotheses about the population, justify the test first, report every element of the result, add a robustness check, read the interval, name confounders and recompute power. A free custom hypothesis test writeup can be ordered through this page.
Get a GB701 Unit 7 example written to your instructions
Provide your data or output, the hypotheses the assignment names and the Unit 7 instructions with the rubric attached. Hypotheses first, procedure defended, result stated at the strength the evidence allows whether significant or not: that is the custom writeup you receive within 24-48h. A first sample is free. The paper above is an original model document written by our desk, not a submitted student paper and not an official Purdue University Global document.
More GB701 papers and MBA samples
- GB701 Unit 1: Discussion Board Post
- GB701 Unit 2: Descriptive Summary Exercise
- GB701 Unit 3: Probability Problem Set
- GB701 Unit 4: Sampling and Power Memo
- GB701 Unit 5: Confidence Interval Exercise
- GB701 Unit 6: Seminar Reflection
- GB701 Unit 8: Regression Model Report
- GB701 Unit 9: Assumption Diagnostics Memo
- GB701 Unit 10: Results Section Draft
- GB700 Unit 7: Scholarly Voice Exercise
- GB601 Unit 7: Implementation Roadmap
- GB600 Unit 7: Stakeholder Response Plan
- GB590 Unit 7: Sustainability Trade-Off Paper
GB701 Unit 7 questions, answered
How should a nonsignificant result be reported in GB701?
Exactly as a significant one would be: the estimate, its confidence interval, the test statistic with degrees of freedom, the p-value and an effect size. Then say what the interval rules out and what it leaves open. Phrases such as no difference exists overstate the evidence, and instructors typically ask for the interval-based reading instead.
Is a one-sided test acceptable when the claim is directional?
Sometimes, if the direction was committed to in writing before any data were examined and an effect in the unexpected direction would count as no effect at all. Many doctoral instructors prefer two-sided tests by default, because a surprising effect in the opposite direction is usually worth knowing about. Whichever is chosen, the justification belongs in the writeup before the results.
Does a hypothesis test writeup need a limitations section?
A short one, usually. For observational data the main limitation is that groups were not randomly formed, so differences may reflect other characteristics. Naming the most important alternative explanation, and saying how a later analysis will address it, reads far better than a generic list of limitations appended at the end.
Where can I find a free GB701 Unit 7 sample paper?
This page carries a complete GB701 Unit 7 hypothesis test writeup comparing portal users and other accounts with Welch's test, a confidence interval, a rank-based check and a discussion of confounding.
Does a p value of .986 mean the groups are the same?
No. It means the data are very consistent with no difference in the unadjusted means. The interval shows which differences remain plausible, and confounding means adjusted results may differ.