Study guide

10.3 Evaluating Data Quality, Signal Integrity, Outliers, and Cleaning Procedures

NSCA Certified Performance and Sport Scientist CPSS

Track your progress as you read

Finished sections get a check mark.

1 Introduction: CPSS Exam Overview, Eligibility, and Case-Study Strategy

1.1 CPSS Exam Format, Eligibility Pathways, and the 2026 Blueprint 1.2 Mastering Athlete Case Studies and Research Review Scenarios

2 Chapter 1: Scientific Process, Adaptation Theory, and Periodization (Domain 1)

2.1 Training Principles, Adaptation Models, and Injury-Risk Mitigation 2.2 Evaluating Performance Programs Using the Scientific Process 2.3 Periodization Paradigms, Fitness-Fatigue Models, and Tapering

3 Chapter 2: Scientific Disciplines in Athlete Preparation (Domain 1)

3.1 Exercise Physiology, Bioenergetics, and Systemic Adaptations 3.2 Applied Biomechanics, Force-Time Kinetics, and Force-Velocity Profiling 3.3 Motor Learning, Skill Acquisition, and Constraints-Led Practice Design 3.4 Performance Nutrition, Recovery Modalities, and Sport Psychology

4 Chapter 3: Technology Review, Research Appraisal, and Knowledge Translation (Domain 1)

4.1 Critically Reviewing Sport Performance Products, Technologies, and Trends 4.2 Evaluating Research Studies: Design, Validity, and Practical Relevance 4.3 Communicating Research Findings to Support Evidence-Based Practice

5 Chapter 4: Sport Demands Analysis, KPIs, and Resource Strategy (Domain 2)

5.1 Researching Biomechanical, Tactical, and Technical Sport Demands 5.2 Physiological Demands, Injury Epidemiology, and Environmental Constraints 5.3 Qualitative and Mixed-Methods Research for Sport Demands Analysis 5.4 Collaborating with Stakeholders to Establish Key Performance Indicators 5.5 Identifying Resources and Strategies for Short- and Long-Term Goals

6 Chapter 5: Profiling Benchmarks and Needs Analysis Communication (Domain 2)

6.1 Developing Normative Benchmarks: Z-Scores, T-Scores, and Percentiles 6.2 Multidimensional Athlete Profiling and Return-to-Performance Milestones 6.3 Communicating Needs Analysis Results to Multidisciplinary Stakeholders

7 Chapter 6: Assessment Ethics, Safety Standards, and Tool Psychometrics (Domain 3)

7.1 Ethical, Legal, and Privacy Considerations in Athlete Data Collection 7.2 Evaluating Assessment Environments, Equipment, and Safety Standards 7.3 Validity, Reliability (ICC, CV%, TE), and Smallest Worthwhile Change

8 Chapter 7: Applied Assessment and Load-Monitoring Systems (Domain 3)

8.1 Movement Tracking Systems: GNSS/GPS, LPS, IMUs, and Optical Tracking 8.2 Neuromuscular Testing: Force Platforms, Isometric Pulls/Pushes, and VBT 8.3 Internal Load, Autonomic Readiness (HRV), Biomarkers, and PROMs

9 Chapter 8: Quality Assurance, Protocols, and Assessment Efficacy (Domain 3)

9.1 Applying Quality Assurance Techniques in Sport Science Data Collection 9.2 Developing, Documenting, and Standardizing Data Collection Protocols 9.3 Systematically Reevaluating Assessments, Protocols, and Interventions

10 Chapter 9: Data Governance, Storage Architecture, and Cleaning (Domain 4)

10.1 Data Privacy, Security Protocols, and Protecting Organizational Information 10.2 Designing Plans for Data Collection, Storage, Analysis, and AMS Workflows 10.3 Evaluating Data Quality, Signal Integrity, Outliers, and Cleaning Procedures

11 Chapter 10: Statistical Modeling, Feedback Delivery, and Analytical Limitations (Domain 4)

11.1 Inferential Statistics, Correlation, Regression, ANOVA, and Effect Sizes 11.2 Longitudinal Load Modeling: ACWR, EWMA, Monotony, Strain, and Change Detection 11.3 Data Visualization, Dashboard Design, and Timely Stakeholder Feedback 11.4 Communicating Analytical Limitations, Bias, and Uncertainty

10.3 Evaluating Data Quality, Signal Integrity, Outliers, and Cleaning Procedures

Key Takeaways Explain This Topic AI

A fourth-order, zero-lag (bidirectional dual-pass) low-pass Butterworth filter attenuates high-frequency noise above the cutoff frequency without shifting the temporal phase of biomechanical peaks.

Over-smoothing force-plate signals with an excessively low cutoff frequency clips true impact peaks and artificially depresses Rate of Force Development (RFD), while under-smoothing amplifies noise during differentiation.

Non-parametric outlier detectors (Tukey IQR fences and Median Absolute Deviation [MAD]) resist the masking effect that corrupts parametric Z-scores when extreme sensor spikes inflate the sample standard deviation.

Practitioners must never automatically delete a statistical outlier without domain plausibility and video/context verification, separating technical artifacts (e.g., HDOP satellite spikes) from true extreme biological workloads.

Missing data must be classified as MCAR, MAR, or MNAR, and missing longitudinal workloads should be imputed via individualized sRPE regression or drill-level position scaling while setting an explicit is_imputed metadata flag.

Last updated: October 2026

10.3 Evaluating Data Quality, Signal Integrity, Outliers, and Cleaning Procedures

Predictive models, workload ratios, and neuromuscular profiles are only as valid as the raw signals feeding them—a principle universally summarized as "Garbage In, Garbage Out" ( GIGO \text{GIGO} GIGO ). Raw streams from force plates, linear position transducers, inertial measurement units ( IMUs \text{IMUs} IMUs ), heart rate straps, and GNSS \text{GNSS} GNSS receivers inevitably contain high-frequency sensor noise, electromagnetic interference, multipath satellite reflections, and missing observations. A sport scientist must master digital signal processing ( DSP \text{DSP} DSP ), robust statistical outlier auditing, and principled missing-data imputation to clean datasets without erasing true biological extremes.

  • Signal Integrity and Digital Signal Processing ( DSP \text{DSP} DSP ) Fundamentals
  • Signal vs. High-Frequency Noise and Zero-Lag Butterworth Filtering

Any measured biomechanical waveform y ( t ) y(t) y ( t ) is the sum of the true underlying biological signal s ( t ) s(t) s ( t ) and additive noise n ( t ) n(t) n ( t ) :

y ( t ) = s ( t ) + n ( t ) y(t) = s(t) + n(t) y ( t ) = s ( t ) + n ( t )

In human movement biomechanics, voluntary segment kinematics and center-of-mass ground reaction forces ( GRF \text{GRF} GRF ) reside predominantly in the low-frequency spectrum below 10 – 20 Hz 10\text{–}20\text{ Hz} 10 – 20 Hz (with rapid passive impact transients during sprinting or landing extending up to 40 – 50 Hz 40\text{–}50\text{ Hz} 40 – 50 Hz ). Conversely, electrical mains hum ( 50 – 60 Hz 50\text{–}60\text{ Hz} 50 – 60 Hz ), cable whip, and force-plate frame resonance occupy higher frequency bands.

To isolate s ( t ) s(t) s ( t ) from high-frequency noise n ( t ) n(t) n ( t ) , the biomechanical gold standard is a fourth-order, zero-lag (bidirectional dual-pass) low-pass Butterworth digital filter :

Low-Pass Attenuation: Passes frequencies below a selected cutoff frequency ( f c f_c f c ​ ) unattenuated while progressively rolling off high-frequency noise above f c f_c f c ​ .

Zero Phase Shift (Dual-Pass Bidirectionality): A standard single-pass (forward-in-time) second-order filter introduces a temporal phase lag, shifting peak force or peak velocity several milliseconds later in time. By passing the second-order Butterworth filter forward through the time series and then passing it backward through the reversed time series, the phase shifts cancel out completely ( zero-lag ) while doubling the effective filter steepness to fourth-order .

Cutoff Frequency Selection (Winter's Residual Analysis) and the Smoothing Paradox

Selecting the optimal cutoff frequency f c f_c f c ​ via Winter's Residual Analysis —which plots the root-mean-square ( RMS \text{RMS} RMS ) residual between unfiltered and filtered signals across a spectrum of candidate cutoff frequencies—requires balancing a fundamental trade-off known as the Smoothing Paradox :

Over-Smoothing ( f c f_c f c ​ Set Too Low): If a practitioner applies an overly aggressive low-pass cutoff (e.g., f c = 4 Hz f_c = 4\text{ Hz} f c ​ = 4 Hz or 6 Hz 6\text{ Hz} 6 Hz ) to a 1,000 Hz 1{,}000\text{ Hz} 1 , 000 Hz vertical ground reaction force trace during an isometric mid-thigh pull or countermovement jump, the filter rounds off sharp physiological force onsets. This clips true peak force and severely underestimates time-derivative explosive metrics such as Rate of Force Development ( RFD = Δ F / Δ t \text{RFD} = \Delta F / \Delta t RFD = Δ F /Δ t ) .

Under-Smoothing ( f c f_c f c ​ Set Too High): If f c f_c f c ​ is set too high (or left unfiltered) when calculating kinematics from displacement data, residual high-frequency noise is catastrophically amplified during numerical differentiation ( displacement → velocity → acceleration \text{displacement} \rightarrow \text{velocity} \rightarrow \text{acceleration} displacement → velocity → acceleration ), because differentiation multiplies noise amplitude proportionally to its frequency ( ω \omega ω ) during the first derivative and by ω 2 \omega^2 ω 2 during the second derivative.

Heart Rate Variability ( HRV \text{HRV} HRV ) Ectopic Beat and Artifact Correction

Signal cleaning is equally critical in autonomic cardiac monitoring. The root mean square of successive differences ( RMSSD \text{RMSSD} RMSSD ) between adjacent R-R intervals ( IBI \text{IBI} IBI ) is highly sensitive to a single missed R-peak or ectopic beat. If an athlete has a steady resting heart rate of 75 bpm 75\text{ bpm} 75 bpm ( R-R interval = 800 ms \text{R-R interval} = 800\text{ ms} R-R interval = 800 ms ) and movement of the chest strap causes the sensor to miss a single beat, the recorded interval doubles to 1,600 ms 1{,}600\text{ ms} 1 , 600 ms . Because RMSSD \text{RMSSD} RMSSD squares successive differences:

RMSSD = 1 N − 1 ∑ i = 1 N − 1 ( IBI i + 1 − IBI i ) 2 \text{RMSSD} = \sqrt{\frac{1}{N - 1} \sum_{i=1}^{N-1} \left(\text{IBI}_{i+1} - \text{IBI}_i\right)^2} RMSSD = N − 1 1 ​ ∑ i = 1 N − 1 ​ ( IBI i + 1 ​ − IBI i ​ ) 2 ​

that single ( 1600 − 800 ) 2 = 640,000 ms 2 (1600 - 800)^2 = 640{,}000\text{ ms}^2 ( 1600 − 800 ) 2 = 640 , 000 ms 2 error term can artificially inflate the athlete's 60 -second 60\text{-second} 60 -second RMSSD \text{RMSSD} RMSSD by > 50 – 100 % > 50\text{–}100\% > 50 – 100% , falsely portraying a fatigued athlete as parasympathetically hyper-recovered! Consequently, raw R-R series must undergo automated threshold artifact detection (flagging intervals that deviate by > 20 % > 20\% > 20% from a local 5-beat moving median) and cubic spline interpolation before computing lnRMSSD \text{lnRMSSD} lnRMSSD .

  • Outlier Detection: Parametric vs. Non-Parametric Rules and Plausibility Auditing
  • Statistical Outlier Detection Frameworks

Parametric Z Z Z -Score Rule ( ∣ Z ∣ > 3.0 |Z| > 3.0 ∣ Z ∣ > 3.0 ): Flags observations falling more than 3.0 3.0 3.0 (or 2.5 2.5 2.5 ) standard deviations from the sample mean ( Z = ( x − x ˉ ) / s Z = (x - \bar{x}) / s Z = ( x − x ˉ ) / s ). Critical Limitation (The Masking Effect): Both the sample mean x ˉ \bar{x} x ˉ and sample standard deviation s s s have a statistical breakdown point of 0 % 0\% 0% —meaning a single massive sensor error (e.g., a corrupted distance value of 95,000 m 95{,}000\text{ m} 95 , 000 m ) inflates s s s so dramatically that the outlier's own Z Z Z -score may remain below 3.0 3.0 3.0 while completely masking legitimate moderate outliers.

Non-Parametric Tukey Interquartile Range ( IQR \text{IQR} IQR ) Fences: Uses order statistics resistant to extreme contamination. Given the 25 th 25\text{th} 25 th percentile ( Q 1 Q_1 Q 1 ​ ) and 75 th 75\text{th} 75 th percentile ( Q 3 Q_3 Q 3 ​ ), where IQR = Q 3 − Q 1 \text{IQR} = Q_3 - Q_1 IQR = Q 3 ​ − Q 1 ​ :

Lower Fence = Q 1 − 1.5 ( IQR ) , Upper Fence = Q 3 + 1.5 ( IQR ) \text{Lower Fence} = Q_1 - 1.5(\text{IQR}), \quad \text{Upper Fence} = Q_3 + 1.5(\text{IQR}) Lower Fence = Q 1 ​ − 1.5 ( IQR ) , Upper Fence = Q 3 ​ + 1.5 ( IQR )

Values outside ± 1.5 ( IQR ) \pm 1.5(\text{IQR}) ± 1.5 ( IQR ) are flagged as mild outliers for inspection, while values beyond Q 1 − 3.0 ( IQR ) Q_1 - 3.0(\text{IQR}) Q 1 ​ − 3.0 ( IQR ) and Q 3 + 3.0 ( IQR ) Q_3 + 3.0(\text{IQR}) Q 3 ​ + 3.0 ( IQR ) are classified as extreme outliers.

3. Median Absolute Deviation ( MAD \text{MAD} MAD ): Possesses the maximum theoretical breakdown point of 50 % 50\% 50% :

MAD = median ( ∣ x i − median ( x ) ∣ ) , Z MAD = 0.6745 × ∣ x i − median ( x ) ∣ MAD \text{MAD} = \text{median}\left(|x_i - \text{median}(x)|\right), \quad Z_{\text{MAD}} = \frac{0.6745 \times |x_i - \text{median}(x)|}{\text{MAD}} MAD = median ( ∣ x i ​ − median ( x ) ∣ ) , Z MAD ​ = MAD 0.6745 × ∣ x i ​ − median ( x ) ∣ ​

Observations exceeding ∣ Z MAD ∣ > 3.5 |Z_{\text{MAD}}| > 3.5 ∣ Z MAD ​ ∣ > 3.5 are flagged for review.

The Cardinal Rule of Sport Science Outlier Cleaning

A statistical outlier flag is a trigger for investigation, never an automatic command for deletion. Before removing or modifying any extreme value, the sport scientist must conduct a Domain Plausibility and Contextual Audit to distinguish between two fundamentally different phenomena:

Technical / Sensor Artifacts (Delete or Re-Interpolate): Values that violate human biological limits or stem from verifiable hardware failure. Examples include a GNSS \text{GNSS} GNSS unit reporting a peak velocity of 14.2 m ⋅ s − 1 14.2\text{ m}\cdot\text{s}^{-1} 14.2 m ⋅ s − 1 ( 51.1 km ⋅ h − 1 51.1\text{ km}\cdot\text{h}^{-1} 51.1 km ⋅ h − 1 , exceeding the human world record of 12.42 m ⋅ s − 1 12.42\text{ m}\cdot\text{s}^{-1} 12.42 m ⋅ s − 1 ) accompanied by poor satellite geometry ( Horizontal Dilution of Precision [ HDOP \text{HDOP} HDOP ] = 4.5 = 4.5 = 4.5 and only 5 5 5 connected satellites near a stadium roof overhang); a heart rate monitor spiking to 248 bpm 248\text{ bpm} 248 bpm due to dry-electrode static interference; or a countermovement jump trial where the athlete stepped off the force plate during the initial quiet weighing phase.

True Biological or Tactical Outliers (Retain in the Dataset!): Values that are statistically extreme ( ∣ Z ∣ > 3.0 |Z| > 3.0 ∣ Z ∣ > 3.0 or beyond Tukey's upper fence) but physiologically authentic and verified by video or session logs. Examples include a box-to-box midfielder recording 1,450 m 1{,}450\text{ m} 1 , 450 m of high-speed running ( + 3.2 SD +3.2\text{ SD} + 3.2 SD above team mean) because their team played 30 minutes 30\text{ minutes} 30 minutes of extra time shorthanded after a red card, or an elite sprinter hitting a personal-best 10.4 m ⋅ s − 1 10.4\text{ m}\cdot\text{s}^{-1} 10.4 m ⋅ s − 1 peak velocity with clean satellite lock ( HDOP = 0.9 \text{HDOP} = 0.9 HDOP = 0.9 , 14 14 14 satellites). Automatically deleting or winsorizing true biological outliers blinds the monitoring system to the exact acute workload spikes and peak neuromuscular outputs that drive injury risk and performance adaptation.

Missing Data Taxonomy ( MCAR \text{MCAR} MCAR , MAR \text{MAR} MAR , MNAR \text{MNAR} MNAR ) and Workload Imputation

When data points are genuinely missing or deleted as corrupted artifacts, the sport scientist must classify the mechanism of missingness under Rubin's Taxonomy :

Missing Completely at Random ( MCAR \text{MCAR} MCAR ): The probability that an observation is missing is completely independent of both observed and unobserved variables (e.g., a GNSS \text{GNSS} GNSS unit battery dies mid-practice or a force-plate USB cable is accidentally unplugged).

Missing at Random ( MAR \text{MAR} MAR ): The probability of missingness is systematically related to observed contextual variables, but not to the missing outcome itself (e.g., offensive linemen are missing outdoor GNSS \text{GNSS} GNSS high-speed running values on Tuesday because their position group was scheduled inside the indoor dome for walk-through pass protection).

Missing Not at Random ( MNAR \text{MNAR} MNAR ): The probability of missingness is directly driven by the unobserved value itself (e.g., the most severely fatigued, sleep-deprived, or muscle-sore athletes intentionally skip the morning subjective wellness survey or countermovement jump screening because they fear a poor score will cause the coaching staff to bench them).

Imputation Strategies for Longitudinal Workload Continuity

In longitudinal workload monitoring (such as 7-day acute loads, 28-day chronic loads, or Exponentially Weighted Moving Averages [ EWMA \text{EWMA} EWMA ]), leaving a missed training session blank or treating it as 0 creates a phantom recovery day that artificially inflates the subsequent week's acute-to-chronic workload ratio ( ACWR \text{ACWR} ACWR ). Practitioners should apply a structured imputation hierarchy while always setting a boolean metadata flag ( is_imputed = TRUE and imputation_method ):

Individualized Session-RPE Regression Imputation: If an athlete's GNSS \text{GNSS} GNSS unit failed during a 90 -minute 90\text{-minute} 90 -minute practice where they reported an RPE \text{RPE} RPE of 7 7 7 ( sRPE = 7 × 90 = 630 AU \text{sRPE} = 7 \times 90 = 630\text{ AU} sRPE = 7 × 90 = 630 AU ), predict their missing total distance or PlayerLoad using that specific athlete's historical linear regression equation relating their sRPE \text{sRPE} sRPE to external load.

Drill-Level Position-Group Scaling: Sum the average distance covered by the athlete's specific position group across the exact drills the athlete participated in, multiplied by that athlete's historical work-rate ratio relative to their position peers (e.g., 1.06 × Positional Drill Mean 1.06 \times \text{Positional Drill Mean} 1.06 × Positional Drill Mean ).

Multiple Imputation ( MICE \text{MICE} MICE ) or Kalman State-Space Interpolation: Used for multivariate research modeling, whereas Last Observation Carried Forward ( LOCF \text{LOCF} LOCF ) should be avoided across variable training days because repeating yesterday's hard session load onto today's recovery day distorts microcycle periodicity.

Worked Example: Tukey IQR \text{IQR} IQR Outlier Audit & Drill-Scaled GNSS \text{GNSS} GNSS Imputation

Suppose a 20-player soccer squad completes a high-intensity tactical session. The quartiles for High-Speed Running ( HSR > 5.5 m ⋅ s − 1 \text{HSR} > 5.5\text{ m}\cdot\text{s}^{-1} HSR > 5.5 m ⋅ s − 1 ) across the squad are:

Q 1 ( 25 th percentile ) = 420 m Q_1 (25\text{th percentile}) = 420\text{ m} Q 1 ​ ( 25 th percentile ) = 420 m

Q 2 ( Median ) = 530 m Q_2 (\text{Median}) = 530\text{ m} Q 2 ​ ( Median ) = 530 m

Q 3 ( 75 th percentile ) = 620 m Q_3 (75\text{th percentile}) = 620\text{ m} Q 3 ​ ( 75 th percentile ) = 620 m

Step 1 — Calculate Tukey's IQR \text{IQR} IQR and Upper Fence:

IQR = Q 3 − Q 1 = 620 − 420 = 200 m \text{IQR} = Q_3 - Q_1 = 620 - 420 = 200\text{ m} IQR = Q 3 ​ − Q 1 ​ = 620 − 420 = 200 m

Upper Fence (1.5 IQR) = Q 3 + 1.5 ( 200 ) = 620 + 300 = 920 m \text{Upper Fence (1.5 IQR)} = Q_3 + 1.5(200) = 620 + 300 = 920\text{ m} Upper Fence (1.5 IQR) = Q 3 ​ + 1.5 ( 200 ) = 620 + 300 = 920 m

Extreme Upper Fence (3.0 IQR) = Q 3 + 3.0 ( 200 ) = 620 + 600 = 1,220 m \text{Extreme Upper Fence (3.0 IQR)} = Q_3 + 3.0(200) = 620 + 600 = 1{,}220\text{ m} Extreme Upper Fence (3.0 IQR) = Q 3 ​ + 3.0 ( 200 ) = 620 + 600 = 1 , 220 m

Step 2 — Audit Two Flagged Outliers ( > 920 m > 920\text{ m} > 920 m ):

Athlete A ( HSR = 1,010 m \text{HSR} = 1{,}010\text{ m} HSR = 1 , 010 m , Peak Speed = 9.4 m ⋅ s − 1 = 9.4\text{ m}\cdot\text{s}^{-1} = 9.4 m ⋅ s − 1 , HDOP = 0.8 \text{HDOP} = 0.8 HDOP = 0.8 , Satellites = 15 = 15 = 15 ): Video confirms Athlete A (a wingback) performed extra repeated overlap recovery runs during an 11v10 tactical overload drill. Decision: Authentic biological/tactical outlier → \rightarrow → Retain 1,010 m 1{,}010\text{ m} 1 , 010 m .

Athlete B ( HSR = 1,380 m \text{HSR} = 1{,}380\text{ m} HSR = 1 , 380 m , Peak Speed = 14.6 m ⋅ s − 1 = 14.6\text{ m}\cdot\text{s}^{-1} = 14.6 m ⋅ s − 1 , HDOP = 4.8 \text{HDOP} = 4.8 HDOP = 4.8 , Satellites = 4 = 4 = 4 ): Sensor diagnostics reveal multipath velocity spikes during a 15-minute period near the stadium bleachers. Decision: Technical artifact → \rightarrow → Delete corrupted segment and impute .

Step 3 — Impute Athlete B's Missing Segment via Individualized Position Scaling:

During the corrupted 15-minute drill, the other three central midfielders averaged 180 m 180\text{ m} 180 m of HSR \text{HSR} HSR . Across the preceding 20 sessions, Athlete B's historical work-rate ratio in that drill category is 1.05 × 1.05\times 1.05 × the central midfield mean. Adding the imputed drill value ( 180 × 1.05 = 189 m 180 \times 1.05 = 189\text{ m} 180 × 1.05 = 189 m ) to Athlete B's clean pre-failure HSR \text{HSR} HSR ( 390 m 390\text{ m} 390 m ) yields a corrected session total of 579 m 579\text{ m} 579 m , stored with is_imputed = TRUE .