Resources

An environmental engineer evaluates whether installing an electrostatic precipitator reduces particulate emissions from a factory exhaust flue. Let XpreX_{\text{pre}} represent the emission rate (in mg/m3\text{mg/m}^3) measured immediately before installation and XpostX_{\text{post}} represent the emission rate measured after installation on matched production cycles. (a) State the null and alternative hypotheses if the parameter is defined as μdiff=μpre−μpost\mu_{\text{diff}} = \mu_{\text{pre}} - \mu_{\text{post}} (reduction defined as Pre minus Post). (b) State the null and alternative hypotheses if the parameter is defined as μΔ=μpost−μpre\mu_{\Delta} = \mu_{\text{post}} - \mu_{\text{pre}} (change defined as Post minus Pre). (c) Describe the sign trap that occurs if a practitioner confuses these two conventions.

Answer and steps

Final answer

(a) Under μdiff=μpre−μpost\mu_{\text{diff}} = \mu_{\text{pre}} - \mu_{\text{post}}: H0:μdiff=0 mg/m3H_0: \mu_{\text{diff}} = 0\text{ mg/m}^3 versus Ha:μdiff>0 mg/m3H_a: \mu_{\text{diff}} > 0\text{ mg/m}^3. (b) Under μΔ=μpost−μpre\mu_{\Delta} = \mu_{\text{post}} - \mu_{\text{pre}}: H0:μΔ=0 mg/m3H_0: \mu_{\Delta} = 0\text{ mg/m}^3 versus Ha:μΔ<0 mg/m3H_a: \mu_{\Delta} < 0\text{ mg/m}^3. (c) The verbal phrase 'particulate emissions decreased' can imply either a positive parameter value (>0> 0) or a negative parameter value (<0< 0) depending entirely on subtraction order; mismatching the subtraction order with the inequality direction causes the analyst to evaluate the opposite tail of the sampling distribution, turning a highly significant result into an erroneous PP-value of 1−P1 - P.

Steps

Understanding the Research Goal

The engineer wants to determine whether the electrostatic precipitator reduces particulate emissions. In terms of population means, an effective reduction means: μpost<μpreor equivalentlyμpre>μpost\mu_{\text{post}} < \mu_{\text{pre}} \quad \text{or equivalently} \quad \mu_{\text{pre}} > \mu_{\text{post}}

Under the null scenario of no effect, there is no difference between the pre- and post-installation emissions: μpre=μpost\mu_{\text{pre}} = \mu_{\text{post}}


Part (a): Reduction Defined as Pre Minus Post

  1. Parameter definition: μdiff=μpre−μpost\mu_{\text{diff}} = \mu_{\text{pre}} - \mu_{\text{post}}
  2. Null Hypothesis (H0H_0): The null hypothesis states that installing the precipitator has no effect on emission levels: H0:μdiff=0 mg/m3H_0: \mu_{\text{diff}} = 0\text{ mg/m}^3
  3. Alternative Hypothesis (HaH_a): If emissions are reduced, the pre-installation level is higher than the post-installation level (μpre>μpost\mu_{\text{pre}} > \mu_{\text{post}}). Subtracting μpost\mu_{\text{post}} from both sides yields: μpre−μpost>0  ⟹  μdiff>0 mg/m3\mu_{\text{pre}} - \mu_{\text{post}} > 0 \implies \mu_{\text{diff}} > 0\text{ mg/m}^3 Therefore: Ha:μdiff>0 mg/m3H_a: \mu_{\text{diff}} > 0\text{ mg/m}^3

Part (b): Change Defined as Post Minus Pre

  1. Parameter definition: μΔ=μpost−μpre\mu_{\Delta} = \mu_{\text{post}} - \mu_{\text{pre}}
  2. Null Hypothesis (H0H_0): The null hypothesis remains that there is zero difference between before and after: H0:μΔ=0 mg/m3H_0: \mu_{\Delta} = 0\text{ mg/m}^3
  3. Alternative Hypothesis (HaH_a): A reduction in emissions means the post-installation level is lower than the pre-installation level (μpost<μpre\mu_{\text{post}} < \mu_{\text{pre}}). Subtracting μpre\mu_{\text{pre}} from both sides yields: μpost−μpre<0  ⟹  μΔ<0 mg/m3\mu_{\text{post}} - \mu_{\text{pre}} < 0 \implies \mu_{\Delta} < 0\text{ mg/m}^3 Therefore: Ha:μΔ<0 mg/m3H_a: \mu_{\Delta} < 0\text{ mg/m}^3

Part (c): The Sign Trap

The phrase "emissions decreased" or "a reduction occurred" intuitively suggests a downward change, but mathematically:

  • Setting the parameter as (Pre−Post)(\text{Pre} - \text{Post}) translates a decrease into a positive difference (>0> 0), requiring a right-tailed test.
  • Setting the parameter as (Post−Pre)(\text{Post} - \text{Pre}) translates a decrease into a negative difference (<0< 0), requiring a left-tailed test.

The Consequence of the Pitfall:
If a researcher calculates sample differences as Post−Pre\text{Post} - \text{Pre} (yielding negative values for successful reductions) but accidentally pairs them with the alternative hypothesis Ha:μ>0H_a: \mu > 0, the test looks in the opposite tail of the sampling distribution. This causes an overwhelmingly successful intervention (which should yield a very small pp-value, like p=0.002p = 0.002) to instead produce 1−p=0.9981 - p = 0.998, leading to the false conclusion that the precipitator had no significant effect.

Typical errors

OverviewACEpath Teacher's Diagnosis

01 The student associates the word 'reduces' with a less-than sign regardless of how the difference is subtracted, reversing the alternative hypothesis in part (a).

Sample Student Work

Reviewed by ACEpath Teacher
Step-by-Step Review
Walk Through It with Your Teacher
Partially correct
  1. Step 1

    H0:μdiff=0  ⟹  Ha:μdiff<0H_0: \mu_{\text{diff}} = 0 \implies H_a: \mu_{\text{diff}} < 0

    Correct

    H0:μdiff=0 mg/m3  ⟹  Ha:μdiff>0 mg/m3H_0: \mu_{\text{diff}} = 0\text{ mg/m}^3 \implies H_a: \mu_{\text{diff}} > 0\text{ mg/m}^3

    Because μdiff=μpre−μpost\mu_{\text{diff}} = \mu_{\text{pre}} - \mu_{\text{post}}, a reduction in emissions means that the pre-installation rate is larger than the post-installation rate (μpre>μpost\mu_{\text{pre}} > \mu_{\text{post}}). Subtracting μpost\mu_{\text{post}} gives μpre−μpost>0\mu_{\text{pre}} - \mu_{\text{post}} > 0, so the alternative hypothesis must be Ha:μdiff>0H_a: \mu_{\text{diff}} > 0.

  2. Step 2

    H0:μΔ=0  ⟹  Ha:μΔ<0H_0: \mu_{\Delta} = 0 \implies H_a: \mu_{\Delta} < 0

  3. Step 3

    Confusing the order reverses the sign of the difference, putting the rejection region in the wrong tail and failing to detect a real reduction.

WHAT YOU DID WELL
Compliments from Your Teacher

You correctly established the null and alternative hypotheses for part (b) using the post-minus-pre convention.

You provided an accurate and articulate description of the sign trap and its statistical consequence in part (c).

THE ERROR
Where Your ACEpath Teacher Spotted the Mistake
Reversed direction of paired difference hypothesis

You treated the word 'reduces' as an automatic less-than sign without tracking the order of subtraction.

WHY IT HAPPENED
ACEpath Teacher's Diagnosis

Associating natural-language direction words (such as 'reduction', 'decrease', or 'drop') directly with the << inequality symbol rather than setting up the underlying physical inequality (μpre>μpost\mu_{\text{pre}} > \mu_{\text{post}}) and algebraically rearranging it based on the defined subtraction order.

REASONING PATH
Your ACEpath teacher’s suggested approach
Establish the Physical Inequality First
  • Before choosing an inequality sign for a difference parameter, write the relationship between the two conditions directly: if emissions decrease after treatment, then μpre>μpost\mu_{\text{pre}} > \mu_{\text{post}}.

  • Always write μpre>μpost\mu_{\text{pre}} > \mu_{\text{post}} (or μpost<μpre\mu_{\text{post}} < \mu_{\text{pre}}) as your anchor statement before applying the difference formula.

Subtract to Match Parameter Order
  • Rearrange the physical inequality algebraically to match the parameter definition.

  • For Pre−Post\text{Pre} - \text{Post}, subtract μpost\mu_{\text{post}} from both sides of μpre>μpost\mu_{\text{pre}} > \mu_{\text{post}} to obtain μpre−μpost>0\mu_{\text{pre}} - \mu_{\text{post}} > 0.

CORRECTED PATH
ACEpath Teacher's Demonstration
  1. State the research question in terms of population means: a reduction means μpost<μpre\mu_{\text{post}} < \mu_{\text{pre}}, or equivalently μpre>μpost\mu_{\text{pre}} > \mu_{\text{post}}.

  2. Apply the parameter definition μdiff=μpre−μpost\mu_{\text{diff}} = \mu_{\text{pre}} - \mu_{\text{post}} by subtracting μpost\mu_{\text{post}} from both sides of μpre>μpost\mu_{\text{pre}} > \mu_{\text{post}}.

  3. Obtain μpre−μpost>0\mu_{\text{pre}} - \mu_{\text{post}} > 0, which translates to μdiff>0\mu_{\text{diff}} > 0.

Final answer

H0:μdiff=0 mg/m3H_0: \mu_{\text{diff}} = 0\text{ mg/m}^3 versus Ha:μdiff>0 mg/m3H_a: \mu_{\text{diff}} > 0\text{ mg/m}^3

NEXT STEPS
Final Takeaways from Your Teacher
  • Practice converting verbal claims (such as 'improved test scores' or 'reduced blood pressure') into two-variable inequalities before subtracting.

  • Check each paired hypothesis by plugging in hypothetical numbers (e.g., Pre =100= 100, Post =80= 80) to see if the resulting difference is positive or negative.

OverviewACEpath Teacher's Diagnosis

02 The student formulates a two-tailed test instead of a one-tailed test because they confuse testing for a directional reduction with testing for any general difference.

Sample Student Work

Reviewed by ACEpath Teacher
Step-by-Step Review
Walk Through It with Your Teacher
Incorrect
  1. Step 1

    H0:μdiff=0H_0: \mu_{\text{diff}} = 0, Ha:μdiff≠0H_a: \mu_{\text{diff}} \neq 0

    Correct

    H0:μdiff=0 mg/m3H_0: \mu_{\text{diff}} = 0\text{ mg/m}^3, Ha:μdiff>0 mg/m3H_a: \mu_{\text{diff}} > 0\text{ mg/m}^3

    You correctly set the null hypothesis to zero difference. However, because the study specifically investigates whether emissions were reduced (meaning μpre>μpost\mu_{\text{pre}} > \mu_{\text{post}}), the difference μdiff=μpre−μpost\mu_{\text{diff}} = \mu_{\text{pre}} - \mu_{\text{post}} must be strictly positive under the alternative hypothesis.

  2. Step 2

    H0:μΔ=0H_0: \mu_{\Delta} = 0, Ha:μΔ≠0H_a: \mu_{\Delta} \neq 0

    Correct

    H0:μΔ=0 mg/m3H_0: \mu_{\Delta} = 0\text{ mg/m}^3, Ha:μΔ<0 mg/m3H_a: \mu_{\Delta} < 0\text{ mg/m}^3

    You correctly stated the null hypothesis as zero. Under the convention μΔ=μpost−μpre\mu_{\Delta} = \mu_{\text{post}} - \mu_{\text{pre}}, an effective reduction means μpost<μpre\mu_{\text{post}} < \mu_{\text{pre}}, which implies μΔ<0\mu_{\Delta} < 0. The alternative hypothesis must therefore be left-tailed rather than two-sided.

  3. Step 3

    If the convention is switched, the sign of the difference flips, but in a two-sided test the p-value is unaffected since both tails are included.

    Correct

    The verbal term 'reduction' requires a one-sided test. If the analyst assumes a positive difference (>0> 0) while calculating differences as Post−Pre\text{Post} - \text{Pre} (which yields negative values), the test calculates the PP-value in the wrong tail, turning a highly significant outcome PP into 1−P1 - P.

    Your reasoning is logically consistent with your two-sided hypotheses from parts (a) and (b), but directional research questions require one-sided tests. When a directional test evaluates the wrong tail due to reversed subtraction, the calculated PP-value becomes 1−P1 - P, leading to an incorrect failure to reject the null hypothesis.

WHAT YOU DID WELL
Compliments from Your Teacher

You correctly identified that the null hypothesis represents no change (H0:μ=0H_0: \mu = 0) in both parameter definitions.

You recognized that changing the subtraction order flips the algebraic sign of the paired difference.

THE ERROR
Where Your ACEpath Teacher Spotted the Mistake
Two-tailed instead of one-tailed directional hypothesis

You used ≠\neq instead of a directional inequality (>> or <<) when testing for a specific reduction in emissions.

WHY IT HAPPENED
ACEpath Teacher's Diagnosis

You treated the research objective as testing for any general difference rather than translating the directional claim ('reduces emissions') into the appropriate inequality based on subtraction order.

REASONING PATH
Your ACEpath teacher’s suggested approach
Translate the research claim before choosing inequality direction
  • Directional verbs such as 'reduces', 'decreases', 'improves', or 'increases' always specify a one-sided alternative hypothesis (>> or <<), never a two-sided one (≠\neq).

  • Write the claim in terms of the original variables (e.g., μpre>μpost\mu_{\text{pre}} > \mu_{\text{post}}) first, then subtract to match the defined difference parameter.

Align tail direction with the subtraction definition
  • For a reduction: (Pre−Post)>0(\text{Pre} - \text{Post}) > 0 requires a right-tailed test, whereas (Post−Pre)<0(\text{Post} - \text{Pre}) < 0 requires a left-tailed test.

  • Check that the sign of your sample difference dˉ\bar{d} aligns with the rejection region of HaH_a before computing test statistics.

CORRECTED PATH
ACEpath Teacher's Demonstration
  1. Express the engineering goal: reduction means μpre>μpost\mu_{\text{pre}} > \mu_{\text{post}}.

  2. Rearrange to match μdiff\mu_{\text{diff}}: subtract μpost\mu_{\text{post}} from both sides to get μpre−μpost>0\mu_{\text{pre}} - \mu_{\text{post}} > 0, so Ha:μdiff>0 mg/m3H_a: \mu_{\text{diff}} > 0\text{ mg/m}^3.

  3. Rearrange to match μΔ\mu_{\Delta}: subtract μpre\mu_{\text{pre}} from both sides of μpost<μpre\mu_{\text{post}} < \mu_{\text{pre}} to get μpost−μpre<0\mu_{\text{post}} - \mu_{\text{pre}} < 0, so Ha:μΔ<0 mg/m3H_a: \mu_{\Delta} < 0\text{ mg/m}^3.

  4. Explain the pitfall: pairing a negative sample difference with a right-tailed hypothesis looks in the opposite tail, transforming a small PP-value into 1−P1 - P.

Final answer

(a) H0:μdiff=0 mg/m3H_0: \mu_{\text{diff}} = 0\text{ mg/m}^3 vs. Ha:μdiff>0 mg/m3H_a: \mu_{\text{diff}} > 0\text{ mg/m}^3; (b) H0:μΔ=0 mg/m3H_0: \mu_{\Delta} = 0\text{ mg/m}^3 vs. Ha:μΔ<0 mg/m3H_a: \mu_{\Delta} < 0\text{ mg/m}^3; (c) Confusing subtraction order inverts the required tail of the sampling distribution, resulting in an erroneous PP-value of 1−P1 - P.

NEXT STEPS
Final Takeaways from Your Teacher
  • Practice converting directional verbal claims into explicit algebraic inequalities before writing HaH_a.

  • Double-check whether each paired tt-test problem asks for a general difference (two-sided, ≠\neq) or an improvement/reduction (one-sided, >> or <<).

  • Trace test statistic signs: ensure that when data favor the researcher's claim, the test statistic falls into the critical tail specified by HaH_a.

Lock it in

Similar question

01

A clinic tests whether a new therapy reduces resting heart rate. Let XpreX_{\text{pre}} and XpostX_{\text{post}} be the heart rates before and after therapy. If the mean difference is defined as μd=μpre−μpost\mu_d = \mu_{\text{pre}} - \mu_{\text{post}}, write the null and alternative hypotheses.

Answer

H0:μd=0 bpmH_0: \mu_d = 0\text{ bpm} versus Ha:μd>0 bpmH_a: \mu_d > 0\text{ bpm}. Key step: A reduction means μpre>μpost\mu_{\text{pre}} > \mu_{\text{post}}, which rearranges to μpre−μpost>0\mu_{\text{pre}} - \mu_{\text{post}} > 0.

02

An athlete tracks sprint times before and after a training program. If the parameter is defined as μΔ=μafter−μbefore\mu_{\Delta} = \mu_{\text{after}} - \mu_{\text{before}}, state H0H_0 and HaH_a for the claim that the training makes athletes faster.

Answer

H0:μΔ=0 sH_0: \mu_{\Delta} = 0\text{ s} versus Ha:μΔ<0 sH_a: \mu_{\Delta} < 0\text{ s}. Key step: Being faster means times decrease (μafter<μbefore\mu_{\text{after}} < \mu_{\text{before}}), so subtracting μbefore\mu_{\text{before}} gives μΔ<0\mu_{\Delta} < 0.

03

An analyst tests if a cooling system reduces operating temperatures using μD=μpre−μpost\mu_D = \mu_{\text{pre}} - \mu_{\text{post}}. The sample mean difference is xˉD=+6.2∘C\bar{x}_D = +6.2^\circ\text{C} with t=+2.45t = +2.45 (p=0.01p = 0.01). If the analyst mistakenly set Ha:μD<0H_a: \mu_D < 0, what pp-value would statistical software report for this lower-tailed test, and what erroneous conclusion would be drawn at α=0.05\alpha = 0.05?

Answer

p=0.99p = 0.99. Key step: Testing the opposite tail yields 1−0.01=0.991 - 0.01 = 0.99. Because 0.99>0.050.99 > 0.05, the analyst would fail to reject H0H_0 and incorrectly conclude that the cooling system has no significant effect.

Loading workspace…