What Are Conclusions and Evaluations?

The final parts of a scientific report answer two different questions.

The conclusion asks:

What did the investigation show?

The evaluation asks:

How good was the investigation, and how could it be improved?

A strong scientific report needs both.

The conclusion uses the evidence to answer the research question.

The evaluation examines the quality of the method and evidence.

---

The Conclusion

A scientific conclusion should provide a direct answer to the investigation question.

Suppose the research question is:

How does water temperature affect the time required for sugar to dissolve?

A conclusion should not simply say:

The experiment worked.

or:

My hypothesis was correct.

Instead, it should explain the relationship found.

For example:

Increasing water temperature decreased the time required for sugar to dissolve over the range tested.

That directly answers the research question.

---

A Strong Conclusion Uses Evidence

A conclusion becomes stronger when it includes specific experimental evidence.

Weak:

Higher temperatures made the sugar dissolve faster.

Better:

Increasing water temperature decreased dissolving time.

Strong:

Increasing water temperature from 20°C to 60°C decreased the mean dissolving time from 84 s to 25 s.

The final statement gives the reader evidence supporting the conclusion.

---

Claim, Evidence, and Reasoning

A useful structure for writing conclusions is:

Claim → Evidence → Reasoning

Claim

Answer the research question.

Evidence

Provide important measurements or trends.

Reasoning

Explain how scientific ideas connect the evidence to the claim.

---

Example of Claim–Evidence–Reasoning

Research question:

How does applied force affect the extension of a spring?

Claim:

Increasing applied force increased spring extension.

Evidence:

Mean extension increased from 2.0 cm at 1 N to 10.1 cm at 5 N.

Reasoning:

Over the tested range, extension increased approximately proportionally with force, consistent with Hooke's law.

A complete conclusion might therefore be:

Increasing applied force increased the extension of the spring. Mean extension increased from 2.0 cm at 1 N to 10.1 cm at 5 N, with an approximately linear relationship between force and extension. These results are consistent with Hooke's law over the tested range.

---

Answer the Research Question Directly

The reader should not need to search through the conclusion to discover the answer.

Research question:

How does pendulum length affect period?

Good opening:

The period of the pendulum increased as pendulum length increased.

Then provide evidence and explanation.

This structure makes the conclusion clear immediately.

---

Refer to Important Numerical Evidence

A strong conclusion usually includes selected evidence rather than repeating every measurement.

Suppose:

20 cm → 0.91 s

40 cm → 1.28 s

60 cm → 1.56 s

80 cm → 1.80 s

100 cm → 2.02 s

You do not need to rewrite the entire table.

Instead:

Increasing pendulum length from 20 cm to 100 cm increased the mean period from 0.91 s to 2.02 s.

This captures the important trend efficiently.

---

Conclusions Should Match the Evidence

Do not make a conclusion stronger than the evidence allows.

Suppose you test temperatures between:

20°C and 60°C

You can conclude:

Reaction rate increased with temperature over the tested range of 20–60°C.

It would be much less justified to conclude:

Increasing temperature always increases reaction rate at every possible temperature.

Scientific conclusions should recognize the conditions and range actually investigated.

---

Hypothesis and Conclusion

The conclusion should usually state whether the evidence:

supports

or:

does not support

the hypothesis.

For example:

The results support the hypothesis that increasing water temperature decreases dissolving time.

This is preferable to:

My hypothesis was proven correct.

One investigation usually provides evidence for or against a hypothesis rather than proving a broad scientific idea with absolute certainty.

---

What If the Hypothesis Was Wrong?

That is completely acceptable scientifically.

Suppose the hypothesis predicted:

Plant growth will continuously increase as fertilizer concentration increases.

But the results show:

A good conclusion would state:

The results do not support the hypothesis that plant growth continuously increases with fertilizer concentration. Growth increased from 5.2 cm without fertilizer to a maximum of 8.4 cm at 2% fertilizer, but decreased at higher concentrations.

Unexpected results are still useful results.

---

Avoid "Proving" a Hypothesis

Scientific investigations involve uncertainty.

It is generally more appropriate to write:

• supports the hypothesis
• does not support the hypothesis
• is consistent with the hypothesis
• provides evidence for
• suggests a relationship

rather than:

• proves
• definitely proves
• shows with absolute certainty

The strength of the wording should match the strength of the evidence.

---

Scientific Reasoning in a Conclusion

A strong conclusion may briefly connect the observed pattern to scientific theory.

For example:

Reaction time decreased as temperature increased. This is consistent with collision theory because particles have greater average kinetic energy at higher temperatures, increasing the frequency of collisions and the proportion of collisions with sufficient energy to react.

The explanation should be relevant and connected directly to the investigation.

---

A Useful Conclusion Structure

A strong conclusion can often be built using four parts:

1. Answer the research question.

2. Describe the main trend.

3. Support the conclusion with numerical evidence.

4. Connect the result to scientific theory and the hypothesis.

For example:

Increasing temperature increased the rate of the reaction. Mean reaction time decreased from 81.6 s at 20°C to 28.2 s at 60°C. This supports the original hypothesis and is consistent with collision theory because particles have greater average kinetic energy at higher temperatures.

---

What Is an Evaluation?

The evaluation examines the quality of the investigation.

It asks:

• What worked well?
• What did not work well?
• How reliable are the results?
• Were important variables controlled?
• What sources of error existed?
• How significant were those errors?
• How could the method be improved?
• What could be investigated next?

An evaluation should be specific and evidence-based.

---

Strengths of an Investigation

A good evaluation should identify strengths as well as weaknesses.

Possible strengths include:

• repeated trials
• controlled variables
• appropriate equipment
• wide range of independent-variable values
• small measurement intervals
• consistent measurements
• clear operational definitions
• suitable sample size
• calibrated equipment
• automated measurements
• appropriate safety procedures

---

Example of a Strength

Weak:

The experiment was good.

Strong:

Three trials were completed at each temperature, allowing a mean to be calculated and making anomalous measurements easier to identify.

The stronger statement identifies exactly why that part of the method was useful.

---

Another Strength

Suppose temperature was maintained using a thermostatically controlled water bath.

A useful evaluation might state:

Using a thermostatically controlled water bath helped maintain the intended temperature throughout each trial, reducing variation in the independent variable.

This connects the method to data quality.

---

Weaknesses and Limitations

A limitation is a feature of the investigation that restricts the quality, accuracy, validity, or interpretation of the evidence.

Examples include:

• small sample size
• few repeated trials
• limited range of values
• poor control of variables
• low-resolution measuring equipment
• subjective endpoint
• human reaction time
• heat loss
• inconsistent technique
• equipment calibration problems

A good evaluation explains the effect of the limitation.

---

Limitation → Effect

Weak:

There was human error.

Better:

The stopwatch was started and stopped manually, so differences in human reaction time may have caused variation in the measured reaction times.

Even better:

Because reaction times were measured manually, differences in starting and stopping the stopwatch could shift individual measurements by a fraction of a second, increasing random variation between trials.

---

Avoid the Phrase "Human Error" by Itself

"Human error" is usually too vague to be useful.

Instead, identify exactly what could have happened.

For example:

Instead of:

Human error affected the experiment.

write:

The maximum rebound height was estimated visually, making it difficult to identify the exact highest point of the ball.

Now the limitation can be addressed.

---

Sources of Error

Experimental errors are not necessarily mistakes.

In science, error often refers to differences between a measured value and the value that would be obtained under ideal conditions.

Two important categories are:

• random error
• systematic error

---

Random Error

Random errors produce unpredictable variation between measurements.

Possible causes include:

• human reaction time
• small differences in reading a scale
• natural variation between biological samples
• small environmental changes
• inconsistent release technique

Repeated measurements can help reduce the effect of random error on a calculated mean.

---

Example of Random Error

Suppose a student measures:

12.1 s

12.6 s

11.9 s

12.4 s

The variation may result partly from reaction time when operating the stopwatch.

Repeating the measurement allows a mean to be calculated and provides information about the variation.

---

Systematic Error

A systematic error shifts measurements consistently in the same direction.

Examples include:

• balance incorrectly zeroed
• thermometer consistently reading 2°C too high
• ruler with a damaged zero point
• incorrectly calibrated sensor

Repeating the experiment does not automatically remove systematic error.

If the same faulty instrument is used every time, the same bias may remain.

---

Random vs Systematic Error

Random error:

Measurements scatter unpredictably.

Can often be reduced by:

repeated measurements and averaging

Systematic error:

Measurements are consistently shifted.

Can often be reduced by:

calibration, improved equipment, or correcting the method

These require different solutions.

---

Measurement Uncertainty

Every measurement has some degree of uncertainty.

Suppose a ruler has markings every:

1 mm

The measurement cannot be known with unlimited precision.

Similarly, a thermometer, measuring cylinder, balance, stopwatch, or sensor has limits.

A good evaluation considers whether uncertainty is significant compared with the measured effect.

---

Absolute Uncertainty

Suppose a length is reported as:

25.0 ± 0.1 cm

The:

±0.1 cm

is the absolute uncertainty.

This communicates a range around the reported measurement.

---

Percentage Uncertainty

Percentage uncertainty can be calculated as:

Percentage uncertainty = (absolute uncertainty ÷ measured value) × 100

Example:

Measurement:

25.0 ± 0.1 cm

Percentage uncertainty:

(0.1 ÷ 25.0) × 100

= 0.4%

---

Why Percentage Uncertainty Matters

Suppose the same absolute uncertainty is:

±0.1 cm

For a measurement of:

1.0 cm

percentage uncertainty is:

10%

For:

100.0 cm

percentage uncertainty is:

0.1%

The same absolute uncertainty can therefore have very different importance depending on the size of the measurement.

---

Accuracy

Accuracy describes how close a measurement is to an accepted or true value.

Suppose an accepted value is:

9.81 m/s²

Experimental result:

9.76 m/s²

The result is quite close to the accepted value.

However, accuracy should not be assumed simply because repeated measurements are similar.

---

Precision

Precision concerns the closeness of repeated measurements and/or the resolution with which measurements are reported, depending on context.

For example:

15.2, 15.2, 15.3, 15.2 cm

show relatively little variation.

But precise measurements can still be systematically inaccurate.

---

Reliability

Reliability concerns the consistency of results.

Evidence for good reliability may include:

• repeated trials producing similar values
• similar results from different groups
• a clear trend despite small variation
• successful replication

Reliability can often be improved by increasing the number of repeated measurements.

---

Validity

Validity asks whether the investigation actually tested what it intended to test.

Suppose the research question asks:

How does temperature affect reaction rate?

But both temperature and concentration change during the experiment.

Then it becomes difficult to determine whether temperature caused the observed difference.

The investigation has a validity problem.

---

Controlling Variables Improves Validity

Important controlled variables should remain as consistent as reasonably possible.

For a reaction-rate experiment, these might include:

• reactant volume
• reactant concentration
• reactant mass
• particle size
• mixing technique
• apparatus

If these change, they may provide alternative explanations for the results.

---

Evaluating Anomalous Results

An anomaly should be considered during evaluation.

Suppose:

Trial 1 = 32.1 s

Trial 2 = 31.8 s

Trial 3 = 49.7 s

Trial 4 = 32.0 s

The 49.7 s result differs substantially from the others.

A useful evaluation might state:

The third trial produced an anomalous value of 49.7 s compared with the other measurements of approximately 32 s. Repeating this condition would help determine whether the result was caused by random experimental variation or represents a reproducible effect.

---

Improvements Must Be Specific

A good improvement addresses a specific weakness.

Use the structure:

Problem → Effect → Improvement

For example:

Problem: The endpoint was judged visually.

Effect: Different observers may stop the timer at slightly different moments.

Improvement: Use a light sensor or colorimeter to define the endpoint objectively.

This is much stronger than:

Use better equipment.

---

Example: Stopwatch Improvement

Weak:

Use a better stopwatch.

This may not solve the problem if human reaction time is the main limitation.

Better:

Use an electronic sensor connected to a data logger so that timing begins and ends automatically, reducing uncertainty caused by human reaction time.

The improvement addresses the actual source of error.

---

Example: Temperature Improvement

Problem:

The solution cooled during the experiment.

Effect:

The actual temperature was not constant throughout each trial.

Improvement:

Place the reaction vessel in a thermostatically controlled water bath throughout each trial.

---

Example: Rebound Height Improvement

Problem:

Students estimated the maximum height of a bouncing ball by eye.

Effect:

The maximum height occurred quickly and was difficult to judge accurately.

Improvement:

Record the bounce using a camera positioned perpendicular to a measurement scale and determine maximum height using frame-by-frame video analysis.

This is realistic and directly addresses the weakness.

---

Example: Spring Investigation Improvement

Problem:

The ruler was positioned several centimetres behind the spring.

Effect:

Viewing the ruler from an angle could produce parallax error.

Improvement:

Position the ruler directly beside the spring and take readings at eye level perpendicular to the scale.

---

Realistic Improvements

Improvements should be practical.

Suppose the limitation is human stopwatch timing.

Possible realistic improvements include:

• light gates
• video analysis
• automatic sensors
• longer timing intervals

An unrealistic improvement might be:

Use perfectly accurate equipment with zero uncertainty.

No real measurement has zero uncertainty.

---

More Trials Are Not Always the Best Improvement

Students often write:

Do more trials.

This can be useful when random variation is a problem.

But repeated trials do not fix every limitation.

For example, if a thermometer always reads 3°C too high, repeating the measurement 100 times does not remove that systematic bias.

The improvement must match the problem.

---

Improving the Range

Suppose an investigation tests:

20°C, 30°C, and 40°C

A useful extension might test:

10°C to 70°C

if this is scientifically relevant and safe.

A larger range may reveal whether the observed relationship continues or changes.

---

Improving the Intervals

Suppose an investigation tests:

0 N, 5 N, and 10 N

A more detailed investigation might test:

0 N, 1 N, 2 N, 3 N...

Smaller intervals can provide more information about the shape of the relationship.

However, more data points should be chosen for a scientific reason rather than simply because "more is better."

---

Improving Sample Size

Biological investigations often involve natural variation.

Suppose a plant experiment uses:

one plant per treatment

A difference could result from natural differences between individual plants.

Using multiple plants at each condition and comparing mean growth would provide stronger evidence.

---

Strengths and Weaknesses Together

A balanced evaluation recognizes both.

Example:

A strength of the investigation was that five different temperatures were tested and three trials were completed at each temperature, producing enough data to identify a clear trend. However, temperature was measured only at the beginning of each trial, so cooling during the reaction may have caused the actual temperature to differ from the intended value.

This provides a more meaningful evaluation than listing only problems.

---

Prioritize Important Limitations

Not every imperfection matters equally.

A useful evaluation focuses on limitations that could significantly affect:

• measurements
• trends
• validity
• conclusion

For example, handwriting quality on the data sheet is unlikely to be an important experimental limitation unless it caused data to be recorded incorrectly.

Focus on scientifically meaningful issues.

---

Direction of Error

When possible, explain how a limitation may affect the measurement.

Suppose heat escapes during an experiment intended to measure energy transferred to water.

The calculated energy change may be:

lower than the true energy released

because some energy was transferred to the surroundings rather than the water.

This is more useful than simply stating:

Heat was lost.

---

Example: Heat Loss

Weak:

Heat loss was an error.

Strong:

Some thermal energy was transferred from the reaction mixture to the surroundings. Therefore, the measured temperature increase was probably smaller than it would have been in a perfectly insulated system, causing the calculated energy transferred to the water to be underestimated.

This explains the direction and consequence of the error.

---

Future Investigations

A scientific investigation often raises new questions.

The final evaluation can therefore suggest how future research could build upon the original experiment.

This is different from simply improving the same method.

An improvement makes the original investigation better.

An extension asks a new or broader question.

---

Improvement vs Extension

Original question:

How does temperature affect reaction rate?

Improvement:

Use a thermostatically controlled water bath to maintain constant temperatures.

Extension:

Investigate whether temperature has the same effect on reaction rate at different reactant concentrations.

The improvement strengthens the existing experiment.

The extension develops a new investigation.

---

Future Investigation Example: Springs

Original investigation:

How does force affect spring extension?

Possible future investigations:

• How does spring material affect spring constant?
• How does spring length affect spring constant?
• At what force does the spring stop obeying Hooke's law?
• How does repeated loading affect the spring?
• How does temperature affect spring behaviour?

These questions build logically from the original investigation.

---

Future Investigation Example: Enzymes

Original:

How does temperature affect enzyme activity?

Possible extensions:

• How does pH affect enzyme activity?
• How does substrate concentration affect activity?
• How does enzyme concentration affect activity?
• Do different enzymes have different optimum temperatures?
• How quickly does activity decrease after denaturation?

A useful extension should connect scientifically to the original results.

---

Future Investigation Example: Plant Growth

Original:

How does light intensity affect plant growth?

Possible extensions:

• How does light colour affect growth?
• Does the effect differ between plant species?
• How does light exposure time affect growth?
• Does increased light still improve growth when water is limited?
• How does light intensity affect leaf number rather than plant height?

---

Future Research Should Have a Reason

Do not simply write:

Next time I would investigate something different.

Explain why the extension matters.

For example:

The results showed that plant growth increased up to 2% fertilizer concentration but decreased at higher concentrations. A future investigation could test concentrations between 1% and 3% at smaller intervals to estimate the concentration associated with maximum growth more precisely.

This future investigation arises directly from the original results.

---

Worked Example 1: Complete Conclusion

Research question:

How does water temperature affect dissolving time?

Results:

20°C → 91 s

30°C → 70 s

40°C → 51 s

50°C → 38 s

60°C → 29 s

Conclusion:

Increasing water temperature decreased the time required for sugar to dissolve. Mean dissolving time decreased from 91 s at 20°C to 29 s at 60°C, supporting the original hypothesis. This trend is consistent with the particle model because particles have greater average kinetic energy at higher temperatures, increasing particle motion and interactions between the water and sugar.

---

Worked Example 2: Complete Evaluation

A strength of the investigation was that three trials were performed at each of five temperatures, allowing mean values to be calculated and making unusual results easier to identify. However, the water temperature was measured only at the start of each trial and may have changed while the sugar dissolved. This means the actual temperature was not perfectly controlled. Future trials could use a thermostatically controlled water bath to maintain a constant temperature. In addition, stirring was performed manually, which may have introduced variation between trials. A mechanical stirrer operating at a fixed rate would make this variable more consistent.

This evaluation connects:

strengths → limitations → effects → improvements

---

Worked Example 3: Future Investigation

Continuing the dissolving investigation:

The original investigation tested only temperature. A future investigation could examine how particle size affects dissolving time while keeping temperature constant. This would determine whether increasing the surface area of the solute produces a measurable change in dissolving behaviour.

The extension builds logically on the original topic.

---

Error Analysis

A student writes:

Conclusion: My hypothesis was correct and the experiment worked.

Problems:

• does not answer the research question
• provides no data
• does not identify a trend
• provides no scientific explanation
• overstates what the experiment established

Improved:

The results support the hypothesis that increasing temperature decreases reaction time. Mean reaction time decreased from 76 s at 20°C to 28 s at 60°C, indicating that reaction rate increased with temperature over the range tested.

---

Another Error Analysis

Student evaluation:

The experiment was bad because of human error.

Problems:

• vague
• does not identify the limitation
• does not explain its effect
• provides no improvement

Improved:

Reaction time was measured manually with a stopwatch, so differences in human reaction time may have increased variation between trials. An automated timing system using an appropriate sensor could reduce this source of measurement uncertainty.

---

Another Error Analysis

Student improvement:

Do more trials.

Question:

What problem does that solve?

If the problem is random variation, more trials may help.

If the problem is:

• incorrect calibration
• uncontrolled temperature
• unsuitable equipment
• poorly defined endpoint

then additional trials alone will not solve it.

Always connect the improvement to the identified limitation.

---

Another Error Analysis

Student writes:

The result was inaccurate because the data were different.

Variation does not automatically mean measurements are inaccurate.

The student should identify:

• what varied
• how much it varied
• whether the variation was expected
• whether an accepted value is available
• what experimental feature may have caused the variation

Use scientific terminology carefully.

---

A Reliable Conclusion Strategy

Use this sequence:

Step 1: Restate the relationship found.

Step 2: Answer the research question directly.

Step 3: Include important numerical evidence.

Step 4: State whether the evidence supports the hypothesis.

Step 5: Connect the result to relevant scientific theory.

Step 6: Avoid making claims beyond the investigated range.

---

A Reliable Evaluation Strategy

Use this sequence:

Step 1: Identify strengths.

Step 2: Identify important limitations.

Step 3: Identify specific sources of error or uncertainty.

Step 4: Explain how each limitation may affect the results.

Step 5: Suggest a realistic improvement for each important limitation.

Step 6: Consider reliability and validity.

Step 7: Suggest a meaningful future investigation.

A particularly useful structure is:

Limitation → Effect → Improvement

---

Conclusion Checklist

Before submitting your conclusion, ask:

• Did I answer the research question?
• Did I state the overall relationship?
• Did I use numerical evidence?
• Did I refer to the hypothesis?
• Did I use appropriate scientific theory?
• Did I avoid simply repeating the results?
• Did I avoid claiming that one experiment "proved" the hypothesis?
• Did I limit my claims to what the evidence supports?

---

Evaluation Checklist

Before submitting your evaluation, ask:

• Did I identify strengths?
• Did I identify specific weaknesses?
• Did I avoid vague statements such as "human error"?
• Did I identify random or systematic errors where appropriate?
• Did I discuss measurement uncertainty?
• Did I explain how important limitations affected the data?
• Did I consider reliability?
• Did I consider validity?
• Does each major weakness have a realistic improvement?
• Does each improvement actually address the problem?
• Did I prioritize important limitations?
• Did I suggest a meaningful future investigation?
• Did I explain how the future investigation builds on the original work?

---

Did You Know?

A result that does not support your hypothesis can still represent excellent science.

The purpose of an investigation is not to make the hypothesis "win."

The purpose is to collect evidence and determine what that evidence supports.

Sometimes unexpected results reveal:

• problems with an existing explanation
• limitations in the experimental method
• variables that were not previously considered
• entirely new questions

In science, a useful investigation often ends by creating the question for the next investigation.

---

Key Terms

• Conclusion: Evidence-based answer to the research question.
• Evaluation: Assessment of the quality and limitations of an investigation.
• Evidence: Measurements and observations supporting a scientific claim.
• Hypothesis: Testable prediction based on scientific reasoning.
• Strength: Feature that improves the quality of an investigation.
• Limitation: Feature restricting the quality or interpretation of an investigation.
• Random error: Unpredictable variation between measurements.
• Systematic error: Consistent bias affecting measurements in the same direction.
• Uncertainty: Limitation in how precisely a quantity can be known.
• Accuracy: Closeness of a measurement to an accepted or true value.
• Precision: Closeness of repeated measurements and/or fineness of measurement resolution, depending on context.
• Reliability: Consistency of measurements or results.
• Validity: Extent to which an investigation appropriately tests the intended question.
• Anomaly: Result that does not fit the general pattern.
• Improvement: Change designed to reduce a limitation in the investigation.
• Extension: New investigation that develops or expands upon the original research.
• Replication: Independent repetition of an investigation.

---

Key Relationships

A strong conclusion follows:

Claim → Evidence → Scientific Reasoning

A strong evaluation follows:

Strength → Limitation → Effect → Improvement

A useful improvement follows:

Specific Problem → Specific Solution

Random error can often be reduced through:

Repeated Measurements → Mean → Reduced Influence of Random Variation

Systematic error usually requires:

Identify Bias → Calibrate or Change Method/Equipment

Scientific investigation continues through:

Question → Investigation → Evidence → Conclusion → Evaluation → New Question

---

Key Takeaways

• A conclusion should directly answer the investigation question.
• Strong conclusions use specific experimental evidence.
• Numerical evidence makes conclusions more precise.
• Conclusions should describe the main relationship rather than repeat every measurement.
• Scientific reasoning should connect the evidence to relevant theory.
• Results can support or fail to support a hypothesis.
• A hypothesis should not usually be described as "proven" by a single experiment.
• Conclusions should not extend beyond the range or conditions actually investigated without justification.
• Unexpected results can still provide valuable scientific evidence.
• Evaluation examines the quality of an investigation.
• Good evaluations identify both strengths and weaknesses.
• Strengths should explain why a feature improved the investigation.
• Weaknesses should identify specific limitations rather than vague "human error."
• Random errors cause unpredictable variation.
• Systematic errors produce consistent bias.
• Repeating trials can reduce the influence of random variation but does not automatically correct systematic error.
• All measurements contain some uncertainty.
• Percentage uncertainty can help compare uncertainty between measurements of different sizes.
• Accuracy, precision, reliability, and validity describe different aspects of experimental quality.
• Controlled variables are important for validity.
• Anomalous results should be identified and investigated.
• Good evaluations explain how limitations may have affected the results.
• When possible, the direction of an error should be discussed.
• Improvements should be specific, realistic, and directly connected to identified limitations.
• "Do more trials" is useful only when additional repetition addresses the actual problem.
• Improvements and extensions are different.
• An improvement strengthens the original investigation.
• An extension asks a new question that builds upon the original work.
• Future investigations should arise logically from the evidence, limitations, or unanswered questions of the original investigation.
• Scientific investigation is an ongoing process in which conclusions often lead to new questions.