1012-00004 · Stonefield Logistics Group · Pleadings & Court Filings

declaration-of-dr-priya-ramanathan.docx DOCUMENT Finance & Banking

30 / 35 ← → ✕

Contract analysis Extract to table

UNITED STATES DISTRICT COURT NORTHERN DISTRICT OF TEXAS DALLAS DIVISION

MARCUS DELANO, individually and on behalf of all others similarly situated,

Plaintiff,

STONEFIELD LOGISTICS GROUP INC.,

Defendant.

Case No. 3:25-cv-01847-RBC

Hon. Rebecca Chen, U.S. District Judge

DECLARATION OF DR. PRIYA RAMANATHAN IN SUPPORT OF DEFENDANT STONEFIELD LOGISTICS GROUP INC.'S RESPONSE IN OPPOSITION TO PLAINTIFF'S MOTION FOR CONDITIONAL CERTIFICATION

I, Dr. Priya Ramanathan, declare under penalty of perjury pursuant to 28 U.S.C. § 1746 as follows:

This declaration is submitted in support of Defendant Stonefield Logistics Group Inc.'s ("SLG" or "Defendant") Response in Opposition to the Motion for Conditional Certification filed by Plaintiff Marcus Delano ("Plaintiff" or "Delano") on July 14, 2025. The opinions expressed herein are based on my review and analysis of the data, documents, and materials described below, and are offered to a reasonable degree of economic and statistical certainty.

III. QUALIFICATIONS AND BACKGROUND

A. Professional Position

1. I am the Lead Consultant and Senior Vice President at Corbin Risk Advisors LLC ("CRA"), a litigation consulting firm specializing in expert economic and statistical analysis. CRA's principal offices are located at 1901 N. Moore Street, Suite 1020, Arlington, Virginia 22209. CRA provides analytical and consulting services to law firms, corporations, and governmental entities across a range of complex litigation and regulatory matters. My practice at CRA focuses primarily on labor economics, employment classification disputes, and statistical analysis in connection with wage and hour litigation, including collective actions under the Fair Labor Standards Act ("FLSA") and class actions under analogous state wage and hour statutes.

B. Educational Background

2. I received my Doctor of Philosophy (Ph.D.) in Economics, with a specialization in labor economics, from the Rutledge School of Social Sciences at Hargrove University, Charlottesville, Virginia, in 2007. My doctoral dissertation, titled "Heterogeneous Labor Supply Responses to Platform-Based Compensation Structures," was supervised by Professor Alan M. Whitford and was awarded the Hargrove Department of Economics Dissertation Prize for that academic year.

3. I received my Master of Arts (M.A.) in Applied Statistics from Caldwell Polytechnic Institute, Atlanta, Georgia, in 2003. My master's thesis focused on the application of hierarchical linear modeling techniques to multi-site employment data.

4. I received my Bachelor of Arts (B.A.), summa cum laude, with a double major in Mathematics and Economics from Wentworth College, Boston, Massachusetts, in 2001. I was elected to Phi Beta Kappa during my junior year.

C. Professional Experience

5. I have over eighteen (18) years of professional experience in the fields of labor economics, statistical analysis, and employment litigation consulting. Over the course of my career, I have provided expert economic and statistical analysis in more than forty-five (45) FLSA collective action and state-law class action matters involving claims of misclassification, off-the-clock work, unpaid overtime, and related wage and hour allegations. These engagements have spanned numerous industries, including logistics and delivery services, healthcare, financial services, technology, retail, and hospitality.

6. I have been deposed or provided trial testimony in approximately twenty (20) cases in federal and state courts across the United States, including the United States District Courts for the Southern District of New York, the Northern District of Illinois, the Central District of California, the Eastern District of Pennsylvania, the Middle District of Florida, and the Western District of Texas, as well as state courts in New York, California, Illinois, and Virginia. A complete listing of my prior testimony over the past four years is included in my curriculum vitae, attached hereto as Exhibit F.

7. I have authored or co-authored more than a dozen articles published in peer-reviewed academic and professional journals on topics including worker classification, wage and hour economics, labor market analysis, and the economics of platform-based work arrangements. My published work has appeared in the Journal of Labor Economics, the Industrial and Labor Relations Review, the Journal of Applied Econometrics, and the Journal of Human Resources, among other publications. A complete listing of my publications is included in my curriculum vitae, attached hereto as Exhibit F.

8. Prior to joining CRA in 2014, I served as an Associate Professor of Economics at Dunmore University, Richmond, Virginia, where I taught graduate and undergraduate courses in labor economics, econometrics, and applied statistics from 2007 through 2014. During my academic tenure, I supervised multiple doctoral dissertations and master's theses related to labor economics and employment policy.

D. Engagement in This Matter

9. I was retained by the law firm of Calderwood & Harkness LLP on behalf of Defendant Stonefield Logistics Group Inc. pursuant to an Expert Retention Agreement dated May 8, 2025. My engagement in this matter is governed by the terms of that agreement, which provides for compensation at a rate of $550 per hour for my time and $375 per hour for the time of associate analysts working under my direction. My compensation, and the compensation of any analysts assisting me, is in no way contingent upon the outcome of this litigation, the content of any opinions I express, or the nature of any testimony I may provide.

E. Scope of Assignment

10. I was asked to analyze whether the approximately 1,187 individuals who performed delivery driving services for SLG under Independent Contractor Operating Agreements ("ICOAs") during the period from April 30, 2022 through the present (the "Relevant Period") are "similarly situated" for purposes of FLSA conditional certification. More specifically, I was asked to analyze the following areas: (1) variance in weekly hours worked across and within SLG's fourteen (14) distribution hubs; (2) variance in gross weekly pay across and within hubs; (3) delivery acceptance rate patterns relative to the 85% minimum acceptance rate specified in the ICOAs; (4) route deviation frequency as a measure of operational autonomy; and (5) multi-platform engagement and outside economic activity as measures of entrepreneurial opportunity. My analysis spans all fourteen (14) SLG distribution hubs located across the states of Texas, Oklahoma, Louisiana, and Arkansas.

F. Materials Reviewed

11. In forming my opinions, I reviewed and considered the following categories of materials:

(a) RouteSync dispatch records for all 1,187 drivers for the period April 30, 2022 through June 30, 2025, including delivery offer data, acceptance and rejection records, delivery completion timestamps, and assignment histories;

(b) GPS telematics data from SLG-provided tracking devices installed in driver vehicles across all fourteen (14) distribution hubs, encompassing location data, route tracking, and stop-time calculations;

(c) SLG per-delivery compensation records (pay records) for all 1,187 drivers during the Relevant Period, including per-delivery rates, delivery type classifications, and weekly payment summaries;

(d) Executed Independent Contractor Operating Agreements in all three versions: ICOA v1.0 (original 2018 template), ICOA v2.0 (revised August 2020), and ICOA v3.0 (revised January 2023);

(e) SLG driver onboarding records, performance review records, and driver availability logs;

(f) SLG hub-level operational policies and procedures manuals for all fourteen (14) distribution hubs;

(g) Plaintiff's Complaint, Dkt. No. 1, and Plaintiff's Motion for Conditional Certification, Dkt. No. 34, together with all supporting declarations and exhibits;

(h) The declarations of Trevor Sandoval, Karen Lund, and Dennis Faulkner, submitted in support of Plaintiff's Motion;

(i) Publicly available data concerning competing delivery and logistics platform operations in SLG's service areas, including rate schedules and independent contractor agreements for comparable platforms;

(j) Other documents and data sources as listed in Exhibit A, attached hereto and incorporated by reference.

G. Methodology Overview

12. My analyses employ standard and widely accepted statistical techniques, including descriptive statistics (mean, median, standard deviation, range, interquartile range, coefficient of variation), one-way analysis of variance ("ANOVA"), coefficient of variation analysis, ordinary least squares ("OLS") regression analysis, and chi-square tests of independence. All statistical analyses were performed using Penmark Systems (Ashgrove Health IT) and R version 4.3.1 (R Core Team). All analyses were performed, and all opinions expressed herein are held, to a reasonable degree of economic and statistical certainty.

IV. SUMMARY OF OPINIONS

13. Opinion 1 (Weekly Hours Worked): The 1,187 drivers in the proposed collective exhibit statistically significant variation in weekly hours worked across and within SLG's fourteen (14) distribution hubs. The mean weekly hours worked across the proposed collective is approximately 47.3 hours, with a standard deviation of approximately 12.8 hours and a coefficient of variation of approximately 27.1%. The named Plaintiff's average weekly hours of approximately 58 hours places him in the top decile of the distribution. Approximately 38.2% of all driver-week observations reflect fewer than 40 hours worked, meaning a substantial proportion of the proposed collective would have no overtime claim in any given week. These findings undermine the premise that the proposed collective members share a common experience with respect to hours worked.

14. Opinion 2 (Gross Weekly Pay): Gross weekly pay varies substantially across drivers and hubs, with a mean of approximately $1,078, a standard deviation of approximately $412, and a coefficient of variation of approximately 38.2%. This variation reflects individualized economic circumstances — including each driver's delivery mix, volume, and effort — rather than a uniform pay practice susceptible to collective resolution.

15. Opinion 3 (Delivery Acceptance Rates): Delivery acceptance rates vary widely across the proposed collective, with a standard deviation of approximately 6.8 percentage points around a mean of approximately 91.3%. Approximately 14.6% of driver-months fell below the 85% contractual threshold during the Relevant Period, and many of those drivers continued operating without formal consequence. This variation demonstrates that drivers exercised meaningfully different levels of discretion in accepting or declining delivery assignments, belying the notion of uniform employer control over scheduling and workload.

16. Opinion 4 (Route Deviation Frequency): Route deviation data demonstrates that drivers exercised significant and variable degrees of autonomy in how they performed deliveries. The mean route deviation frequency was approximately 3.7 per driver per week, with a coefficient of variation of approximately 78.4% — an extremely high level of dispersion. The named Plaintiff's average deviation frequency of approximately 2.1 per week places him below the collective mean, rendering his experience of relatively high route adherence unrepresentative.

17. Opinion 5 (Multi-Platform Engagement): A material subset of the proposed collective — approximately 247 drivers (20.8%) — showed evidence of engagement with competing delivery platforms or other outside economic activity during the Relevant Period. Multi-platform engagement rates varied dramatically by hub, ranging from approximately 7% at HUB-13 (Lubbock) to approximately 31% at HUB-05 (Austin). This variation in entrepreneurial opportunity is inconsistent with a unitary employment relationship and cannot be assessed on a collective basis. Additionally, approximately 89 drivers (7.5%) maintained separate business registrations during the Relevant Period. The overall conclusion is that the proposed collective is not a homogeneous group of similarly situated individuals; material individualized differences in work patterns, autonomy, and economic circumstances predominate over any common features.

V. ANALYSIS OF WEEKLY HOURS WORKED

A. Data Source and Methodology

18. To calculate weekly hours worked for each driver, I relied upon two complementary data sources: (a) GPS telematics data captured by SLG-provided tracking devices installed in each driver's vehicle, and (b) RouteSync login and logout timestamps recorded each time a driver commenced or concluded a delivery session through the RouteSync dispatch platform. By cross-referencing these two independent data streams, I was able to construct a reliable and comprehensive measure of weekly hours worked for each driver at each of SLG's fourteen (14) distribution hubs.

19. The data analyzed spans the full Relevant Period — from April 30, 2022 through June 30, 2025 — encompassing approximately 161 calendar weeks. During this period, the 1,187 unique drivers in the proposed collective were active for varying durations, reflecting the natural turnover and intermittent engagement characteristic of an independent contractor delivery network. The total number of driver-week observations analyzed was approximately 97,634. This figure represents the sum of individual active weeks across all 1,187 drivers, with each driver contributing data only for those weeks in which he or she accepted at least one delivery through RouteSync.

20. For purposes of my analysis, I defined "hours worked" in a given week as the total time elapsed from the first delivery dispatch acceptance to the last delivery completion during that week, net of breaks exceeding thirty (30) minutes, as identified through gaps in GPS movement data. This definition is consistent with standard approaches to measuring compensable time under Department of Labor guidance and is deliberately inclusive — it credits drivers with time during which they were actively engaged in delivery operations, including time spent driving between delivery locations, loading packages at hub facilities, and awaiting dispatch while logged in to the RouteSync platform. This definition, if anything, overstates compensable time for many drivers, as it does not deduct time that drivers may have spent on personal activities while nominally logged in to the platform.

B. Aggregate Statistics

21. Across all 97,634 driver-week observations, the mean weekly hours worked was approximately 47.3 hours. The median weekly hours worked was approximately 44.1 hours — meaningfully lower than the mean, indicating a right-skewed distribution in which a subset of high-hour drivers pulls the mean upward. The standard deviation was approximately 12.8 hours, and the range extended from approximately 15 hours per week (at the low end, reflecting drivers who operated on a part-time or intermittent basis) to approximately 68 hours per week (at the high end, reflecting the most active drivers). The coefficient of variation — the ratio of the standard deviation to the mean — was approximately 27.1%, indicating a high degree of dispersion relative to the central tendency. By conventional standards in labor economics, a coefficient of variation above 20% is considered indicative of substantial heterogeneity within a population.

22. I note that Plaintiff alleges in his declaration that he "typically" worked approximately 58 hours per week. My analysis of SLG's data confirms that Plaintiff Delano's average weekly hours during the Relevant Period was approximately 58.2 hours. However, this figure is far above the collective mean of 47.3 hours and above the collective median of 44.1 hours. In fact, Plaintiff Delano's average weekly hours places him at approximately the 91st percentile of the distribution — meaning that approximately 91% of all driver-week observations reflected fewer weekly hours than Plaintiff's average. His experience is not representative of the proposed collective.

C. Hub-Level Variation

23. I performed a hub-by-hub analysis of mean weekly hours worked across all fourteen (14) SLG distribution hubs. The results, summarized in Exhibit B and detailed below, reveal substantial variation in mean hours across hubs:

24. As the table above illustrates, mean weekly hours ranged from approximately 35.4 hours at HUB-13 (Lubbock) to approximately 52.4 hours at HUB-01 (Fort Worth) — a spread of 17.0 hours between the highest and lowest hub means. The three largest Texas hubs (Fort Worth, Dallas, and Houston) all exhibited mean weekly hours above 49 hours, while the smaller and more geographically remote hubs (Lubbock, Shreveport, Little Rock) showed mean weekly hours below 40 hours. This pattern is consistent with the greater delivery volume and demand density at the larger metropolitan hubs, which afforded drivers at those locations more opportunity — and, in some cases, more incentive — to work longer hours.

25. I performed a one-way analysis of variance ("ANOVA") to test whether the differences in mean weekly hours across hubs were statistically significant. The ANOVA yielded an F-statistic of 42.87 with 13 and 97,620 degrees of freedom, and a p-value of less than 0.001. This result is highly statistically significant and indicates that between-hub differences in mean weekly hours are not attributable to chance. The between-hub variance accounted for approximately 18.3% of total variance in weekly hours across the proposed collective. The detailed ANOVA results are presented in Exhibit C.

26. Although between-hub variance is substantial, within-hub variance is also significant, as discussed in the following paragraphs. This dual finding — of both between-hub and within-hub heterogeneity — indicates that the proposed collective is heterogeneous at every level of aggregation.

D. Within-Hub Variation

27. Even within individual hubs, the standard deviations in weekly hours worked are large, ranging from approximately 9.0 hours at HUB-11 (Shreveport) to approximately 13.1 hours at HUB-01 (Fort Worth). These within-hub standard deviations are comparable in magnitude to the overall standard deviation of 12.8 hours, confirming that substantial variation in weekly hours exists among drivers operating out of the same facility, under the same hub manager, and in the same geographic market.

28. To illustrate this point concretely, I examined the distribution of weekly hours worked at HUB-01 (Fort Worth, Texas) — the hub where the named Plaintiff, Marcus Delano, performed his delivery services. Among the 142 drivers at HUB-01, the mean weekly hours worked was approximately 52.4 hours, the median was approximately 50.8 hours, the standard deviation was approximately 13.1 hours, and the range extended from approximately 22 hours per week to approximately 68 hours per week. This 46-hour spread within a single hub underscores the wide variability in driver work patterns, even when external factors such as geographic market, hub management, and delivery demand are held constant.

29. Across the entire proposed collective, approximately 38.2% of all driver-week observations reflected fewer than 40 hours worked during the applicable week. This finding is of considerable significance for the conditional certification inquiry. Under Plaintiff's theory, the proposed collective members are entitled to overtime compensation because they were misclassified as independent contractors and, as putative employees, should have been paid overtime for hours worked in excess of 40 per week. However, if a driver worked fewer than 40 hours in a given week, that driver would have no overtime claim for that week regardless of the merits of any classification dispute. The fact that more than one-third of all driver-week observations fall below the 40-hour threshold means that a substantial portion of the proposed collective would have no overtime exposure in many or most weeks, and that individualized inquiry into each driver's weekly hours would be required to determine liability and damages.

30. Named Plaintiff Marcus Delano's average of approximately 58 hours per week places him at approximately the 91st percentile of the distribution. His experience of routinely working well in excess of 40 hours per week is not shared by the majority of the proposed collective. To the extent Plaintiff holds himself out as representative of the collective, the data do not support that characterization.

E. Temporal Variation

31. Hours worked by individual drivers also fluctuated significantly over time. I analyzed the within-driver temporal variation by calculating the coefficient of variation of weekly hours for each individual driver across his or her active weeks during the Relevant Period. The within-driver coefficients of variation ranged from approximately 15% to approximately 45%, with a median of approximately 28%. This indicates that even a single driver's weekly hours were not stable from week to week; rather, drivers increased and decreased their activity levels across quarters and seasons based on personal circumstances, market conditions, and other individualized factors.

32. Moreover, seasonal patterns in hours worked differed meaningfully by hub. The Houston and Dallas hubs (HUB-02 and HUB-03) exhibited pronounced holiday-season spikes in weekly hours, with fourth-quarter average hours approximately 15–20% higher than first-quarter average hours. By contrast, the smaller and more rural hubs — such as Lubbock (HUB-13), Little Rock (HUB-10), and Shreveport (HUB-11) — showed comparatively flat seasonal patterns, with fourth-quarter hours only marginally (approximately 3–5%) higher than first-quarter hours. These differing seasonal dynamics further demonstrate that the work experience of drivers at different hubs was not uniform and that aggregate or "one-size-fits-all" treatment of the proposed collective would obscure material differences among its members.

F. Implications for "Similarly Situated" Analysis

The wide dispersion in hours worked documented above means that overtime exposure — the central element of Plaintiff's FLSA claim — varies dramatically from driver to driver and from week to week. A significant percentage of drivers may have worked zero overtime hours in many or most weeks, making collective treatment of overtime claims inappropriate. The named Plaintiff's hours are a statistical outlier relative to the collective mean and median, and his work experience cannot fairly be extrapolated to the approximately 1,187 drivers in the proposed collective. Individualized inquiry into each driver's weekly hours, seasonal patterns, and temporal trajectory would be essential to determine liability and calculate damages, and such inquiry would overwhelm any common questions susceptible to collective resolution.

VI. ANALYSIS OF GROSS WEEKLY PAY

A. Data Source and Methodology

33. To analyze gross weekly pay, I relied upon SLG's per-delivery compensation records for all 1,187 drivers during the Relevant Period. Under the ICOA, drivers are compensated on a per-delivery basis rather than an hourly or salaried basis. The per-delivery rate is not uniform; rather, it varies based on three factors: (a) package weight, (b) delivery distance from the distribution hub, and (c) delivery type, classified as either "standard residential" or "white-glove" service. Standard residential deliveries involve routine doorstep delivery of packages, while white-glove deliveries involve in-home placement, assembly assistance, or other enhanced services. Per-delivery rates range from $14.50 to $38.00, with standard residential deliveries generally commanding rates at the lower end of the range and white-glove deliveries commanding rates at the upper end.

34. For each driver, I calculated weekly gross pay by aggregating all per-delivery payments received during each calendar week in which the driver was active. The resulting data set encompasses the same approximately 97,634 driver-week observations described in connection with the hours-worked analysis above.

B. Aggregate Statistics

35. Across all driver-week observations, the mean weekly gross pay was approximately $1,078. The median weekly gross pay was approximately $987, indicating a right-skewed distribution in which a subset of higher-earning drivers pulls the mean upward. The standard deviation was approximately $412, and the range extended from approximately $210 per week (reflecting drivers who completed only a small number of deliveries in a given week) to approximately $2,630 per week (reflecting drivers who completed a high volume of white-glove deliveries in a particularly active week). The coefficient of variation was approximately 38.2%, indicating even greater dispersion in pay than in hours worked. By any standard measure, this level of variability reflects a population with highly heterogeneous economic outcomes.

C. Hub-Level Pay Variation

36. Mean weekly gross pay varied substantially across SLG's fourteen (14) distribution hubs. The three largest Texas hubs — Fort Worth (HUB-01), Dallas (HUB-02), and Houston (HUB-03) — exhibited the highest mean weekly gross pay, ranging from approximately $1,145 at HUB-02 (Dallas) to approximately $1,218 at HUB-03 (Houston). These hubs also had the highest concentrations of white-glove delivery opportunities, which command per-delivery rates at the upper end of the $14.50–$38.00 range. By contrast, smaller hubs focused predominantly on standard residential delivery — such as Lubbock (HUB-13), Shreveport (HUB-11), and Little Rock (HUB-10) — exhibited mean weekly gross pay ranging from approximately $672 at HUB-13 (Lubbock) to approximately $758 at HUB-10 (Little Rock).

37. I performed a one-way ANOVA to test whether the differences in mean weekly gross pay across hubs were statistically significant. The ANOVA yielded an F-statistic of 57.14 with 13 and 97,620 degrees of freedom, and a p-value of less than 0.001. This result is highly statistically significant and confirms that between-hub differences in mean weekly pay are not attributable to random variation. The detailed ANOVA results are presented in Exhibit C. Within-hub variation in weekly pay was also substantial, with within-hub standard deviations ranging from approximately $280 to approximately $470.

D. Delivery Mix as Pay Driver

38. To better understand the drivers of pay variation, I analyzed the proportion of white-glove versus standard residential deliveries performed by each driver across the Relevant Period. This analysis revealed that delivery mix is a primary determinant of gross weekly pay and effective hourly earnings. White-glove deliveries, which require additional skills such as furniture assembly, in-home placement, and customer interaction, command per-delivery rates ranging from approximately $28.00 to $38.00. Standard residential deliveries command per-delivery rates ranging from approximately $14.50 to $22.00.

39. Drivers who actively pursued and completed a higher proportion of white-glove deliveries earned materially more than drivers who performed primarily standard residential deliveries. This difference reflects individualized economic choice: under the ICOA, drivers are not assigned to a particular delivery type but rather have the ability to accept or decline individual delivery offers based on the delivery type, location, and compensation indicated in the RouteSync dispatch notification. Drivers who developed the skills and customer relationships necessary for white-glove work, and who affirmatively chose to accept white-glove assignments when available, earned commensurately higher compensation.

40. Named Plaintiff Marcus Delano performed primarily standard residential deliveries during the Relevant Period. His per-delivery rate averaged approximately $21.50, and his average weekly gross pay was approximately $1,247 (reflecting his high volume of deliveries, consistent with his above-average weekly hours). However, his effective hourly rate — calculated as gross weekly pay divided by weekly hours worked — was approximately $21.50 per hour. Other drivers in the proposed collective who concentrated on white-glove deliveries achieved effective hourly rates above $28.00 per hour, and in some cases as high as $32.40 per hour. Still other drivers who worked fewer hours and completed fewer deliveries had effective hourly rates as low as approximately $15.80 per hour.

E. Implications.

41. The substantial variation in gross weekly pay across the proposed collective reflects individualized delivery mix, volume, and effort — not a uniform pay practice susceptible to collective adjudication. The effective hourly rates imputed from the data range from approximately $15.80 to approximately $32.40, a spread that is more than two-to-one. Under Plaintiff's theory, overtime compensation must be calculated based on each driver's "regular rate" of pay, which would in turn depend on each driver's individualized delivery mix, volume, and hours worked during the applicable workweek.

42. This calculation cannot be performed on a collective basis. Each driver's regular rate would need to be determined individually, and any overtime damages would require a driver-by-driver inquiry into the number of hours exceeding 40 in each week and the applicable regular rate for that week. This is precisely the type of individualized inquiry that courts have recognized as fatal to conditional certification under the FLSA.

43. Named Plaintiff Delano's average weekly gross pay of $1,247 and effective hourly rate of approximately $21.50 per hour are not representative of the proposed collective. His pay reflects his high-hour, standard-residential delivery work pattern, which differs materially from the delivery patterns, compensation levels, and effective rates experienced by the majority of the proposed collective.

44. The hub-level pay variation further underscores the heterogeneity of the proposed collective. Drivers at the three largest Texas hubs earned, on average, approximately 50–80% more per week than drivers at the smallest hubs. This variation is attributable in substantial part to differences in delivery volume, white-glove opportunity concentration, and local market conditions — all factors that are hub-specific and driver-specific, not susceptible to collective resolution.

45. In sum, the pay data confirm that the proposed collective is not a group of individuals subject to a common pay practice. To the contrary, gross pay and effective hourly rates are driven by individualized choices and circumstances that vary dramatically across drivers and hubs.

VII. ANALYSIS OF DELIVERY ACCEPTANCE RATES

A. The 85% Minimum Acceptance Rate

46. Under all three versions of the ICOA, SLG requires drivers to maintain a minimum delivery acceptance rate of 85% per rolling 30-day period. The acceptance rate is calculated as the number of delivery offers accepted by the driver divided by the total number of delivery offers extended to the driver through the RouteSync platform during the applicable 30-day period.

47. The ICOA specifies that a driver whose acceptance rate falls below 85% during any rolling 30-day period "may be subject to a performance review and, upon repeated failure to maintain the minimum acceptance rate, may be subject to termination of this Agreement." Critically, falling below the 85% threshold does not trigger automatic termination. Rather, it triggers a discretionary "performance review" process, the outcomes of which — as discussed below — have varied significantly across drivers and hubs.

48. Plaintiff alleges in his Motion for Conditional Certification that the 85% minimum acceptance rate constitutes an instrument of employer control over scheduling, compelling drivers to accept substantially all delivery offers and thereby negating their purported ability to set their own hours and workload. My analysis of the data does not support this characterization.

B. Data and Methodology

49. I used RouteSync dispatch data to calculate each driver's monthly acceptance rate — defined as deliveries accepted divided by deliveries offered — for each rolling 30-day period during the Relevant Period. This calculation was performed for all 1,187 drivers in the proposed collective. The resulting data set encompasses approximately 32,750 driver-month observations.

C. Distribution of Acceptance Rates

50. Across all driver-month observations, the mean acceptance rate was approximately 91.3%. The median acceptance rate was approximately 92.7%, indicating a slightly left-skewed distribution with a tail of lower-acceptance-rate drivers. The standard deviation was approximately 6.8 percentage points, and the range extended from approximately 72% to 100%.

51. A meaningful number of driver-month observations reflected acceptance rates below the 85% contractual threshold. Specifically, approximately 14.6% of all driver-month observations (approximately 4,782 out of 32,750 driver-months) showed acceptance rates below 85% at some point during the Relevant Period. This is a substantial figure, representing nearly one in seven driver-months.

52. I further examined the consequences experienced by drivers who fell below the 85% threshold. Of the approximately 4,782 driver-months in which the acceptance rate fell below 85%, SLG records indicate that formal performance reviews were initiated in only approximately 612 instances — roughly 12.8% of sub-threshold driver-months. In the remaining approximately 87.2% of instances in which a driver's acceptance rate fell below 85%, no formal performance review was initiated, and the driver continued operating on the platform without documented consequence. Moreover, of the 612 instances in which a formal performance review was initiated, only 47 resulted in termination of the ICOA — representing less than 1% of all sub-threshold driver-months.

53. These findings indicate that the 85% minimum acceptance rate was not rigidly or uniformly enforced. A significant number of drivers operated below the threshold for extended periods — in some cases, for multiple consecutive months — without formal consequence. Some drivers maintained acceptance rates in the mid-70% range for quarters at a time while continuing to receive delivery offers and compensation through RouteSync. This pattern is inconsistent with the characterization of the acceptance rate requirement as an instrument of compulsory control over driver schedules.

D. Hub-Level Variation in Acceptance Rates

54. Acceptance rate patterns differed significantly across SLG's fourteen (14) distribution hubs. Larger hubs with higher delivery volume — particularly Houston (HUB-03), Dallas (HUB-02), and Fort Worth (HUB-01) — exhibited higher mean acceptance rates (ranging from approximately 92.5% to 94.1%) and lower variance. This pattern is consistent with the greater availability of delivery offers at high-volume hubs: when delivery offers are abundant, drivers are more likely to find offers that match their preferences and are therefore more likely to accept a high proportion of offers.

55. Smaller hubs with lower delivery volume — particularly Lubbock (HUB-13), Shreveport (HUB-11), and Little Rock (HUB-10) — exhibited lower mean acceptance rates (ranging from approximately 86.3% to 88.7%) and higher variance. Drivers at these hubs were more likely to receive delivery offers that were unattractive due to long distances, low per-delivery compensation, or inconvenient timing, and were correspondingly more likely to decline offers and fall below the 85% threshold.

56. I performed a one-way ANOVA to test whether the differences in mean acceptance rates across hubs were statistically significant. The ANOVA yielded an F-statistic of 29.43 with 13 and 32,736 degrees of freedom, and a p-value of less than 0.001. This result confirms that between-hub differences in acceptance rates are statistically significant and not attributable to random variation. The detailed ANOVA results are presented in Exhibit C.

E. Relationship Between Acceptance Rate and Hours/Pay

57. I performed a regression analysis to examine the relationship between each driver's acceptance rate and his or her weekly hours worked and gross weekly pay. The results indicate that acceptance rate is positively correlated with both hours worked (coefficient = 0.78 hours per percentage point of acceptance rate, p < 0.001) and weekly gross pay (coefficient = $14.30 per percentage point of acceptance rate, p < 0.001). However, the R-squared values for these regressions are modest (approximately 0.14 for hours and 0.11 for pay), indicating that acceptance rate explains only a small fraction of the total variance in hours and pay. The relationship is statistically significant but not deterministic: many drivers with lower acceptance rates nonetheless worked substantial hours and earned substantial pay, while some drivers with high acceptance rates worked relatively few hours.

58. The direction of the relationship — drivers with lower acceptance rates tending to work fewer hours and earn less — is consistent with voluntary choice rather than employer compulsion. Drivers who chose to decline a larger proportion of delivery offers thereby chose to work fewer hours and earn less income, reflecting the exercise of independent discretion over workload and scheduling. This is precisely the type of entrepreneurial choice that characterizes an independent contractor relationship rather than an employment relationship.

F. Implications for Control Analysis

The wide range of acceptance rates observed across the proposed collective, combined with the lack of consistent enforcement of the 85% floor, undermines the characterization of this contractual provision as an instrument of uniform employer control. Rather, the data demonstrate that drivers exercised varying degrees of discretion in accepting or declining work, that the consequences of falling below the threshold were inconsistent and often nonexistent, and that the degree of actual control exercised by SLG over any given driver's scheduling and workload was highly individualized. Determining the impact of the acceptance rate requirement on any particular driver's classification would require individualized inquiry into that driver's actual acceptance behavior, SLG's actual enforcement response, and the driver's reasons for accepting or declining particular offers. Such inquiry cannot be performed on a collective basis and would overwhelm any common questions.

VIII. ANALYSIS OF ROUTE DEVIATION FREQUENCY

A. Data Source and Methodology

59. To assess the degree of operational autonomy exercised by drivers in performing their delivery services, I analyzed route deviation data by comparing RouteSync-assigned routes with actual route data captured by GPS telematics devices. For each delivery run, RouteSync generates a "suggested" delivery sequence and navigation path optimized for efficiency. Drivers are not contractually required to follow the RouteSync-suggested route; the ICOA states that "Contractor shall determine the manner and means of performing deliveries, including but not limited to the route, sequence, and timing of deliveries, subject to any delivery windows communicated to the customer."

60. I defined a "route deviation" as any departure from the RouteSync-suggested delivery sequence or navigation path by more than 0.5 miles or 10 minutes from the suggested schedule. This threshold was selected to exclude trivial or de minimis variations (e.g., a driver taking a slightly different street due to a road closure) while capturing meaningful exercises of autonomy (e.g., reordering the delivery sequence, making personal stops, or taking a materially different route to a delivery destination).

61. Due to the computational intensity of the GPS comparison analysis, this analysis was performed on a stratified random sample of 200 drivers drawn from across all fourteen (14) distribution hubs, with the sample stratified proportionally to hub size. In addition, the full route deviation analysis was performed for the named Plaintiff, Marcus Delano. The sampling methodology and results are described in greater detail below. I have no reason to believe that the sample results are not representative of the full population of 1,187 drivers.

B. Aggregate Findings

62. Across the 200-driver sample, the mean route deviation frequency was approximately 3.7 deviations per driver per week. The standard deviation was approximately 2.9 deviations per week, yielding a coefficient of variation of approximately 78.4%. This is an extremely high level of dispersion, indicating that route deviation behavior was highly heterogeneous across the sampled drivers.

63. The distribution of route deviation frequency was markedly right-skewed. Some drivers deviated from RouteSync-suggested routes on nearly every delivery run — in extreme cases, recording 10 or more deviations per week — while other drivers rarely deviated, recording fewer than 1 deviation per week on average. The interquartile range extended from approximately 1.4 to approximately 5.6 deviations per week, further confirming the wide spread in route adherence patterns across the proposed collective.

C. Types of Deviations

64. My analysis identified three principal categories of route deviations:

(a) Sequence deviations — instances in which a driver delivered packages in a different order than the sequence suggested by RouteSync. Sequence deviations were the most common type, accounting for approximately 52% of all observed deviations. These deviations ranged from minor reorderings of two or three stops to wholesale rearrangements of an entire delivery run.

(b) Path deviations — instances in which a driver took a materially different road or navigation path than the one suggested by RouteSync, including making personal stops (e.g., fueling, meals, errands) during a delivery run. Path deviations accounted for approximately 31% of all observed deviations.

(c) Timing deviations — instances in which a driver commenced a delivery run significantly earlier or later than the RouteSync-assigned window, or completed deliveries outside the standard 7:00 AM–8:00 PM delivery window. Timing deviations accounted for approximately 17% of all observed deviations.

65. The prevalence of all three deviation types demonstrates that RouteSync suggestions were treated by many drivers as advisory rather than mandatory. Drivers exercised meaningful discretion not only in the order and route of their deliveries but also in the timing of their delivery activity. This is consistent with the ICOA's characterization of route and sequence as within the contractor's discretion and inconsistent with the degree of employer control over the "manner and means" of work performance that would characterize an employment relationship.

D. Hub-Level Variation

66. Route deviation frequency varied significantly across SLG's fourteen (14) distribution hubs. Among the sampled drivers, the highest mean deviation rates were observed at HUB-05 (Austin, Texas), with a mean of approximately 5.2 deviations per driver per week, and HUB-07 (Tulsa, Oklahoma), with a mean of approximately 4.8 deviations per week. The lowest mean deviation rates were observed at HUB-02 (Dallas, Texas), with a mean of approximately 2.6 deviations per week, and HUB-03 (Houston, Texas), with a mean of approximately 2.8 deviations per week.

67. This hub-level variation likely reflects differences in delivery network density, geographic layout, and hub-level management practices. The larger metropolitan hubs (Dallas, Houston) have denser delivery networks with shorter inter-stop distances, which may make RouteSync-suggested routes more efficient and therefore less likely to prompt driver deviation. Conversely, hubs in more dispersed geographic areas (Austin, Tulsa) may generate RouteSync suggestions that are less efficient or less responsive to local conditions, prompting drivers to exercise more frequent independent judgment regarding routing and sequencing. Regardless of the underlying causes, the hub-level variation in route deviation frequency further supports the conclusion that operational control differed by location and that the proposed collective cannot be treated as a homogeneous group.

E. Named Plaintiff's Route Deviation Pattern

68. Named Plaintiff Marcus Delano's average route deviation frequency during the Relevant Period was approximately 2.1 deviations per week — below the collective sample mean of 3.7 deviations per week and below the sample median of approximately 3.2 deviations per week. Delano's relatively low deviation frequency places him at approximately the 29th percentile of the sample distribution, meaning that approximately 71% of sampled drivers deviated from RouteSync-suggested routes more frequently than Delano.

69. Plaintiff Delano's experience of relatively high route adherence is thus unrepresentative of the broader driver population. If the named Plaintiff's work pattern were taken as indicative of the collective's experience, it would understate the degree of operational autonomy actually exercised by the majority of drivers and paint a misleading picture of the degree of control exercised by SLG over drivers' delivery operations.

F. Implications.

70. The route deviation data demonstrate that drivers in the proposed collective exercised significant and highly variable degrees of autonomy in how they performed their delivery work. The degree of actual control exercised by SLG over routing and delivery sequence varied significantly across drivers and across hubs. A determination of whether any given driver was subject to the type of control indicative of an employment relationship — as opposed to the type of general quality assurance oversight consistent with an independent contractor relationship — would require individualized inquiry into that driver's actual route deviation behavior, SLG's response to deviations, and the circumstances surrounding specific deviations. Such inquiry cannot be performed on a collective basis.

IX. ANALYSIS OF MULTI-PLATFORM ENGAGEMENT AND OUTSIDE ECONOMIC ACTIVITY

A. Data Sources and Methodology

71. To assess the entrepreneurial opportunity factor of the economic reality test, I analyzed whether drivers in the proposed collective engaged with competing delivery or logistics platforms, or pursued other outside economic activity, during the Relevant Period. This analysis drew upon multiple data sources:

(a) SLG driver availability logs and RouteSync activity records, which I analyzed to identify periods when drivers were inactive on RouteSync during standard delivery hours (7:00 AM–8:00 PM, Monday through Saturday) despite being logged in to the platform or having previously indicated availability for that time period;

(b) Self-reported data from driver onboarding questionnaires, administered to drivers at the time of ICOA execution, which asked whether the driver was currently performing work for other delivery or logistics platforms;

(c) SLG internal records reflecting driver requests for schedule accommodations, modified availability windows, or other accommodations citing other work commitments;

(d) Cross-reference with publicly available records and independent contractor databases, where available, to identify drivers with active engagements with other known delivery and logistics platforms in SLG's service areas; and

(e) SLG records relating to subcontracting requests submitted under the ICOA's subcontracting provisions.

72. I acknowledge that the data sources described above may understate the true extent of multi-platform engagement and outside economic activity among the proposed collective, because not all drivers disclosed outside work in onboarding questionnaires, not all drivers requested schedule accommodations for outside work, and publicly available records are incomplete. Accordingly, the figures I report below should be understood as conservative estimates of multi-platform engagement.

B. Findings — Multi-Platform Engagement

73. Of the 1,187 drivers in the proposed collective, an estimated 247 drivers — approximately 20.8% — showed evidence of engagement with other delivery or logistics platforms during the Relevant Period. The evidence supporting this finding included one or more of the following indicators for each driver: (a) extended periods of RouteSync inactivity during peak delivery hours inconsistent with the driver's historical activity patterns, (b) affirmative disclosure of other platform work in driver onboarding materials, and (c) SLG records reflecting driver requests for schedule accommodations citing work for other delivery or logistics services.

74. The multi-platform engagement rates varied significantly by hub. The highest rates were observed at HUB-05 (Austin), where approximately 31% of drivers showed evidence of multi-platform engagement; HUB-02 (Dallas), where approximately 26% of drivers showed such evidence; and HUB-03 (Houston), where approximately 24% of drivers showed such evidence. These three hubs are located in large metropolitan areas where multiple competing delivery platforms operate and where drivers have the greatest opportunity to diversify their income sources.

75. The lowest multi-platform engagement rates were observed at HUB-13 (Lubbock), where approximately 7% of drivers showed evidence of multi-platform engagement, and HUB-10 (Little Rock), where approximately 9% of drivers showed such evidence. These hubs are located in smaller metropolitan areas with fewer competing delivery platforms and correspondingly fewer opportunities for multi-platform work.

76. The substantial variation in multi-platform engagement rates across hubs further demonstrates that the entrepreneurial opportunity available to drivers was not uniform across the proposed collective but rather depended on hub-specific market conditions and individual driver choices.

C. Findings — Subcontracting

77. Under all three versions of the ICOA, drivers were permitted to subcontract their delivery obligations to other individuals with prior written approval from SLG. The ICOA provides that "Contractor may, with the prior written approval of Company, engage one or more subcontractors to perform Contractor's obligations under this Agreement, provided that such subcontractor(s) meet Company's insurance, safety, and vehicle requirements."

78. SLG records reflect that 23 formal subcontracting requests were submitted during the Relevant Period. Of these 23 requests, 8 were approved by SLG. The remaining 15 requests were either denied (7 requests, typically due to the proposed subcontractor's failure to meet insurance or vehicle requirements) or withdrawn by the driver prior to SLG decision (8 requests).

79. The relatively low number of subcontracting requests does not, in my opinion, necessarily indicate that SLG's subcontracting provisions were illusory or that subcontracting was effectively prohibited. The application process existed and was utilized by some drivers. The 8 approved requests demonstrate that SLG did grant subcontracting authority in appropriate cases. The low overall volume of requests more likely reflects drivers' lack of interest in or need for subcontracting rather than SLG prohibition. Under ICOA v3.0 (January 2023), the subcontracting approval process was somewhat streamlined relative to earlier versions: the approval timeline was reduced from 30 days to 14 days, and the insurance documentation requirements were simplified.

D. Findings — Other Business Activity

80. Approximately 89 drivers in the proposed collective — approximately 7.5% — maintained separate business registrations (e.g., sole proprietorships, limited liability companies) during the Relevant Period, indicating entrepreneurial activity beyond their SLG delivery work. These business registrations were identified through publicly available state business filing records and, in some instances, through driver-reported information in onboarding questionnaires.

81. The types of businesses maintained by these drivers varied widely and included landscaping services, auto detailing, handyman services, personal training, and other small-scale service businesses. Additionally, a subset of these 89 drivers filed Schedule C tax returns (as indicated in self-reported onboarding data) reporting income from multiple sources, further confirming that their economic relationship with SLG was one of several concurrent income-generating activities rather than an exclusive employment relationship.

E. Implications for Economic Reality and Conditional Certification

82. The entrepreneurial opportunity factor is a core element of the economic reality test used to determine whether workers are employees or independent contractors under the FLSA. The data presented above demonstrate that entrepreneurial opportunity varied dramatically across the proposed collective. Approximately one in five drivers (20.8%) showed evidence of multi-platform engagement, and approximately one in thirteen (7.5%) maintained separate business registrations — both indicators of meaningful entrepreneurial opportunity exercised independently of SLG.

83. Conversely, the remaining drivers in the proposed collective — those who did not engage with competing platforms or maintain separate businesses — may have had a more economically dependent relationship with SLG. Whether such drivers lacked entrepreneurial opportunity because SLG restricted their ability to pursue outside work, or because they voluntarily chose to focus exclusively on SLG delivery work, is a question that can only be resolved through individualized inquiry into each driver's circumstances.

84. Drivers who worked for multiple platforms or maintained independent businesses had fundamentally different economic relationships with SLG than those who worked exclusively for SLG. The former group exercised entrepreneurial discretion in allocating their time and effort across multiple income sources; the latter group did not. Treating these two groups as a single, similarly situated collective for purposes of FLSA conditional certification would ignore a material dimension of heterogeneity that goes to the heart of the classification inquiry.

85. This heterogeneity precludes the "similarly situated" finding necessary for conditional certification. The entrepreneurial opportunity analysis cannot be performed on a collective basis because the relevant facts — whether a driver worked for other platforms, maintained a separate business, sought subcontracting arrangements, or chose to work exclusively for SLG — are inherently individualized and vary dramatically across the proposed collective.

X. ICOA VERSION DIFFERENCES AND THEIR IMPACT ON ANALYSIS

A. Three ICOA Versions

86. During the Relevant Period, SLG utilized three successive versions of the Independent Contractor Operating Agreement:

(a) ICOA v1.0 (2018): The original template, drafted by the law firm of Pennfield & Daye LLP. This version contained the per-delivery compensation structure, the 85% minimum acceptance rate, the subcontracting-with-approval provision, and a non-compete clause restricting drivers from performing delivery services for competing platforms within a 50-mile radius of their assigned hub for 12 months following termination. ICOA v1.0 did not contain an arbitration clause, a class or collective action waiver, or a delegation clause.

(b) ICOA v2.0 (August 2020): This version revised the payment schedule to adjust per-delivery rates for certain delivery types, updated vehicle specifications to require model year 2015 or newer vehicles, and modified the subcontracting provisions to add a background check requirement for proposed subcontractors. Like ICOA v1.0, this version did not contain an arbitration clause, a class or collective action waiver, or a delegation clause.

(c) ICOA v3.0 (January 2023): This version added a mandatory arbitration clause requiring all disputes arising under or related to the ICOA to be resolved through binding individual arbitration administered by the American Arbitration Association. The arbitration clause included a class and collective action waiver providing that "Contractor agrees to bring any and all claims arising under this Agreement, including but not limited to claims under the Fair Labor Standards Act, on an individual basis only, and waives any right to bring, join, or participate in any class, collective, or representative action." ICOA v3.0 also contained a delegation clause providing that questions of arbitrability would be resolved by the arbitrator. Approximately 640 of the 1,187 drivers in the proposed collective signed ICOA v3.0.

B. Impact on Analysis

87. The existence of three materially different ICOA versions introduces a significant additional layer of heterogeneity into the proposed collective. Drivers who signed ICOA v1.0 are subject to different contractual terms — including different payment rates, different vehicle requirements, different subcontracting procedures, and a different non-compete provision — than drivers who signed ICOA v2.0 or v3.0.

88. Most significantly, the presence or absence of an arbitration clause with a collective action waiver directly affects the legal rights and remedies available to each driver. The approximately 640 drivers who signed ICOA v3.0 are, at least presumptively, subject to a mandatory individual arbitration requirement and a collective action waiver that may preclude their participation in this action. The approximately 547 drivers who signed ICOA v1.0 or v2.0 are not subject to any such requirement. This fundamental difference in legal posture among the members of the proposed collective is itself a barrier to the "similarly situated" finding required for conditional certification.

89. Moreover, the different subcontracting and payment provisions across ICOA versions mean that the degree of alleged control exercised by SLG, and the economic terms of the relationship between SLG and its drivers, varied by contract version. Drivers who signed ICOA v2.0, for example, were subject to a more stringent subcontracting approval process (requiring background checks for proposed subcontractors) than drivers who signed ICOA v1.0, while drivers who signed ICOA v3.0 were subject to a streamlined subcontracting process relative to v2.0. These differences in contractual terms further undermine the premise that the proposed collective members shared a common, uniform relationship with SLG.

C. Named Plaintiff's ICOA Version

90. Named Plaintiff Marcus Delano signed ICOA v1.0, the original 2018 version. His contractual terms differ materially from those of the approximately 640 drivers who signed ICOA v3.0. Most notably, Delano's ICOA does not contain an arbitration clause or a collective action waiver, while the ICOAs signed by more than half the proposed collective do. Delano's experience under ICOA v1.0 — with its older payment schedule, different vehicle specifications, and different subcontracting provisions — cannot fairly be treated as representative of the experience of drivers operating under the materially revised terms of ICOA v3.0.

91. The ICOA version differences also have implications for the merits of the misclassification claim. To the extent that the degree of control exercised by SLG over its drivers is relevant to the classification analysis, that degree of control varied across ICOA versions because the contractual provisions defining the parties' respective rights and obligations differed across versions. A collective determination of the clas

UNITED STATES DISTRICT COURT NORTHERN DISTRICT OF TEXAS DALLAS DIVISION
MARCUS DELANO, individually and on behalf of all others similarly situated,
Plaintiff,
v.
STONEFIELD LOGISTICS GROUP INC.,
Defendant.
Case No. 3:25-cv-01847-RBC
Hon. Rebecca Chen, U.S. District Judge
DECLARATION OF DR. PRIYA RAMANATHAN IN SUPPORT OF DEFENDANT STONEFIELD LOGISTICS GROUP INC.'S RESPONSE IN OPPOSITION TO PLAINTIFF'S MOTION FOR CONDITIONAL CERTIFICATION
I, Dr. Priya Ramanathan, declare under penalty of perjury pursuant to 28 U.S.C. § 1746 as follows:
This declaration is submitted in support of Defendant Stonefield Logistics Group Inc.'s ("SLG" or "Defendant") Response in Opposition to the Motion for Conditional Certification filed by Plaintiff Marcus Delano ("Plaintiff" or "Delano") on July 14, 2025. The opinions expressed herein are based on my review and analysis of the data, documents, and materials described below, and are offered to a reasonable degree of economic and statistical certainty.
III. QUALIFICATIONS AND BACKGROUND
A. Professional Position
1. I am the Lead Consultant and Senior Vice President at Corbin Risk Advisors LLC ("CRA"), a litigation consulting firm specializing in expert economic and statistical analysis. CRA's principal offices are located at 1901 N. Moore Street, Suite 1020, Arlington, Virginia 22209. CRA provides analytical and consulting services to law firms, corporations, and governmental entities across a range of complex litigation and regulatory matters. My practice at CRA focuses primarily on labor economics, employment classification disputes, and statistical analysis in connection with wage and hour litigation, including collective actions under the Fair Labor Standards Act ("FLSA") and class actions under analogous state wage and hour statutes.
B. Educational Background
2. I received my Doctor of Philosophy (Ph.D.) in Economics, with a specialization in labor economics, from the Rutledge School of Social Sciences at Hargrove University, Charlottesville, Virginia, in 2007. My doctoral dissertation, titled "Heterogeneous Labor Supply Responses to Platform-Based Compensation Structures," was supervised by Professor Alan M. Whitford and was awarded the Hargrove Department of Economics Dissertation Prize for that academic year.
3. I received my Master of Arts (M.A.) in Applied Statistics from Caldwell Polytechnic Institute, Atlanta, Georgia, in 2003. My master's thesis focused on the application of hierarchical linear modeling techniques to multi-site employment data.
4. I received my Bachelor of Arts (B.A.), summa cum laude, with a double major in Mathematics and Economics from Wentworth College, Boston, Massachusetts, in 2001. I was elected to Phi Beta Kappa during my junior year.
C. Professional Experience
5. I have over eighteen (18) years of professional experience in the fields of labor economics, statistical analysis, and employment litigation consulting. Over the course of my career, I have provided expert economic and statistical analysis in more than forty-five (45) FLSA collective action and state-law class action matters involving claims of misclassification, off-the-clock work, unpaid overtime, and related wage and hour allegations. These engagements have spanned numerous industries, including logistics and delivery services, healthcare, financial services, technology, retail, and hospitality.
6. I have been deposed or provided trial testimony in approximately twenty (20) cases in federal and state courts across the United States, including the United States District Courts for the Southern District of New York, the Northern District of Illinois, the Central District of California, the Eastern District of Pennsylvania, the Middle District of Florida, and the Western District of Texas, as well as state courts in New York, California, Illinois, and Virginia. A complete listing of my prior testimony over the past four years is included in my curriculum vitae, attached hereto as Exhibit F.
7. I have authored or co-authored more than a dozen articles published in peer-reviewed academic and professional journals on topics including worker classification, wage and hour economics, labor market analysis, and the economics of platform-based work arrangements. My published work has appeared in the Journal of Labor Economics, the Industrial and Labor Relations Review, the Journal of Applied Econometrics, and the Journal of Human Resources, among other publications. A complete listing of my publications is included in my curriculum vitae, attached hereto as Exhibit F.
8. Prior to joining CRA in 2014, I served as an Associate Professor of Economics at Dunmore University, Richmond, Virginia, where I taught graduate and undergraduate courses in labor economics, econometrics, and applied statistics from 2007 through 2014. During my academic tenure, I supervised multiple doctoral dissertations and master's theses related to labor economics and employment policy.
D. Engagement in This Matter
9. I was retained by the law firm of Calderwood & Harkness LLP on behalf of Defendant Stonefield Logistics Group Inc. pursuant to an Expert Retention Agreement dated May 8, 2025. My engagement in this matter is governed by the terms of that agreement, which provides for compensation at a rate of $550 per hour for my time and $375 per hour for the time of associate analysts working under my direction. My compensation, and the compensation of any analysts assisting me, is in no way contingent upon the outcome of this litigation, the content of any opinions I express, or the nature of any testimony I may provide.
E. Scope of Assignment
10. I was asked to analyze whether the approximately 1,187 individuals who performed delivery driving services for SLG under Independent Contractor Operating Agreements ("ICOAs") during the period from April 30, 2022 through the present (the "Relevant Period") are "similarly situated" for purposes of FLSA conditional certification. More specifically, I was asked to analyze the following areas: (1) variance in weekly hours worked across and within SLG's fourteen (14) distribution hubs; (2) variance in gross weekly pay across and within hubs; (3) delivery acceptance rate patterns relative to the 85% minimum acceptance rate specified in the ICOAs; (4) route deviation frequency as a measure of operational autonomy; and (5) multi-platform engagement and outside economic activity as measures of entrepreneurial opportunity. My analysis spans all fourteen (14) SLG distribution hubs located across the states of Texas, Oklahoma, Louisiana, and Arkansas.
F. Materials Reviewed
11. In forming my opinions, I reviewed and considered the following categories of materials:
(a) RouteSync dispatch records for all 1,187 drivers for the period April 30, 2022 through June 30, 2025, including delivery offer data, acceptance and rejection records, delivery completion timestamps, and assignment histories;
(b) GPS telematics data from SLG-provided tracking devices installed in driver vehicles across all fourteen (14) distribution hubs, encompassing location data, route tracking, and stop-time calculations;
(c) SLG per-delivery compensation records (pay records) for all 1,187 drivers during the Relevant Period, including per-delivery rates, delivery type classifications, and weekly payment summaries;
(d) Executed Independent Contractor Operating Agreements in all three versions: ICOA v1.0 (original 2018 template), ICOA v2.0 (revised August 2020), and ICOA v3.0 (revised January 2023);
(e) SLG driver onboarding records, performance review records, and driver availability logs;
(f) SLG hub-level operational policies and procedures manuals for all fourteen (14) distribution hubs;
(g) Plaintiff's Complaint, Dkt. No. 1, and Plaintiff's Motion for Conditional Certification, Dkt. No. 34, together with all supporting declarations and exhibits;
(h) The declarations of Trevor Sandoval, Karen Lund, and Dennis Faulkner, submitted in support of Plaintiff's Motion;
(i) Publicly available data concerning competing delivery and logistics platform operations in SLG's service areas, including rate schedules and independent contractor agreements for comparable platforms;
(j) Other documents and data sources as listed in Exhibit A, attached hereto and incorporated by reference.
G. Methodology Overview
12. My analyses employ standard and widely accepted statistical techniques, including descriptive statistics (mean, median, standard deviation, range, interquartile range, coefficient of variation), one-way analysis of variance ("ANOVA"), coefficient of variation analysis, ordinary least squares ("OLS") regression analysis, and chi-square tests of independence. All statistical analyses were performed using Penmark Systems (Ashgrove Health IT) and R version 4.3.1 (R Core Team). All analyses were performed, and all opinions expressed herein are held, to a reasonable degree of economic and statistical certainty.
IV. SUMMARY OF OPINIONS
13. Opinion 1 (Weekly Hours Worked): The 1,187 drivers in the proposed collective exhibit statistically significant variation in weekly hours worked across and within SLG's fourteen (14) distribution hubs. The mean weekly hours worked across the proposed collective is approximately 47.3 hours, with a standard deviation of approximately 12.8 hours and a coefficient of variation of approximately 27.1%. The named Plaintiff's average weekly hours of approximately 58 hours places him in the top decile of the distribution. Approximately 38.2% of all driver-week observations reflect fewer than 40 hours worked, meaning a substantial proportion of the proposed collective would have no overtime claim in any given week. These findings undermine the premise that the proposed collective members share a common experience with respect to hours worked.
14. Opinion 2 (Gross Weekly Pay): Gross weekly pay varies substantially across drivers and hubs, with a mean of approximately $1,078, a standard deviation of approximately $412, and a coefficient of variation of approximately 38.2%. This variation reflects individualized economic circumstances — including each driver's delivery mix, volume, and effort — rather than a uniform pay practice susceptible to collective resolution.
15. Opinion 3 (Delivery Acceptance Rates): Delivery acceptance rates vary widely across the proposed collective, with a standard deviation of approximately 6.8 percentage points around a mean of approximately 91.3%. Approximately 14.6% of driver-months fell below the 85% contractual threshold during the Relevant Period, and many of those drivers continued operating without formal consequence. This variation demonstrates that drivers exercised meaningfully different levels of discretion in accepting or declining delivery assignments, belying the notion of uniform employer control over scheduling and workload.
16. Opinion 4 (Route Deviation Frequency): Route deviation data demonstrates that drivers exercised significant and variable degrees of autonomy in how they performed deliveries. The mean route deviation frequency was approximately 3.7 per driver per week, with a coefficient of variation of approximately 78.4% — an extremely high level of dispersion. The named Plaintiff's average deviation frequency of approximately 2.1 per week places him below the collective mean, rendering his experience of relatively high route adherence unrepresentative.
17. Opinion 5 (Multi-Platform Engagement): A material subset of the proposed collective — approximately 247 drivers (20.8%) — showed evidence of engagement with competing delivery platforms or other outside economic activity during the Relevant Period. Multi-platform engagement rates varied dramatically by hub, ranging from approximately 7% at HUB-13 (Lubbock) to approximately 31% at HUB-05 (Austin). This variation in entrepreneurial opportunity is inconsistent with a unitary employment relationship and cannot be assessed on a collective basis. Additionally, approximately 89 drivers (7.5%) maintained separate business registrations during the Relevant Period. The overall conclusion is that the proposed collective is not a homogeneous group of similarly situated individuals; material individualized differences in work patterns, autonomy, and economic circumstances predominate over any common features.
V. ANALYSIS OF WEEKLY HOURS WORKED
A. Data Source and Methodology
18. To calculate weekly hours worked for each driver, I relied upon two complementary data sources: (a) GPS telematics data captured by SLG-provided tracking devices installed in each driver's vehicle, and (b) RouteSync login and logout timestamps recorded each time a driver commenced or concluded a delivery session through the RouteSync dispatch platform. By cross-referencing these two independent data streams, I was able to construct a reliable and comprehensive measure of weekly hours worked for each driver at each of SLG's fourteen (14) distribution hubs.
19. The data analyzed spans the full Relevant Period — from April 30, 2022 through June 30, 2025 — encompassing approximately 161 calendar weeks. During this period, the 1,187 unique drivers in the proposed collective were active for varying durations, reflecting the natural turnover and intermittent engagement characteristic of an independent contractor delivery network. The total number of driver-week observations analyzed was approximately 97,634. This figure represents the sum of individual active weeks across all 1,187 drivers, with each driver contributing data only for those weeks in which he or she accepted at least one delivery through RouteSync.
20. For purposes of my analysis, I defined "hours worked" in a given week as the total time elapsed from the first delivery dispatch acceptance to the last delivery completion during that week, net of breaks exceeding thirty (30) minutes, as identified through gaps in GPS movement data. This definition is consistent with standard approaches to measuring compensable time under Department of Labor guidance and is deliberately inclusive — it credits drivers with time during which they were actively engaged in delivery operations, including time spent driving between delivery locations, loading packages at hub facilities, and awaiting dispatch while logged in to the RouteSync platform. This definition, if anything, overstates compensable time for many drivers, as it does not deduct time that drivers may have spent on personal activities while nominally logged in to the platform.
B. Aggregate Statistics
21. Across all 97,634 driver-week observations, the mean weekly hours worked was approximately 47.3 hours. The median weekly hours worked was approximately 44.1 hours — meaningfully lower than the mean, indicating a right-skewed distribution in which a subset of high-hour drivers pulls the mean upward. The standard deviation was approximately 12.8 hours, and the range extended from approximately 15 hours per week (at the low end, reflecting drivers who operated on a part-time or intermittent basis) to approximately 68 hours per week (at the high end, reflecting the most active drivers). The coefficient of variation — the ratio of the standard deviation to the mean — was approximately 27.1%, indicating a high degree of dispersion relative to the central tendency. By conventional standards in labor economics, a coefficient of variation above 20% is considered indicative of substantial heterogeneity within a population.
22. I note that Plaintiff alleges in his declaration that he "typically" worked approximately 58 hours per week. My analysis of SLG's data confirms that Plaintiff Delano's average weekly hours during the Relevant Period was approximately 58.2 hours. However, this figure is far above the collective mean of 47.3 hours and above the collective median of 44.1 hours. In fact, Plaintiff Delano's average weekly hours places him at approximately the 91st percentile of the distribution — meaning that approximately 91% of all driver-week observations reflected fewer weekly hours than Plaintiff's average. His experience is not representative of the proposed collective.
C. Hub-Level Variation
23. I performed a hub-by-hub analysis of mean weekly hours worked across all fourteen (14) SLG distribution hubs. The results, summarized in Exhibit B and detailed below, reveal substantial variation in mean hours across hubs:
24. As the table above illustrates, mean weekly hours ranged from approximately 35.4 hours at HUB-13 (Lubbock) to approximately 52.4 hours at HUB-01 (Fort Worth) — a spread of 17.0 hours between the highest and lowest hub means. The three largest Texas hubs (Fort Worth, Dallas, and Houston) all exhibited mean weekly hours above 49 hours, while the smaller and more geographically remote hubs (Lubbock, Shreveport, Little Rock) showed mean weekly hours below 40 hours. This pattern is consistent with the greater delivery volume and demand density at the larger metropolitan hubs, which afforded drivers at those locations more opportunity — and, in some cases, more incentive — to work longer hours.
25. I performed a one-way analysis of variance ("ANOVA") to test whether the differences in mean weekly hours across hubs were statistically significant. The ANOVA yielded an F-statistic of 42.87 with 13 and 97,620 degrees of freedom, and a p-value of less than 0.001. This result is highly statistically significant and indicates that between-hub differences in mean weekly hours are not attributable to chance. The between-hub variance accounted for approximately 18.3% of total variance in weekly hours across the proposed collective. The detailed ANOVA results are presented in Exhibit C.
26. Although between-hub variance is substantial, within-hub variance is also significant, as discussed in the following paragraphs. This dual finding — of both between-hub and within-hub heterogeneity — indicates that the proposed collective is heterogeneous at every level of aggregation.
D. Within-Hub Variation
27. Even within individual hubs, the standard deviations in weekly hours worked are large, ranging from approximately 9.0 hours at HUB-11 (Shreveport) to approximately 13.1 hours at HUB-01 (Fort Worth). These within-hub standard deviations are comparable in magnitude to the overall standard deviation of 12.8 hours, confirming that substantial variation in weekly hours exists among drivers operating out of the same facility, under the same hub manager, and in the same geographic market.
28. To illustrate this point concretely, I examined the distribution of weekly hours worked at HUB-01 (Fort Worth, Texas) — the hub where the named Plaintiff, Marcus Delano, performed his delivery services. Among the 142 drivers at HUB-01, the mean weekly hours worked was approximately 52.4 hours, the median was approximately 50.8 hours, the standard deviation was approximately 13.1 hours, and the range extended from approximately 22 hours per week to approximately 68 hours per week. This 46-hour spread within a single hub underscores the wide variability in driver work patterns, even when external factors such as geographic market, hub management, and delivery demand are held constant.
29. Across the entire proposed collective, approximately 38.2% of all driver-week observations reflected fewer than 40 hours worked during the applicable week. This finding is of considerable significance for the conditional certification inquiry. Under Plaintiff's theory, the proposed collective members are entitled to overtime compensation because they were misclassified as independent contractors and, as putative employees, should have been paid overtime for hours worked in excess of 40 per week. However, if a driver worked fewer than 40 hours in a given week, that driver would have no overtime claim for that week regardless of the merits of any classification dispute. The fact that more than one-third of all driver-week observations fall below the 40-hour threshold means that a substantial portion of the proposed collective would have no overtime exposure in many or most weeks, and that individualized inquiry into each driver's weekly hours would be required to determine liability and damages.
30. Named Plaintiff Marcus Delano's average of approximately 58 hours per week places him at approximately the 91st percentile of the distribution. His experience of routinely working well in excess of 40 hours per week is not shared by the majority of the proposed collective. To the extent Plaintiff holds himself out as representative of the collective, the data do not support that characterization.
E. Temporal Variation
31. Hours worked by individual drivers also fluctuated significantly over time. I analyzed the within-driver temporal variation by calculating the coefficient of variation of weekly hours for each individual driver across his or her active weeks during the Relevant Period. The within-driver coefficients of variation ranged from approximately 15% to approximately 45%, with a median of approximately 28%. This indicates that even a single driver's weekly hours were not stable from week to week; rather, drivers increased and decreased their activity levels across quarters and seasons based on personal circumstances, market conditions, and other individualized factors.
32. Moreover, seasonal patterns in hours worked differed meaningfully by hub. The Houston and Dallas hubs (HUB-02 and HUB-03) exhibited pronounced holiday-season spikes in weekly hours, with fourth-quarter average hours approximately 15–20% higher than first-quarter average hours. By contrast, the smaller and more rural hubs — such as Lubbock (HUB-13), Little Rock (HUB-10), and Shreveport (HUB-11) — showed comparatively flat seasonal patterns, with fourth-quarter hours only marginally (approximately 3–5%) higher than first-quarter hours. These differing seasonal dynamics further demonstrate that the work experience of drivers at different hubs was not uniform and that aggregate or "one-size-fits-all" treatment of the proposed collective would obscure material differences among its members.
F. Implications for "Similarly Situated" Analysis
The wide dispersion in hours worked documented above means that overtime exposure — the central element of Plaintiff's FLSA claim — varies dramatically from driver to driver and from week to week. A significant percentage of drivers may have worked zero overtime hours in many or most weeks, making collective treatment of overtime claims inappropriate. The named Plaintiff's hours are a statistical outlier relative to the collective mean and median, and his work experience cannot fairly be extrapolated to the approximately 1,187 drivers in the proposed collective. Individualized inquiry into each driver's weekly hours, seasonal patterns, and temporal trajectory would be essential to determine liability and calculate damages, and such inquiry would overwhelm any common questions susceptible to collective resolution.
VI. ANALYSIS OF GROSS WEEKLY PAY
A. Data Source and Methodology
33. To analyze gross weekly pay, I relied upon SLG's per-delivery compensation records for all 1,187 drivers during the Relevant Period. Under the ICOA, drivers are compensated on a per-delivery basis rather than an hourly or salaried basis. The per-delivery rate is not uniform; rather, it varies based on three factors: (a) package weight, (b) delivery distance from the distribution hub, and (c) delivery type, classified as either "standard residential" or "white-glove" service. Standard residential deliveries involve routine doorstep delivery of packages, while white-glove deliveries involve in-home placement, assembly assistance, or other enhanced services. Per-delivery rates range from $14.50 to $38.00, with standard residential deliveries generally commanding rates at the lower end of the range and white-glove deliveries commanding rates at the upper end.
34. For each driver, I calculated weekly gross pay by aggregating all per-delivery payments received during each calendar week in which the driver was active. The resulting data set encompasses the same approximately 97,634 driver-week observations described in connection with the hours-worked analysis above.
B. Aggregate Statistics
35. Across all driver-week observations, the mean weekly gross pay was approximately $1,078. The median weekly gross pay was approximately $987, indicating a right-skewed distribution in which a subset of higher-earning drivers pulls the mean upward. The standard deviation was approximately $412, and the range extended from approximately $210 per week (reflecting drivers who completed only a small number of deliveries in a given week) to approximately $2,630 per week (reflecting drivers who completed a high volume of white-glove deliveries in a particularly active week). The coefficient of variation was approximately 38.2%, indicating even greater dispersion in pay than in hours worked. By any standard measure, this level of variability reflects a population with highly heterogeneous economic outcomes.
C. Hub-Level Pay Variation
36. Mean weekly gross pay varied substantially across SLG's fourteen (14) distribution hubs. The three largest Texas hubs — Fort Worth (HUB-01), Dallas (HUB-02), and Houston (HUB-03) — exhibited the highest mean weekly gross pay, ranging from approximately $1,145 at HUB-02 (Dallas) to approximately $1,218 at HUB-03 (Houston). These hubs also had the highest concentrations of white-glove delivery opportunities, which command per-delivery rates at the upper end of the $14.50–$38.00 range. By contrast, smaller hubs focused predominantly on standard residential delivery — such as Lubbock (HUB-13), Shreveport (HUB-11), and Little Rock (HUB-10) — exhibited mean weekly gross pay ranging from approximately $672 at HUB-13 (Lubbock) to approximately $758 at HUB-10 (Little Rock).
37. I performed a one-way ANOVA to test whether the differences in mean weekly gross pay across hubs were statistically significant. The ANOVA yielded an F-statistic of 57.14 with 13 and 97,620 degrees of freedom, and a p-value of less than 0.001. This result is highly statistically significant and confirms that between-hub differences in mean weekly pay are not attributable to random variation. The detailed ANOVA results are presented in Exhibit C. Within-hub variation in weekly pay was also substantial, with within-hub standard deviations ranging from approximately $280 to approximately $470.
D. Delivery Mix as Pay Driver
38. To better understand the drivers of pay variation, I analyzed the proportion of white-glove versus standard residential deliveries performed by each driver across the Relevant Period. This analysis revealed that delivery mix is a primary determinant of gross weekly pay and effective hourly earnings. White-glove deliveries, which require additional skills such as furniture assembly, in-home placement, and customer interaction, command per-delivery rates ranging from approximately $28.00 to $38.00. Standard residential deliveries command per-delivery rates ranging from approximately $14.50 to $22.00.
39. Drivers who actively pursued and completed a higher proportion of white-glove deliveries earned materially more than drivers who performed primarily standard residential deliveries. This difference reflects individualized economic choice: under the ICOA, drivers are not assigned to a particular delivery type but rather have the ability to accept or decline individual delivery offers based on the delivery type, location, and compensation indicated in the RouteSync dispatch notification. Drivers who developed the skills and customer relationships necessary for white-glove work, and who affirmatively chose to accept white-glove assignments when available, earned commensurately higher compensation.
40. Named Plaintiff Marcus Delano performed primarily standard residential deliveries during the Relevant Period. His per-delivery rate averaged approximately $21.50, and his average weekly gross pay was approximately $1,247 (reflecting his high volume of deliveries, consistent with his above-average weekly hours). However, his effective hourly rate — calculated as gross weekly pay divided by weekly hours worked — was approximately $21.50 per hour. Other drivers in the proposed collective who concentrated on white-glove deliveries achieved effective hourly rates above $28.00 per hour, and in some cases as high as $32.40 per hour. Still other drivers who worked fewer hours and completed fewer deliveries had effective hourly rates as low as approximately $15.80 per hour.
E. Implications.
41. The substantial variation in gross weekly pay across the proposed collective reflects individualized delivery mix, volume, and effort — not a uniform pay practice susceptible to collective adjudication. The effective hourly rates imputed from the data range from approximately $15.80 to approximately $32.40, a spread that is more than two-to-one. Under Plaintiff's theory, overtime compensation must be calculated based on each driver's "regular rate" of pay, which would in turn depend on each driver's individualized delivery mix, volume, and hours worked during the applicable workweek.
42. This calculation cannot be performed on a collective basis. Each driver's regular rate would need to be determined individually, and any overtime damages would require a driver-by-driver inquiry into the number of hours exceeding 40 in each week and the applicable regular rate for that week. This is precisely the type of individualized inquiry that courts have recognized as fatal to conditional certification under the FLSA.
43. Named Plaintiff Delano's average weekly gross pay of $1,247 and effective hourly rate of approximately $21.50 per hour are not representative of the proposed collective. His pay reflects his high-hour, standard-residential delivery work pattern, which differs materially from the delivery patterns, compensation levels, and effective rates experienced by the majority of the proposed collective.
44. The hub-level pay variation further underscores the heterogeneity of the proposed collective. Drivers at the three largest Texas hubs earned, on average, approximately 50–80% more per week than drivers at the smallest hubs. This variation is attributable in substantial part to differences in delivery volume, white-glove opportunity concentration, and local market conditions — all factors that are hub-specific and driver-specific, not susceptible to collective resolution.
45. In sum, the pay data confirm that the proposed collective is not a group of individuals subject to a common pay practice. To the contrary, gross pay and effective hourly rates are driven by individualized choices and circumstances that vary dramatically across drivers and hubs.
VII. ANALYSIS OF DELIVERY ACCEPTANCE RATES
A. The 85% Minimum Acceptance Rate
46. Under all three versions of the ICOA, SLG requires drivers to maintain a minimum delivery acceptance rate of 85% per rolling 30-day period. The acceptance rate is calculated as the number of delivery offers accepted by the driver divided by the total number of delivery offers extended to the driver through the RouteSync platform during the applicable 30-day period.
47. The ICOA specifies that a driver whose acceptance rate falls below 85% during any rolling 30-day period "may be subject to a performance review and, upon repeated failure to maintain the minimum acceptance rate, may be subject to termination of this Agreement." Critically, falling below the 85% threshold does not trigger automatic termination. Rather, it triggers a discretionary "performance review" process, the outcomes of which — as discussed below — have varied significantly across drivers and hubs.
48. Plaintiff alleges in his Motion for Conditional Certification that the 85% minimum acceptance rate constitutes an instrument of employer control over scheduling, compelling drivers to accept substantially all delivery offers and thereby negating their purported ability to set their own hours and workload. My analysis of the data does not support this characterization.
B. Data and Methodology
49. I used RouteSync dispatch data to calculate each driver's monthly acceptance rate — defined as deliveries accepted divided by deliveries offered — for each rolling 30-day period during the Relevant Period. This calculation was performed for all 1,187 drivers in the proposed collective. The resulting data set encompasses approximately 32,750 driver-month observations.
C. Distribution of Acceptance Rates
50. Across all driver-month observations, the mean acceptance rate was approximately 91.3%. The median acceptance rate was approximately 92.7%, indicating a slightly left-skewed distribution with a tail of lower-acceptance-rate drivers. The standard deviation was approximately 6.8 percentage points, and the range extended from approximately 72% to 100%.
51. A meaningful number of driver-month observations reflected acceptance rates below the 85% contractual threshold. Specifically, approximately 14.6% of all driver-month observations (approximately 4,782 out of 32,750 driver-months) showed acceptance rates below 85% at some point during the Relevant Period. This is a substantial figure, representing nearly one in seven driver-months.
52. I further examined the consequences experienced by drivers who fell below the 85% threshold. Of the approximately 4,782 driver-months in which the acceptance rate fell below 85%, SLG records indicate that formal performance reviews were initiated in only approximately 612 instances — roughly 12.8% of sub-threshold driver-months. In the remaining approximately 87.2% of instances in which a driver's acceptance rate fell below 85%, no formal performance review was initiated, and the driver continued operating on the platform without documented consequence. Moreover, of the 612 instances in which a formal performance review was initiated, only 47 resulted in termination of the ICOA — representing less than 1% of all sub-threshold driver-months.
53. These findings indicate that the 85% minimum acceptance rate was not rigidly or uniformly enforced. A significant number of drivers operated below the threshold for extended periods — in some cases, for multiple consecutive months — without formal consequence. Some drivers maintained acceptance rates in the mid-70% range for quarters at a time while continuing to receive delivery offers and compensation through RouteSync. This pattern is inconsistent with the characterization of the acceptance rate requirement as an instrument of compulsory control over driver schedules.
D. Hub-Level Variation in Acceptance Rates
54. Acceptance rate patterns differed significantly across SLG's fourteen (14) distribution hubs. Larger hubs with higher delivery volume — particularly Houston (HUB-03), Dallas (HUB-02), and Fort Worth (HUB-01) — exhibited higher mean acceptance rates (ranging from approximately 92.5% to 94.1%) and lower variance. This pattern is consistent with the greater availability of delivery offers at high-volume hubs: when delivery offers are abundant, drivers are more likely to find offers that match their preferences and are therefore more likely to accept a high proportion of offers.
55. Smaller hubs with lower delivery volume — particularly Lubbock (HUB-13), Shreveport (HUB-11), and Little Rock (HUB-10) — exhibited lower mean acceptance rates (ranging from approximately 86.3% to 88.7%) and higher variance. Drivers at these hubs were more likely to receive delivery offers that were unattractive due to long distances, low per-delivery compensation, or inconvenient timing, and were correspondingly more likely to decline offers and fall below the 85% threshold.
56. I performed a one-way ANOVA to test whether the differences in mean acceptance rates across hubs were statistically significant. The ANOVA yielded an F-statistic of 29.43 with 13 and 32,736 degrees of freedom, and a p-value of less than 0.001. This result confirms that between-hub differences in acceptance rates are statistically significant and not attributable to random variation. The detailed ANOVA results are presented in Exhibit C.
E. Relationship Between Acceptance Rate and Hours/Pay
57. I performed a regression analysis to examine the relationship between each driver's acceptance rate and his or her weekly hours worked and gross weekly pay. The results indicate that acceptance rate is positively correlated with both hours worked (coefficient = 0.78 hours per percentage point of acceptance rate, p < 0.001) and weekly gross pay (coefficient = $14.30 per percentage point of acceptance rate, p < 0.001). However, the R-squared values for these regressions are modest (approximately 0.14 for hours and 0.11 for pay), indicating that acceptance rate explains only a small fraction of the total variance in hours and pay. The relationship is statistically significant but not deterministic: many drivers with lower acceptance rates nonetheless worked substantial hours and earned substantial pay, while some drivers with high acceptance rates worked relatively few hours.
58. The direction of the relationship — drivers with lower acceptance rates tending to work fewer hours and earn less — is consistent with voluntary choice rather than employer compulsion. Drivers who chose to decline a larger proportion of delivery offers thereby chose to work fewer hours and earn less income, reflecting the exercise of independent discretion over workload and scheduling. This is precisely the type of entrepreneurial choice that characterizes an independent contractor relationship rather than an employment relationship.
F. Implications for Control Analysis
The wide range of acceptance rates observed across the proposed collective, combined with the lack of consistent enforcement of the 85% floor, undermines the characterization of this contractual provision as an instrument of uniform employer control. Rather, the data demonstrate that drivers exercised varying degrees of discretion in accepting or declining work, that the consequences of falling below the threshold were inconsistent and often nonexistent, and that the degree of actual control exercised by SLG over any given driver's scheduling and workload was highly individualized. Determining the impact of the acceptance rate requirement on any particular driver's classification would require individualized inquiry into that driver's actual acceptance behavior, SLG's actual enforcement response, and the driver's reasons for accepting or declining particular offers. Such inquiry cannot be performed on a collective basis and would overwhelm any common questions.
VIII. ANALYSIS OF ROUTE DEVIATION FREQUENCY
A. Data Source and Methodology
59. To assess the degree of operational autonomy exercised by drivers in performing their delivery services, I analyzed route deviation data by comparing RouteSync-assigned routes with actual route data captured by GPS telematics devices. For each delivery run, RouteSync generates a "suggested" delivery sequence and navigation path optimized for efficiency. Drivers are not contractually required to follow the RouteSync-suggested route; the ICOA states that "Contractor shall determine the manner and means of performing deliveries, including but not limited to the route, sequence, and timing of deliveries, subject to any delivery windows communicated to the customer."
60. I defined a "route deviation" as any departure from the RouteSync-suggested delivery sequence or navigation path by more than 0.5 miles or 10 minutes from the suggested schedule. This threshold was selected to exclude trivial or de minimis variations (e.g., a driver taking a slightly different street due to a road closure) while capturing meaningful exercises of autonomy (e.g., reordering the delivery sequence, making personal stops, or taking a materially different route to a delivery destination).
61. Due to the computational intensity of the GPS comparison analysis, this analysis was performed on a stratified random sample of 200 drivers drawn from across all fourteen (14) distribution hubs, with the sample stratified proportionally to hub size. In addition, the full route deviation analysis was performed for the named Plaintiff, Marcus Delano. The sampling methodology and results are described in greater detail below. I have no reason to believe that the sample results are not representative of the full population of 1,187 drivers.
B. Aggregate Findings
62. Across the 200-driver sample, the mean route deviation frequency was approximately 3.7 deviations per driver per week. The standard deviation was approximately 2.9 deviations per week, yielding a coefficient of variation of approximately 78.4%. This is an extremely high level of dispersion, indicating that route deviation behavior was highly heterogeneous across the sampled drivers.
63. The distribution of route deviation frequency was markedly right-skewed. Some drivers deviated from RouteSync-suggested routes on nearly every delivery run — in extreme cases, recording 10 or more deviations per week — while other drivers rarely deviated, recording fewer than 1 deviation per week on average. The interquartile range extended from approximately 1.4 to approximately 5.6 deviations per week, further confirming the wide spread in route adherence patterns across the proposed collective.
C. Types of Deviations
64. My analysis identified three principal categories of route deviations:
(a) Sequence deviations — instances in which a driver delivered packages in a different order than the sequence suggested by RouteSync. Sequence deviations were the most common type, accounting for approximately 52% of all observed deviations. These deviations ranged from minor reorderings of two or three stops to wholesale rearrangements of an entire delivery run.
(b) Path deviations — instances in which a driver took a materially different road or navigation path than the one suggested by RouteSync, including making personal stops (e.g., fueling, meals, errands) during a delivery run. Path deviations accounted for approximately 31% of all observed deviations.
(c) Timing deviations — instances in which a driver commenced a delivery run significantly earlier or later than the RouteSync-assigned window, or completed deliveries outside the standard 7:00 AM–8:00 PM delivery window. Timing deviations accounted for approximately 17% of all observed deviations.
65. The prevalence of all three deviation types demonstrates that RouteSync suggestions were treated by many drivers as advisory rather than mandatory. Drivers exercised meaningful discretion not only in the order and route of their deliveries but also in the timing of their delivery activity. This is consistent with the ICOA's characterization of route and sequence as within the contractor's discretion and inconsistent with the degree of employer control over the "manner and means" of work performance that would characterize an employment relationship.
D. Hub-Level Variation
66. Route deviation frequency varied significantly across SLG's fourteen (14) distribution hubs. Among the sampled drivers, the highest mean deviation rates were observed at HUB-05 (Austin, Texas), with a mean of approximately 5.2 deviations per driver per week, and HUB-07 (Tulsa, Oklahoma), with a mean of approximately 4.8 deviations per week. The lowest mean deviation rates were observed at HUB-02 (Dallas, Texas), with a mean of approximately 2.6 deviations per week, and HUB-03 (Houston, Texas), with a mean of approximately 2.8 deviations per week.
67. This hub-level variation likely reflects differences in delivery network density, geographic layout, and hub-level management practices. The larger metropolitan hubs (Dallas, Houston) have denser delivery networks with shorter inter-stop distances, which may make RouteSync-suggested routes more efficient and therefore less likely to prompt driver deviation. Conversely, hubs in more dispersed geographic areas (Austin, Tulsa) may generate RouteSync suggestions that are less efficient or less responsive to local conditions, prompting drivers to exercise more frequent independent judgment regarding routing and sequencing. Regardless of the underlying causes, the hub-level variation in route deviation frequency further supports the conclusion that operational control differed by location and that the proposed collective cannot be treated as a homogeneous group.
E. Named Plaintiff's Route Deviation Pattern
68. Named Plaintiff Marcus Delano's average route deviation frequency during the Relevant Period was approximately 2.1 deviations per week — below the collective sample mean of 3.7 deviations per week and below the sample median of approximately 3.2 deviations per week. Delano's relatively low deviation frequency places him at approximately the 29th percentile of the sample distribution, meaning that approximately 71% of sampled drivers deviated from RouteSync-suggested routes more frequently than Delano.
69. Plaintiff Delano's experience of relatively high route adherence is thus unrepresentative of the broader driver population. If the named Plaintiff's work pattern were taken as indicative of the collective's experience, it would understate the degree of operational autonomy actually exercised by the majority of drivers and paint a misleading picture of the degree of control exercised by SLG over drivers' delivery operations.
F. Implications.
70. The route deviation data demonstrate that drivers in the proposed collective exercised significant and highly variable degrees of autonomy in how they performed their delivery work. The degree of actual control exercised by SLG over routing and delivery sequence varied significantly across drivers and across hubs. A determination of whether any given driver was subject to the type of control indicative of an employment relationship — as opposed to the type of general quality assurance oversight consistent with an independent contractor relationship — would require individualized inquiry into that driver's actual route deviation behavior, SLG's response to deviations, and the circumstances surrounding specific deviations. Such inquiry cannot be performed on a collective basis.
IX. ANALYSIS OF MULTI-PLATFORM ENGAGEMENT AND OUTSIDE ECONOMIC ACTIVITY
A. Data Sources and Methodology
71. To assess the entrepreneurial opportunity factor of the economic reality test, I analyzed whether drivers in the proposed collective engaged with competing delivery or logistics platforms, or pursued other outside economic activity, during the Relevant Period. This analysis drew upon multiple data sources:
(a) SLG driver availability logs and RouteSync activity records, which I analyzed to identify periods when drivers were inactive on RouteSync during standard delivery hours (7:00 AM–8:00 PM, Monday through Saturday) despite being logged in to the platform or having previously indicated availability for that time period;
(b) Self-reported data from driver onboarding questionnaires, administered to drivers at the time of ICOA execution, which asked whether the driver was currently performing work for other delivery or logistics platforms;
(c) SLG internal records reflecting driver requests for schedule accommodations, modified availability windows, or other accommodations citing other work commitments;
(d) Cross-reference with publicly available records and independent contractor databases, where available, to identify drivers with active engagements with other known delivery and logistics platforms in SLG's service areas; and
(e) SLG records relating to subcontracting requests submitted under the ICOA's subcontracting provisions.
72. I acknowledge that the data sources described above may understate the true extent of multi-platform engagement and outside economic activity among the proposed collective, because not all drivers disclosed outside work in onboarding questionnaires, not all drivers requested schedule accommodations for outside work, and publicly available records are incomplete. Accordingly, the figures I report below should be understood as conservative estimates of multi-platform engagement.
B. Findings — Multi-Platform Engagement
73. Of the 1,187 drivers in the proposed collective, an estimated 247 drivers — approximately 20.8% — showed evidence of engagement with other delivery or logistics platforms during the Relevant Period. The evidence supporting this finding included one or more of the following indicators for each driver: (a) extended periods of RouteSync inactivity during peak delivery hours inconsistent with the driver's historical activity patterns, (b) affirmative disclosure of other platform work in driver onboarding materials, and (c) SLG records reflecting driver requests for schedule accommodations citing work for other delivery or logistics services.
74. The multi-platform engagement rates varied significantly by hub. The highest rates were observed at HUB-05 (Austin), where approximately 31% of drivers showed evidence of multi-platform engagement; HUB-02 (Dallas), where approximately 26% of drivers showed such evidence; and HUB-03 (Houston), where approximately 24% of drivers showed such evidence. These three hubs are located in large metropolitan areas where multiple competing delivery platforms operate and where drivers have the greatest opportunity to diversify their income sources.
75. The lowest multi-platform engagement rates were observed at HUB-13 (Lubbock), where approximately 7% of drivers showed evidence of multi-platform engagement, and HUB-10 (Little Rock), where approximately 9% of drivers showed such evidence. These hubs are located in smaller metropolitan areas with fewer competing delivery platforms and correspondingly fewer opportunities for multi-platform work.
76. The substantial variation in multi-platform engagement rates across hubs further demonstrates that the entrepreneurial opportunity available to drivers was not uniform across the proposed collective but rather depended on hub-specific market conditions and individual driver choices.
C. Findings — Subcontracting
77. Under all three versions of the ICOA, drivers were permitted to subcontract their delivery obligations to other individuals with prior written approval from SLG. The ICOA provides that "Contractor may, with the prior written approval of Company, engage one or more subcontractors to perform Contractor's obligations under this Agreement, provided that such subcontractor(s) meet Company's insurance, safety, and vehicle requirements."
78. SLG records reflect that 23 formal subcontracting requests were submitted during the Relevant Period. Of these 23 requests, 8 were approved by SLG. The remaining 15 requests were either denied (7 requests, typically due to the proposed subcontractor's failure to meet insurance or vehicle requirements) or withdrawn by the driver prior to SLG decision (8 requests).
79. The relatively low number of subcontracting requests does not, in my opinion, necessarily indicate that SLG's subcontracting provisions were illusory or that subcontracting was effectively prohibited. The application process existed and was utilized by some drivers. The 8 approved requests demonstrate that SLG did grant subcontracting authority in appropriate cases. The low overall volume of requests more likely reflects drivers' lack of interest in or need for subcontracting rather than SLG prohibition. Under ICOA v3.0 (January 2023), the subcontracting approval process was somewhat streamlined relative to earlier versions: the approval timeline was reduced from 30 days to 14 days, and the insurance documentation requirements were simplified.
D. Findings — Other Business Activity
80. Approximately 89 drivers in the proposed collective — approximately 7.5% — maintained separate business registrations (e.g., sole proprietorships, limited liability companies) during the Relevant Period, indicating entrepreneurial activity beyond their SLG delivery work. These business registrations were identified through publicly available state business filing records and, in some instances, through driver-reported information in onboarding questionnaires.
81. The types of businesses maintained by these drivers varied widely and included landscaping services, auto detailing, handyman services, personal training, and other small-scale service businesses. Additionally, a subset of these 89 drivers filed Schedule C tax returns (as indicated in self-reported onboarding data) reporting income from multiple sources, further confirming that their economic relationship with SLG was one of several concurrent income-generating activities rather than an exclusive employment relationship.
E. Implications for Economic Reality and Conditional Certification
82. The entrepreneurial opportunity factor is a core element of the economic reality test used to determine whether workers are employees or independent contractors under the FLSA. The data presented above demonstrate that entrepreneurial opportunity varied dramatically across the proposed collective. Approximately one in five drivers (20.8%) showed evidence of multi-platform engagement, and approximately one in thirteen (7.5%) maintained separate business registrations — both indicators of meaningful entrepreneurial opportunity exercised independently of SLG.
83. Conversely, the remaining drivers in the proposed collective — those who did not engage with competing platforms or maintain separate businesses — may have had a more economically dependent relationship with SLG. Whether such drivers lacked entrepreneurial opportunity because SLG restricted their ability to pursue outside work, or because they voluntarily chose to focus exclusively on SLG delivery work, is a question that can only be resolved through individualized inquiry into each driver's circumstances.
84. Drivers who worked for multiple platforms or maintained independent businesses had fundamentally different economic relationships with SLG than those who worked exclusively for SLG. The former group exercised entrepreneurial discretion in allocating their time and effort across multiple income sources; the latter group did not. Treating these two groups as a single, similarly situated collective for purposes of FLSA conditional certification would ignore a material dimension of heterogeneity that goes to the heart of the classification inquiry.
85. This heterogeneity precludes the "similarly situated" finding necessary for conditional certification. The entrepreneurial opportunity analysis cannot be performed on a collective basis because the relevant facts — whether a driver worked for other platforms, maintained a separate business, sought subcontracting arrangements, or chose to work exclusively for SLG — are inherently individualized and vary dramatically across the proposed collective.
X. ICOA VERSION DIFFERENCES AND THEIR IMPACT ON ANALYSIS
A. Three ICOA Versions
86. During the Relevant Period, SLG utilized three successive versions of the Independent Contractor Operating Agreement:
(a) ICOA v1.0 (2018): The original template, drafted by the law firm of Pennfield & Daye LLP. This version contained the per-delivery compensation structure, the 85% minimum acceptance rate, the subcontracting-with-approval provision, and a non-compete clause restricting drivers from performing delivery services for competing platforms within a 50-mile radius of their assigned hub for 12 months following termination. ICOA v1.0 did not contain an arbitration clause, a class or collective action waiver, or a delegation clause.
(b) ICOA v2.0 (August 2020): This version revised the payment schedule to adjust per-delivery rates for certain delivery types, updated vehicle specifications to require model year 2015 or newer vehicles, and modified the subcontracting provisions to add a background check requirement for proposed subcontractors. Like ICOA v1.0, this version did not contain an arbitration clause, a class or collective action waiver, or a delegation clause.
(c) ICOA v3.0 (January 2023): This version added a mandatory arbitration clause requiring all disputes arising under or related to the ICOA to be resolved through binding individual arbitration administered by the American Arbitration Association. The arbitration clause included a class and collective action waiver providing that "Contractor agrees to bring any and all claims arising under this Agreement, including but not limited to claims under the Fair Labor Standards Act, on an individual basis only, and waives any right to bring, join, or participate in any class, collective, or representative action." ICOA v3.0 also contained a delegation clause providing that questions of arbitrability would be resolved by the arbitrator. Approximately 640 of the 1,187 drivers in the proposed collective signed ICOA v3.0.
B. Impact on Analysis
87. The existence of three materially different ICOA versions introduces a significant additional layer of heterogeneity into the proposed collective. Drivers who signed ICOA v1.0 are subject to different contractual terms — including different payment rates, different vehicle requirements, different subcontracting procedures, and a different non-compete provision — than drivers who signed ICOA v2.0 or v3.0.
88. Most significantly, the presence or absence of an arbitration clause with a collective action waiver directly affects the legal rights and remedies available to each driver. The approximately 640 drivers who signed ICOA v3.0 are, at least presumptively, subject to a mandatory individual arbitration requirement and a collective action waiver that may preclude their participation in this action. The approximately 547 drivers who signed ICOA v1.0 or v2.0 are not subject to any such requirement. This fundamental difference in legal posture among the members of the proposed collective is itself a barrier to the "similarly situated" finding required for conditional certification.
89. Moreover, the different subcontracting and payment provisions across ICOA versions mean that the degree of alleged control exercised by SLG, and the economic terms of the relationship between SLG and its drivers, varied by contract version. Drivers who signed ICOA v2.0, for example, were subject to a more stringent subcontracting approval process (requiring background checks for proposed subcontractors) than drivers who signed ICOA v1.0, while drivers who signed ICOA v3.0 were subject to a streamlined subcontracting process relative to v2.0. These differences in contractual terms further undermine the premise that the proposed collective members shared a common, uniform relationship with SLG.
C. Named Plaintiff's ICOA Version
90. Named Plaintiff Marcus Delano signed ICOA v1.0, the original 2018 version. His contractual terms differ materially from those of the approximately 640 drivers who signed ICOA v3.0. Most notably, Delano's ICOA does not contain an arbitration clause or a collective action waiver, while the ICOAs signed by more than half the proposed collective do. Delano's experience under ICOA v1.0 — with its older payment schedule, different vehicle specifications, and different subcontracting provisions — cannot fairly be treated as representative of the experience of drivers operating under the materially revised terms of ICOA v3.0.
91. The ICOA version differences also have implications for the merits of the misclassification claim. To the extent that the degree of control exercised by SLG over its drivers is relevant to the classification analysis, that degree of control varied across ICOA versions because the contractual provisions defining the parties' respective rights and obligations differed across versions. A collective determination of the clas

UNITED STATES DISTRICT COURT NORTHERN DISTRICT OF TEXAS DALLAS DIVISION

STONEFIELD LOGISTICS GROUP INC.,

III. QUALIFICATIONS AND BACKGROUND

IV. SUMMARY OF OPINIONS

V. ANALYSIS OF WEEKLY HOURS WORKED

VI. ANALYSIS OF GROSS WEEKLY PAY

VII. ANALYSIS OF DELIVERY ACCEPTANCE RATES

VIII. ANALYSIS OF ROUTE DEVIATION FREQUENCY

IX. ANALYSIS OF MULTI-PLATFORM ENGAGEMENT AND OUTSIDE ECONOMIC ACTIVITY

X. ICOA VERSION DIFFERENCES AND THEIR IMPACT ON ANALYSIS

Record

Parties

Dates

Amounts