GB701 Unit 10 Results Section Draft Example

Results as a Committee Reads Them: A Unit 10 Results Section Draft for the Quote Conversion Study

[Student Name]

Purdue University Global

GB701: Business Statistics

Unit 10 Results Section Draft

[Instructor Name]

[Date]

The distributor, its accounts and every figure are composite and illustrative; the reporting standards and sources cited are real.

Results

# Sample

Of the 1,240 active commercial accounts in the sampling frame, 234 were drawn by stratified random sampling with proportional allocation by order type, 66 special-order and 168 stock-order accounts. Eighteen were excluded before analysis, 13 where one portal login served several branches of the customer and 5 with incomplete quote histories. The analytic sample comprised 216 accounts, 61 special-order and 155 stock-order. A second coder independently coded 22 randomly selected accounts; agreement on portal use was 100 percent and on the count of heavily revised quotes 95 percent.

# Descriptive Statistics

Table 1 reports means, standard deviations and intercorrelations for the outcome and the five predictors. Mean quote-to-order time was 8.44 days (SD = 4.63), with a median of 7.50. Of the 216 accounts, 74 (34.3%, 95% Wilson CI [28.2, 40.8]) used the customer portal for most quote requests, and 61 (28.2%) were special-order accounts. Quote-to-order time correlated with special-order status, r = .41, and with the number of heavily revised quotes, r = .41, but not with portal use, r = .00. Portal use correlated with special-order status, r = .35.

The table is numbered in bold, Table 1, with its title in italic title case on the line below, *Means, Standard Deviations and Intercorrelations of Study Variables*, following the APA format for tables (American Psychological Association, 2020).

# Assumption Checks

Variance inflation factors ranged from 1.01 to 1.41. Residual spread increased modestly with fitted values; Koenker's test was not significant, chi-square(5) = 7.89, p = .162, but heteroscedasticity-consistent (HC3) standard errors were used for all inference as specified in advance. Residuals were right-skewed, W = .945, p < .001. Fifteen accounts exceeded the screening threshold for Cook's distance (maximum = 0.055). Sensitivity analyses addressing normality and influence are reported below.

# Primary Analysis

Quote-to-order time was regressed on portal use, special-order status, the number of heavily revised quotes, tenure and annual purchases. The model explained 25.4% of the variance, R-squared = .254, adjusted R-squared = .236, F(5, 210) = 14.29, p < .001.

Holding the other predictors constant, portal use was associated with shorter quote-to-order time, b = -1.63 days, 95% CI [-2.80, -0.46], HC3 SE = 0.60, p = .007. Portal use alone accounted for 2.5% of the variance beyond the other predictors, f-squared = .033. Special-order accounts took longer, b = 3.44 days, 95% CI [1.95, 4.93], p < .001, as did accounts with more heavily revised quotes, b = 0.93 days per quote, 95% CI [0.49, 1.37], p < .001. Tenure, b = 0.00, 95% CI [-0.06, 0.06], p = .99, and annual purchases, b = -0.10 days per $100,000, 95% CI [-0.27, 0.07], p = .27, were not associated with quote-to-order time.

The portal result appears once, with its interval and effect size, and every later mention points back to this sentence.

# Secondary Analysis

Without adjustment, mean quote-to-order time did not differ between portal users (M = 8.45, SD = 4.58) and other accounts (M = 8.43, SD = 4.67), Welch's t(150.7) = 0.02, p = .986, mean difference = 0.01 days, 95% CI [-1.29, 1.32], d = 0.00. The contrast between the adjusted and unadjusted results reflects the concentration of portal use among special-order accounts, which converted more slowly.

# Sensitivity Analyses

With the logarithm of quote-to-order time as the outcome, portal use was associated with 17.7% shorter conversion time, b = -0.195. Excluding the 15 influential accounts, the portal coefficient was -1.66 days. Both analyses agreed with the primary model in direction and approximate size.

Limitations Carried to the Discussion

Three limitations from earlier units are carried forward for the discussion section rather than reported here. The study uses one distributor's records from one year. A few accounts share a parent company, which may violate independence. And portal use was chosen by customers, so the association may reflect unmeasured differences between the customers who chose it and those who did not.

Notes on the Draft

The section reports what was found and stops. Interpretation, including why portal users might convert faster and what the distributor should do, belongs in the discussion. Three decisions shaped the draft. The primary result is stated once, with an interval, a p value and an effect size, following the American Statistical Association's guidance that p values not be reported alone (Wasserstein & Lazar, 2016). Assumption checks are reported briefly, since Unit 9 already established them and a committee reads results for findings, not for the diagnostics' history. And every claim is limited to association, because customers chose whether to use the portal. Gelman and Hill (2007) caution that adjusted estimates from observational data remain associations unless the design supports a causal reading, and nothing in this design does.

References

American Psychological Association. (2020). Publication manual of the American Psychological Association (7th ed.).

Gelman, A., & Hill, J. (2007). Data analysis using regression and multilevel/hierarchical models. Cambridge University Press.

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