GELE Adjustment Computations (Least Squares) — Theory of Errors, Weights and Most Probable ValueMisconception Buster
If you have been missing Theory of Errors, Weights and Most Probable Value questions on your GELE mocks, the cause is almost always a misconception. This page lists the ones Professional Regulation Commission (PRC) — Board of Geodetic Engineering exploits most often in the GELE Adjustment Computations (Least Squares) subtest and shows how to correct them before exam day.
Exam context
On the GELE 2026, the Adjustment Computations (Least Squares) subtest carries a "Core" weight in Professional Regulation Commission (PRC) — Board of Geodetic Engineering's pattern. Theory of Errors, Weights and Most Probable Value lands at position 1st out of 5 in the standard review order. Target score is 70% weighted average, no sub-test below 50%, and roughly a meaningful share of items come from Adjustment Computations (Least Squares) on a typical GELE paper.
Theory of Errors, Weights and Most Probable Value - Misconception Buster
In the PRC Geodetic Engineer Licensure Examination, the Theory of Errors and Weights chapter is a high-yield topic in Adjustment Computations. Examiners deliberately craft questions that exploit predictable student errors — wrong formula substitution, inverted weight relationships, and confusion between standard deviation types. Many examinees lose critical marks not because they lack knowledge, but because they hold subtly wrong beliefs that feel correct until a carefully designed trap question exposes them. This guide identifies the 10 most dangerous misconceptions, shows exactly why students fall for them, and arms you with the correct understanding so that no board-exam trap question can catch you off guard. Study each misconception carefully — the trap questions mirror actual board-exam item formats.
Summary
The following are the most important takeaways to avoid losing marks in the PRC board examination on Theory of Errors, Weights and Most Probable Value: (1) ALWAYS use w = 1/σ² — never w = 1/σ. The variance, not the standard deviation, is in the denominator. (2) For leveling, w ∝ 1/K — the SHORTEST route has the HIGHEST weight. (3) The arithmetic mean is the MPV only for EQUALLY weighted observations. When precisions differ, use the weighted mean. (4) Standard deviation of the mean is σ/√n — precision improves with the SQUARE ROOT of n, not n itself. To double precision, QUADRUPLE the observations. (5) Probable error PE = 0.6745σ — memorize 0.6745, not 0.5 and not 0.7071. (6) Always use n−1 (not n) in the denominator for the sample standard deviation in surveying adjustment. (7) Systematic errors CANNOT be eliminated by averaging — they require correction by calibration or proper procedure. (8) Blunders must be DETECTED and REMOVED before adjustment, not assigned a low weight and included. (9) Residuals always satisfy Σvᵢ = 0 — use this as a mandatory arithmetic check. (10) Residuals (vᵢ = xᵢ − x̄) are NOT the same as true errors (εᵢ = xᵢ − μ); we work with residuals because the true value is unknown. Mastering these ten points will prevent the most common and costly errors on the licensure examination.
Misconceptions
Weight is inversely proportional to the standard deviation (w = 1/σ), NOT to the variance.
Tags
- critical_formula_error
- weight_definition
- common_error
Topic
Weights and Weighted Mean
Severity
critical
Exam Impact
Using w = 1/σ instead of w = 1/σ² changes all weight values and therefore gives a wrong weighted mean. In a four-choice board question, this error often leads to a specific distractor that the examiners placed intentionally.
The Reality
Weight is defined as w = 1/σ². This comes directly from the normal distribution and the principle of maximum likelihood. Variance σ² is the second central moment that quantifies spread in the probability density function. Using 1/σ instead of 1/σ² will give completely different numerical weights and a wrong weighted mean. For example, if σ₁ = 0.02 m and σ₂ = 0.04 m, the correct weight ratio is w₁:w₂ = (1/0.0004):(1/0.0016) = 4:1, NOT (1/0.02):(1/0.04) = 2:1.
Trap Question
Question
A distance is measured twice: x₁ = 150.30 m with σ₁ = 0.04 m, and x₂ = 150.35 m with σ₂ = 0.02 m. What is the weight ratio w₁:w₂?
Explanation
Weight must be computed using variance (σ²), not standard deviation (σ). Since σ₂ = σ₁/2, the variance ratio is σ₁²/σ₂² = (0.04)²/(0.02)² = 4, so w₁:w₂ = 1:4. The second observation, being twice as precise in standard deviation terms, is four times more heavily weighted.
Wrong Answer
w₁:w₂ = 1:2 (using w = 1/σ → 1/0.04 : 1/0.02 = 25:50 = 1:2)
Correct Answer
w₁:w₂ = 1:4 (using w = 1/σ² → 1/0.0016 : 1/0.0004 = 625:2500 = 1:4)
Misconception Id
M1
Correct Vs Incorrect
Correct Approach
Given σ₁ = 0.02 m, σ₂ = 0.01 m → w₁ = 1/(0.02)² = 1/0.0004 = 2500, w₂ = 1/(0.01)² = 1/0.0001 = 10000 → w₁:w₂ = 1:4. CORRECT.
Incorrect Approach
Given σ₁ = 0.02 m, σ₂ = 0.01 m → w₁ = 1/0.02 = 50, w₂ = 1/0.01 = 100 → w₁:w₂ = 1:2. WRONG.
Why Students Believe It
Students see that a higher standard deviation means less reliable measurement, so they intuitively write w ∝ 1/σ. The logic 'less precise → lower weight' is correct, but the mathematical relationship used is wrong. Many review books list the rule qualitatively without emphasizing that variance — not standard deviation — is the denominator.
The standard deviation of the mean equals the standard deviation of a single observation (σ_mean = σ).
Tags
- formula_confusion
- standard_deviation
- common_error
Topic
Most Probable Value and Precision of the Mean
Severity
critical
Exam Impact
This misconception causes incorrect answers in questions asking for the precision or uncertainty of an averaged quantity. It also leads to errors in determining how many observations are needed to achieve a target precision.
The Reality
The standard deviation of the mean is σ_mean = σ/√n. This is derived from the propagation of random errors through the averaging operation. Taking the mean of n independent observations reduces the standard deviation by a factor of √n, not n. For example, with σ = 0.030 m and n = 9 observations, σ_mean = 0.030/√9 = 0.030/3 = 0.010 m — three times smaller, not nine times smaller.
Trap Question
Question
A surveyor measures a horizontal angle 16 times and obtains a standard deviation of a single reading of σ = 8". What is the standard deviation of the mean angle?
Explanation
The standard deviation of the mean is σ/√n = 8"/√16 = 8"/4 = 2". Averaging 16 independent observations improves precision by a factor of √16 = 4, not 16. If the answer were 0.5" (dividing by 16), that would also be wrong — the improvement factor is √n, not n.
Wrong Answer
8" (same as single observation standard deviation)
Correct Answer
2" (σ_mean = 8"/√16 = 8"/4 = 2")
Misconception Id
M2
Correct Vs Incorrect
Correct Approach
σ_mean = σ/√n = 0.027/√9 = 0.027/3 = 0.009 m. The mean of 9 observations is three times more precise than a single observation.
Incorrect Approach
A distance measured 9 times has σ = 0.027 m. Precision of the mean = 0.027 m. WRONG — this is only the single-observation precision.
Why Students Believe It
Students learn that averaging reduces error and know that σ_mean is smaller than σ, but under exam pressure they forget the √n factor and simply use σ directly as the precision of the mean. This is especially common when the question asks for the 'precision of the average' rather than explicitly saying 'standard deviation of the mean.'
For differential leveling, weight is proportional to the distance traveled (w ∝ K), so a longer route has higher weight.
Tags
- inverse_relationship
- leveling
- critical_conceptual_error
Topic
Weights — Differential Leveling
Severity
critical
Exam Impact
This misconception reverses all weight assignments in leveling problems. The weighted mean elevation will be pulled toward the wrong observation, and all subsequent calculations will be incorrect. This is one of the most commonly exploited errors in board-exam leveling weight problems.
The Reality
For differential leveling, weight is inversely proportional to the length of the level route: w ∝ 1/K. A longer route accumulates more random errors (rod reading errors, instrument settlement, refraction effects) and is therefore less reliable. The shortest route between two benchmarks is the most precise and carries the highest weight. This is directly analogous to w = 1/σ² because σ² ∝ K for leveling.
Trap Question
Question
Three level routes connect the same two benchmarks with route lengths of 3 km, 5 km, and 1 km, yielding elevations of 45.123 m, 45.131 m, and 45.119 m respectively. Which observation carries the highest weight in computing the weighted mean elevation?
Explanation
In differential leveling, random errors accumulate along the route. A 5 km route accumulates roughly 5 times more random error variance than a 1 km route. Therefore w ∝ 1/K: w₁ = 1/3, w₂ = 1/5, w₃ = 1/1. The 1 km route has weight 1.0 — the highest of the three — and pulls the weighted mean closest to 45.119 m.
Wrong Answer
The 5 km route, because it is the longest and covered the most ground.
Correct Answer
The 1 km route, because weight ∝ 1/K and the shortest route has the least accumulated error.
Misconception Id
M3
Correct Vs Incorrect
Correct Approach
Level routes of 2, 4, 8 km → w₁ = 1/2, w₂ = 1/4, w₃ = 1/8. Multiply by LCM (8) → w₁:w₂:w₃ = 4:2:1. CORRECT — shorter route has higher weight.
Incorrect Approach
Level routes of 2, 4, 8 km → w₁:w₂:w₃ = 2:4:8 = 1:2:4 (proportional to distance). WRONG — this gives the longest route the highest weight.
Why Students Believe It
Students confuse 'more work done = more reliable' thinking, or they reason that a longer route covers more ground and thus provides more information. The word 'proportional' without the inverse is easy to misremember under stress.
The arithmetic mean is always the Most Probable Value (MPV), regardless of observation quality.
Tags
- conceptual_gap
- arithmetic_mean
- weighted_mean
Topic
Most Probable Value
Severity
critical
Exam Impact
In problems that provide standard deviations or route lengths alongside measurements, students who apply the simple arithmetic mean will get the wrong MPV. Board questions frequently test whether examinees know when to switch from arithmetic mean to weighted mean.
The Reality
The arithmetic mean x̄ = Σx/n is the MPV only when all observations are of equal reliability (equal weights). When observations have different precisions, different numbers of repeats, or are obtained from level routes of different lengths, the MPV is the weighted mean x̄_w = Σ(wᵢxᵢ)/Σwᵢ. Using the arithmetic mean for unequally weighted observations violates the principle of least squares.
Trap Question
Question
Two surveyors independently measure the same distance. Surveyor A measures 200.14 m with σ = 0.06 m. Surveyor B measures 200.08 m with σ = 0.02 m. What is the Most Probable Value of the distance?
Explanation
Since σ_A = 3 × σ_B, the weights are w_A = 1/(0.06)² = 277.78 and w_B = 1/(0.02)² = 2500. The weight ratio is approximately 1:9. x̄_w = (277.78×200.14 + 2500×200.08)/(277.78+2500) = (55,594.9 + 500,200)/2777.78 ≈ 200.093 m. Surveyor B's result dominates because their measurement is nine times more heavily weighted.
Wrong Answer
200.11 m (arithmetic mean: (200.14 + 200.08)/2 = 200.11 m)
Correct Answer
200.095 m (weighted mean using w = 1/σ²)
Misconception Id
M4
Correct Vs Incorrect
Correct Approach
w₁=1/0.0016=625, w₂=1/0.0004=2500, w₃=1/0.0001=10000. x̄_w = (625×100.02 + 2500×100.06 + 10000×100.04)/(625+2500+10000) = 100.043 m. CORRECT.
Incorrect Approach
Three measurements: 100.02 m (σ=0.04 m), 100.06 m (σ=0.02 m), 100.04 m (σ=0.01 m). MPV = (100.02 + 100.06 + 100.04)/3 = 100.040 m. WRONG — ignores unequal reliability.
Why Students Believe It
Students learn 'MPV = arithmetic mean' as a fact and apply it universally without remembering the key condition: this is only valid when all observations have equal reliability (equal weights). The simplified formula is taught first and often never properly qualified.
Using n instead of n−1 in the denominator of the sample standard deviation formula gives the correct result.
Tags
- degrees_of_freedom
- formula_confusion
- sample_vs_population
Topic
Standard Deviation and Residuals
Severity
major
Exam Impact
Using n instead of n−1 gives a numerically smaller (and wrong) standard deviation. In small samples (n=4, 5), the difference is significant enough to choose the wrong multiple-choice option.
The Reality
In surveying, the mean x̄ is computed from the same dataset and is unknown beforehand. This consumes one degree of freedom, leaving n−1. The formula σ = √(Σv²/(n−1)) is the unbiased estimator of the population standard deviation. Using n underestimates the true variability. The distinction matters for board questions that ask for the 'standard deviation of a single observation' from a set of measurements.
Trap Question
Question
Six repeated measurements of a distance yield residuals of +5, −3, +2, −6, +4, −2 mm. What is the standard deviation of a single measurement?
Explanation
Always use n−1 in the denominator for the sample standard deviation in surveying adjustment. With n=6 measurements, there are n−1=5 degrees of freedom because one parameter (the mean) was estimated from the data. Σv² = 25+9+4+36+16+4 = 94 mm². σ = √(94/5) = √18.8 = 4.34 mm.
Wrong Answer
3.26 mm (using Σv²/n = 90/6 = 15, √15 = 3.87 mm — or similar error with n in denominator)
Correct Answer
4.24 mm (Σv² = 25+9+4+36+16+4 = 94 mm², σ = √(94/5) = √18.8 = 4.34 mm)
Misconception Id
M5
Correct Vs Incorrect
Correct Approach
σ = √(54/(4−1)) = √(54/3) = √18 = 4.24 mm. CORRECT — uses n−1 = 3 degrees of freedom.
Incorrect Approach
4 residuals: v = +3, −2, +4, −5 mm. Σv² = 9+4+16+25 = 54 mm². σ = √(54/4) = √13.5 = 3.67 mm. WRONG — uses n.
Why Students Believe It
Students encounter two forms: σ = √(Σv²/n) for a population and s = √(Σv²/(n−1)) for a sample. In survey adjustment, the mean is estimated from the data itself, so n−1 (degrees of freedom) is correct. But some textbooks use σ loosely for both, and students default to n because it seems simpler.
Systematic errors are random in nature and can be reduced by averaging multiple observations.
Tags
- error_classification
- systematic_vs_random
- conceptual_gap
Topic
Types of Errors
Severity
major
Exam Impact
Board questions test whether students know which error type is addressed by which method. Stating that averaging corrects systematic error is a definitional error that reveals a fundamental misunderstanding of adjustment theory.
The Reality
Systematic errors are constant or follow a deterministic pattern — they do NOT cancel out with averaging. In fact, averaging more observations only reinforces the systematic error because every measurement contains the same bias in the same direction. Examples: an EDM with an uncorrected zero error, a level with a non-horizontal line of sight, or a steel tape that is shorter than its nominal length. Systematic errors must be identified and corrected through calibration, field procedures (e.g., double-centering, reciprocal leveling), or mathematical models — not by averaging.
Trap Question
Question
A surveyor levels a section 20 times with an instrument having a small collimation error but uses balanced backsight and foresight distances throughout. Is the collimation error systematic or random, and does averaging the 20 level runs eliminate it?
Explanation
A collimation error is systematic — it affects every measurement the same way. However, when backsight and foresight distances are balanced, the collimation error cancels geometrically within each setup, not by averaging. If sights were unbalanced, averaging the 20 runs would not help because the bias would accumulate in every run in the same direction.
Wrong Answer
It is systematic, and averaging 20 runs will reduce its effect on the final elevation difference.
Correct Answer
It is systematic, but balanced sight distances — not averaging — eliminate it. Averaging will NOT remove a systematic error.
Misconception Id
M6
Correct Vs Incorrect
Correct Approach
Apply a tape correction to each measurement before averaging. The systematic error is removed by a deterministic correction, and only residual random errors are then reduced by averaging.
Incorrect Approach
A steel tape is 0.005 m too short. A surveyor measures a distance 10 times and averages the results, expecting the short-tape error to cancel. WRONG — every measurement is biased by the same amount; the average is equally biased.
Why Students Believe It
Students know that averaging reduces errors, and they generalize this to all error types. If you measure the same thing 100 times, surely the average must be correct — right? This confusion stems from not clearly distinguishing the statistical nature of each error type.
A blunder (mistake) is just a large random error and should be included in the adjustment with reduced weight.
Tags
- blunder_vs_random_error
- error_classification
- adjustment_sequence
Topic
Types of Errors — Blunders
Severity
major
Exam Impact
Board questions on error classification and the sequence of adjustment steps will penalize students who do not know the correct order: eliminate blunders → correct systematic errors → adjust random errors.
The Reality
Blunders are not random errors at all — they are mistakes arising from human error (transposing digits, misreading a rod, recording a wrong value). They do not follow the normal distribution and have no place in the adjustment model. Standard procedure: identify and eliminate blunders using statistical tests (e.g., tau test, 3σ rejection criterion) BEFORE performing any least squares adjustment. Assigning a low weight to a blunder does not correct the mathematics — it corrupts the adjustment with an outlier.
Trap Question
Question
During processing of 6 repeated EDM distance measurements, one value is 3.5 standard deviations from the mean. The surveyor reduces its weight to 0.1 and includes it in the weighted mean. Is this acceptable practice?
Explanation
The adjustment model assumes all included observations are governed by the same normal distribution of random errors. An outlier at 3.5σ almost certainly represents a blunder rather than a random fluctuation (probability of occurrence by chance < 0.05%). Including it even with a low weight violates the assumptions of least squares. The correct action is to apply a blunder detection test (tau criterion), reject the outlier, and re-compute the mean from the remaining observations.
Wrong Answer
Yes, giving it a low weight minimizes its influence and the weighted mean will still be valid.
Correct Answer
No. A value beyond the 3σ threshold should be flagged as a blunder and removed before adjustment, not weighted down and included.
Misconception Id
M7
Correct Vs Incorrect
Correct Approach
Detect 54.22 m as a blunder using the 3σ test or tau test, remove it from the dataset, then compute the mean of the four valid observations: (45.22+45.19+45.23+45.21)/4 = 45.2125 m.
Incorrect Approach
Five measurements: 45.22, 45.19, 45.23, 45.21, 54.22 m (last value is clearly a blunder — digits transposed). Include all five in a weighted mean with the outlier given weight 0.1. WRONG — contaminates the adjustment.
Why Students Believe It
Students see that some errors are small and some are large, and they think of this as a continuous spectrum. A large residual 'must be a big random error' and they assign it a low weight to reduce its influence rather than rejecting it outright.
Probable error (PE) equals 0.5σ, so PE = σ/2.
Tags
- probable_error
- formula_confusion
- numerical_factor
Topic
Error Measures — Probable Error
Severity
major
Exam Impact
Any board question asking to compute the probable error from σ (or vice versa) will yield the wrong answer if 0.5 is used instead of 0.6745. The factor 0.6745 must be memorized precisely.
The Reality
The probable error is PE = 0.6745σ, not 0.5σ. The probable error is defined as the value r such that exactly 50% of random errors lie within ±r of the true value. For a normal distribution, this corresponds to the 25th-to-75th percentile range, which gives the factor 0.6745 — not 0.5. Using 0.5σ underestimates the probable error and will give wrong numerical answers.
Trap Question
Question
The standard deviation of a surveyed distance is 0.030 m. What is the probable error of a single measurement?
Explanation
The probable error is PE = 0.6745σ. This factor (0.6745) comes from the inverse normal distribution at the 75th percentile. It means that 50% of all random errors fall within ±PE of zero. For σ = 0.030 m: PE = 0.6745 × 0.030 = 0.02024 m. This is approximately two-thirds of σ, not one-half. Memorize: PE ≈ 0.6745σ or equivalently σ ≈ 1.4826 × PE.
Wrong Answer
0.015 m (using PE = 0.5σ)
Correct Answer
0.020 m (PE = 0.6745 × 0.030 = 0.02024 m ≈ 0.020 m)
Misconception Id
M8
Correct Vs Incorrect
Correct Approach
PE = 0.6745 × 0.020 = 0.01349 m ≈ 0.013 m. CORRECT — uses the proper normal distribution factor.
Incorrect Approach
σ = 0.020 m → PE = 0.5 × 0.020 = 0.010 m. WRONG — uses incorrect factor.
Why Students Believe It
Students vaguely remember that probable error is 'about half' of the standard deviation from seeing that 0.6745 ≈ 0.7 ≈ roughly half, and the simplified mnemonic '50-50 chance' is misremembered as 'half of sigma.' Some confuse the 50th percentile of the error distribution with the midpoint between 0 and σ.
Increasing the number of observations n always improves precision proportionally — doubling n doubles precision.
Tags
- square_root_law
- precision
- common_error
Topic
Precision Improvement with Repeated Observations
Severity
major
Exam Impact
Questions asking 'how many observations are needed to improve precision from X to Y' require solving n = (σ/σ_mean)². Students who use a linear model will get a dramatically different (and wrong) answer.
The Reality
Precision improves as √n, not n. To halve the standard deviation of the mean (double precision), you must quadruple the number of observations: σ_mean = σ/√n. Doubling n only reduces σ_mean by a factor of √2 ≈ 1.414, not 2. This has profound practical implications: beyond a certain point, the cost of additional measurements far outweighs the precision gained.
Trap Question
Question
A leveling section run once achieves a standard deviation of the mean elevation of 12 mm. How many times must the section be run to reduce this to 4 mm?
Explanation
Setting up: σ_mean₂ = σ_mean₁/√n → 4 = 12/√n → √n = 3 → n = 9. You need 9 runs, not 3. This is the square-root law of precision improvement. Tripling the precision (from 12 to 4 mm) requires nine times the number of observations. This is why high-precision surveying uses careful instrument selection and procedures rather than simply repeating mediocre measurements many times.
Wrong Answer
3 times (because 12/4 = 3, so 3 times as many runs)
Correct Answer
9 times (because σ_mean = σ/√n → n = (12/4)² = 9)
Misconception Id
M9
Correct Vs Incorrect
Correct Approach
σ_mean = σ/√n → 0.005 = 0.020/√n → √n = 4 → n = 16 observations. CORRECT — need 4 times as many observations to double precision.
Incorrect Approach
Current precision with n=4 is σ_mean = 0.010 m. To achieve 0.005 m (double precision), need n = 4 × 2 = 8 observations. WRONG — linear thinking.
Why Students Believe It
Students think 'more data = twice as good' because they associate improvement with a simple multiplicative relationship. The qualitative idea that more observations is better is correct, but the quantitative relationship is misunderstood as linear rather than square-root.
Residuals vᵢ = xᵢ − x̄ and errors εᵢ = xᵢ − μ are the same thing.
Tags
- conceptual_gap
- residuals
- true_errors
- degrees_of_freedom
Topic
Residuals vs. Errors
Severity
minor
Exam Impact
While this is less commonly tested numerically, board questions on the theory of least squares and degrees of freedom hinge on understanding that we minimize Σv² (sum of squared residuals), not Σε². Misidentifying these causes errors in conceptual questions.
The Reality
Errors εᵢ = xᵢ − μ are the deviations from the TRUE (population) mean μ, which is unknown in practice. Residuals vᵢ = xᵢ − x̄ are deviations from the ESTIMATED mean x̄ (the MPV computed from the sample). Because x̄ is estimated from the data, residuals are slightly smaller than true errors on average — this is why we use n−1 instead of n in the denominator of the standard deviation formula. In adjustment computations, we ALWAYS work with residuals (corrections), never with true errors.
Trap Question
Question
After computing the arithmetic mean x̄ from 5 measurements, a student finds the residuals v₁=+2, v₂=−3, v₃=+1, v₄=+4, v₅=−4 mm. The student says 'Σv = 0, so these are the true errors.' Is this statement correct?
Explanation
By the algebraic property of the arithmetic mean, Σ(xᵢ − x̄) = Σxᵢ − nx̄ = 0 always, regardless of the true value μ. True errors εᵢ = xᵢ − μ sum to a random value. Residuals are our computable estimates of true errors. The distinction matters: it is why σ uses n−1 (one degree of freedom is lost estimating x̄) and why we say we are minimizing Σv² in least squares, not Σε².
Wrong Answer
Yes, because true errors also sum to zero.
Correct Answer
No. These are residuals (deviations from the estimated mean x̄), not true errors (deviations from the unknown true value μ). The zero sum is a consequence of using x̄ as the reference, not proof that they equal true errors.
Misconception Id
M10
Correct Vs Incorrect
Correct Approach
Residuals vᵢ = xᵢ − x̄ always satisfy Σvᵢ = 0 (by the property of the arithmetic mean). True errors εᵢ = xᵢ − μ sum to a random quantity with expected value zero. They are related but not identical.
Incorrect Approach
A student states: 'The sum of residuals equals zero, so the residuals are the true errors.' WRONG — true errors sum to zero only in expectation; residuals computed from the sample mean always sum to exactly zero by construction.
Why Students Believe It
Both residuals and errors measure how far an observation is from a reference value. Students use 'error' and 'residual' interchangeably in casual speech. In many introductory texts, the distinction is glossed over, so students carry this confusion into advanced topics.
In a weighted mean problem, if one observation has zero weight (w=0), it should be deleted from the dataset entirely.
Tags
- zero_weight
- weighted_mean
- formula_application
Topic
Weights and Weighted Mean
Severity
minor
Exam Impact
Board questions may test the mathematical consequence of w=0 and whether students know the weighted mean formula still holds (the term simply vanishes). The conceptual question of when w=0 is valid is more likely to appear in essay-type items.
The Reality
An observation with w=0 truly contributes nothing to the weighted mean — mathematically it is equivalent to having no observation. However, the act of assigning w=0 must be justified: it means infinite variance, i.e., the observation carries absolutely no information. In practice, even a poor observation has SOME information and should be given a small but nonzero weight based on its actual σ. Setting w=0 arbitrarily because you distrust an observation (without statistical justification) is not adjustment — it is selective data rejection.
Trap Question
Question
Three distances are measured with weights w₁=3, w₂=0, w₃=1 and values x₁=100.10 m, x₂=100.50 m, x₃=100.20 m. What is the weighted mean?
Explanation
The weighted mean formula naturally handles zero weights. When w₂=0, the term 0×100.50=0 contributes nothing to the numerator, and 0 adds nothing to the denominator. The result is determined entirely by x₁ and x₃ with their respective weights 3 and 1. No special handling or deletion is needed — the formula is self-consistent.
Wrong Answer
(3×100.10 + 0×100.50 + 1×100.20)/(3+0+1) = (300.30 + 0 + 100.20)/4 = 400.50/4 = 100.125 m — but the student then worries about x₂ being ignored and manually includes it somehow.
Correct Answer
100.125 m. The formula handles w=0 correctly: x̄_w = Σ(wᵢxᵢ)/Σwᵢ = (3×100.10 + 0×100.50 + 1×100.20)/(3+0+1) = 400.50/4 = 100.125 m. x₂ contributes nothing.
Misconception Id
M11
Correct Vs Incorrect
Correct Approach
All observations with nonzero variance receive their proper weight w=1/σ². An observation is excluded (effectively w=0) only after a formal statistical outlier test confirms it is a blunder.
Incorrect Approach
A surveyor assigns w=0 to one of three measurements because they 'feel' it is wrong, without any statistical test. This is data manipulation, not adjustment.
Why Students Believe It
Students see that w=0 contributes nothing to the numerator (wᵢxᵢ = 0) and nothing to the denominator (wᵢ = 0 adds nothing), so they reason 'it has no effect, just remove it.' This is correct procedurally but students misunderstand WHY — and sometimes they set w=0 for observations they merely distrust rather than for observations that are physically absent or completely unreliable.
The sum of residuals Σvᵢ can be non-zero after computing the arithmetic mean from a dataset.
Tags
- residual_check
- arithmetic_identity
- quality_control
Topic
Residuals and Arithmetic Check
Severity
minor
Exam Impact
Board questions sometimes include a table of residuals and ask which set is consistent with the arithmetic mean. Knowing Σv = 0 is a mandatory check allows instant identification of the correct answer or detection of a planted error.
The Reality
For unweighted observations, Σvᵢ = Σ(xᵢ − x̄) = 0 ALWAYS and EXACTLY. This is an algebraic identity — it is one of the best arithmetic checks available. If Σv ≠ 0 after computing the arithmetic mean, there is a computational error and the work must be rechecked. For weighted observations, the check is Σ(wᵢvᵢ) = 0.
Trap Question
Question
Four measurements and their residuals are listed: x₁=50.12 (v₁=+0.01), x₂=50.08 (v₂=−0.03), x₃=50.15 (v₃=+0.04), x₄=50.10 (v₄=−0.01). Is this set of residuals consistent with the arithmetic mean?
Explanation
Σvᵢ = +0.01 − 0.03 + 0.04 − 0.01 = +0.01 m ≠ 0. This immediately reveals an inconsistency. The arithmetic mean should be x̄ = (50.12+50.08+50.15+50.10)/4 = 200.45/4 = 50.1125 m. Recomputing: v₁=50.12−50.1125=+0.0075, v₂=−0.0325, v₃=+0.0375, v₄=−0.0125. Σv=0. The original residuals were incorrectly computed, proving Σv=0 is a vital check.
Wrong Answer
Yes, the residuals look reasonable and are all small.
Correct Answer
No. Σv = +0.01 − 0.03 + 0.04 − 0.01 = +0.01 ≠ 0. There is a computational error somewhere.
Misconception Id
M12
Correct Vs Incorrect
Correct Approach
Σvᵢ = 0 is an exact algebraic identity. Any non-zero sum (even 0.001 mm) signals an error. Recheck the computation of x̄ and each vᵢ = xᵢ − x̄.
Incorrect Approach
Student computes residuals and gets Σv = +0.003 m, accepts this as 'close enough due to rounding.' WRONG — this indicates a computational error in either the mean or the residuals.
Why Students Believe It
Students who rush through computations sometimes make arithmetic errors and get Σv ≠ 0, which they accept as a rounding issue. Others simply do not know this is a mandatory check. Some students confuse the weighted mean (where Σwᵢvᵢ = 0, but Σvᵢ ≠ 0 in general) with the unweighted case.
Quick Self Check
Weight is inversely proportional to the VARIANCE: w = 1/σ². Using σ instead of σ² gives a wrong weight ratio. For example, if σ doubles, the variance quadruples and the weight becomes one-fourth, not one-half.
Statement
The weight of an observation is inversely proportional to its standard deviation (w = 1/σ).
Weight for leveling is w ∝ 1/K, where K is the route length. A shorter route accumulates fewer random errors, is more reliable, and therefore receives the highest weight. This is consistent with w = 1/σ² because σ² ∝ K for leveling.
Statement
For differential leveling, the level route with the shortest distance should be assigned the highest weight.
Systematic errors are deterministic and constant — they do not cancel by averaging. Every repeated observation contains the same bias. Averaging 1000 observations with a systematic error still gives a biased result. Systematic errors must be corrected by calibration or proper field procedures.
Statement
Averaging a large number of observations will eliminate a systematic error if enough observations are taken.
σ_mean = σ/√n = σ/√9 = σ/3. With n=9, the square root of 9 is exactly 3, so precision improves by a factor of 3. This is a commonly tested special case of the √n law.
Statement
The standard deviation of the mean from 9 observations is exactly one-third of the standard deviation of a single observation.
The probable error PE = 0.6745σ, not 0.5σ. The factor 0.6745 comes from the 75th percentile of the standard normal distribution and defines the value within which 50% of random errors lie. Using 0.5 underestimates the probable error.
Statement
The probable error of an observation equals 0.5 times its standard deviation.
By definition, Σvᵢ = Σ(xᵢ − x̄) = Σxᵢ − nx̄ = nx̄ − nx̄ = 0. This is an algebraic identity that serves as an essential arithmetic check. A non-zero sum always indicates a computational error.
Statement
After computing the arithmetic mean from any set of equally-weighted observations, the sum of the residuals must always equal exactly zero.
The arithmetic mean is the MPV only when all observations have equal reliability (equal weights). When observations have different precisions (different σ values) or come from leveling routes of different lengths, the MPV is the weighted mean x̄_w = Σ(wᵢxᵢ)/Σwᵢ.
Statement
The arithmetic mean is the Most Probable Value (MPV) for ALL sets of direct observations.
To halve σ_mean, the number of observations must be QUADRUPLED. Since σ_mean = σ/√n, halving σ_mean requires √n to double, which means n must increase by a factor of 4. This is the square-root law: precision improves as √n, not n.
Statement
To halve the standard deviation of the mean, the number of observations must be doubled.
Ready to practise for the GELE 2026?
Super Tutor's AI review plan adapts to your weak areas and builds a weekly practice schedule around your target GELE exam date.