RSM 2/26/26

Objective of the Study

  • Determine if location alters water consumption.
  • Study conducted at Armstrong Equine Center and NMZU horse farm with water sourced from both locations.

Experimental Design

  • Subjects: 12 mares
  • Random Assignment: Mares assigned to two treatments.
    • Period One:
    • 6 mares (identified as group 6A) assigned to treatment one (water from NMSU horse farm).
    • 6 mares (identified as group 6B) assigned to treatment two (water from Armstrong).
    • Period Two:
    • The mares that received treatment one in period one (6A) received treatment two in period two.
    • The mares that received treatment two in period one (6B) received treatment one in period two.

Treatment Objective

  • Objective was to induce thirst upon arrival by providing water from specified sources.

Blocking Mechanisms

  • Period as Blocking Factor: Period one and period two can be viewed as distinct blocks.
  • Treatment as Blocking Factor: The different water sources also serve as blocking factors.

Experimental Units and Interaction Testing

  • Units: 12 mares
    • Blocking consideration suggests that 12 experimental units minus one (degrees of freedom associated with experimental design).
  • Interaction Testing: Interaction between period and treatment is something that could be explored, although in this study, direct interaction testing may not have been conducted due to complexity.

Design Analysis

  • Aim to remove variability to yield cleaner evidence for stronger conclusions based on treatment effects.
  • Recognized differences between crossover designs and Latin square designs.
    • Crossover Design: Each horse becomes its own control, allowing it to experience both treatments across periods, controlling for individual variability (e.g., age, first treatment effects).
    • Latin Square Design: Considers interactions but can lead to loss of degrees of freedom due to the complexity of managing multiple treatments.

Benefits of Crossover Design

  • Allows for greater sensitivity and control over variability inherent to individual horses (variability in age, weight, etc.).
  • Helpful when available animals are heterogeneous (different ages/sizes), as it allows powerful conclusions to be drawn even with fewer uniform animals.

Managing Carryover Effects

  • To mitigate carryover effects in crossover studies, it is crucial to include an adaptation period when switching treatments (at least 10 days in this context).

Data Analysis

  • Data Collection: Specific water consumption in kilograms per day recorded for each animal and treatment.
  • Consumption per horse and across treatments totaled:
    • Period One water consumption ranged (sample values):
    • Horse 1: 0.12, Horse 2: 3.2, Horse 3: 3.1, Horse 4: 1.4, Horse 5: 4.3, etc.
    • Period Two switched treatment sources, data collection strategy remained consistent.

Statistical Treatment of Data

  • Total water consumption calculated by summing across periods:
    • Example total for one horse might have been computed, yielding totals 72.3 (specific values to be calculated by participants).
  • Total mean sum of squares calculated, with components related to:
    • Horse effect, treatment effect, period effect, and residual variance.
  • F-values calculated for treatment and interactions to determine statistical significance:
    • Formula used: F = \frac{\text{Mean Sum of Squares Treatment}}{\text{Mean Sum of Squares Error}}

Results Interpretation

  • Examined p-values to assess significance of treatment effects; p-values greater than thresholds indicated no significant difference in water consumption based on treatment.
  • Recognized general rule of thumb: If the difference between two means exceeds double the standard error, it suggests significance. However, due to data outcomes, this rule did not hold in the case of the study.

Conclusion

  • Final determination made based on analysis: water source has no significant effect on water consumption in mares, demonstrating effectiveness of the crossover design in yielding clear evidence despite inherent variability in horse populations.