Validity, Reliability, Accuracy

Validity

Validity of a first-hand investigation relates to the extent to which the method planned and followed allows accurate data to be gathered that addresses the AIM of the investigation.

 

Validity for 1st hand investigation depends on making sure it is a fair test. The following points are important:

  • controlling relevant variables (only the independent variables should change).

  • Using the most appropriate equipment (as sensitive as possible) to improve accuracy and ensure the equipment is used correctly.

  • using a suitable range of independent variables to ensure the aim is tested.

 

Mnemonic to help recall factors to discuss when assessing validity: ACER

Reliability

Reliability describes the relative consistency of the data measured in repeated trials conducted.

 

To assess reliability, you must repeat the procedure for each independent variable at least 3 times and see if the results are fairly consistent.

 

If outliers are measured, repeat the experiment again.

 

If the procedure cannot produce consistent results in the trials, it is deemed unreliable and aspects of the method must be considered and modifications made in order for reliability to be improved. That is, an investigation which cannot produce reliable results must also be invalid.

 

To improve reliability of an investigation so that another scientist will get similar results, you should repeat several times, remove any outliers (with justification) and average the set of consistent results.

 

Repeating an investigation and removing outliers and averaging results will likely increase the accuracy of the experiment, because it minimises the effect of outliers due to random errors.

Accuracy

Accuracy for 1st hand investigations refers to how close the experimental result obtained is to the theoretical or accepted value. Accuracy depends on the design of the experiment (i.e. the validity of the method) and the sensitivity and correct use of the instruments used. If the aim of an experiment is to measure something quantitatively, and the result is inaccurate, the method’s validity must be questioned. And as the picture shows, having reliable results does not guarantee the experiment is producing accurate data.

Constructing Scientific Tables

All tables should have a title (heading) that tells the person analysing the table what information is contained in the table.

 

2.    A border surrounds the entire table. All lines in the table must be ruled with a ruler.

 

3.    The table should be written in pencil, so that changes can be erased. Planning the format of the table may assist is reducing mistakes.

 

4.    Each column must have a heading. This will be the top cell in each column.

 

5.    If a column contains measurements, the units of the measurement should be placed in the column heading (within brackets). The unit should NOT be repeated in each cell underneath the heading.

 

For example: If the temperature of water was measured each minute, the column heading would be: “Temperature (0C)”.

 

6.    Observations and/or data in each column should be aligned.

 

7.    A zero should only be included when a measurement of zero was obtained; a dash should be used when no reading was recorded.

 

8.    Identical results should be written again-not shown with ‘ditto marks’.

 

9.    It is often appropriate and useful to show data for each trial along with average values. Subtotals, totals and/or percentages can be included.

 

Drawing Graphs

Line graphs are the most common type of graph you draw in Science. They are used when presenting continuous data where two variables are used.

  

An appropriate line graph should:

  • be drawn using a sharp pencil

  • have a clear and descriptive title

  • display the independent variable on the x-axis (the variable the experimenter purposely changes in an experiment)

  • display the dependent variable on the y-axis (the variable which changes in response to changes in the independent variable)

  • have axes with a clearly-marked, even scale that increases with even increments

  • always have the vertex representing (0,0)

  • have a scale which allows the range of data displayed to extend over more than half the grid in both directions (if the data is spread over a small range this can be done by starting at zero at the vertex, using a hash or a zig-zag, then starting with a value close to that of your smallest value)

  • have axes labelled with the appropriate quantity

  • have axes labelled with the appropriate units

  • have accurately plotted points drawn as neat, small crosses

  • a line or curve of best fit is required to represent a trend (unless told to not join the centre of each cross)

  • have a clear key if more than one line is shown on a graph using different lines (not points) for different data sets

Scientific Report

  1. Give your report a descriptive title.

 

  1. Include an aim.

  • Start your aim with the words To investigate, To determine or To observe.

  • The aim should include both variables that you are investigating.

  • Your dependant variable is usually written first, the independent variable written second.

 

  1. Include an equipment list.

 

  1. Writing a detailed and concise method.

A method should be written such that anyone who reads it could carry out the experiment without difficulty. A method should also mention how each aspect of the experiment is controlled.

  1. Use numbered steps (to be clear, use one numbered step for each action carried out)

  2. Do not omit any steps carried out in your method or write them out of order, but keep each step brief and to the point.

  3. Name all equipment used in each step.

  4. Mention exact quantities of measurements taken and the units used.

  5. If possible, record how accurately you took your measurements (ie “…to two decimal places”, or “…to the nearest mm”).

  6. For reliability, each set of results should be repeated at least 3 times and then averaged, to reduce inconsistent readings.

 

A method can be written in two ways, depending on when you conduct the experiment. If you are simply planning a method (or writing it before you carry it out) then it can be written as a series of steps in the present tense such as:

EXAMPLE: “Using an electronic balance, record the mass of one teaspoon of sodium chloride, in grams, correct to two decimal places”.

 

 

However, if you are writing a scientific report AFTER you have carried out an experiment, then it is known as a procedural recount.  A procedural recount should be written in past tense.

 

EXAMPLE: “An electronic balance was used to record the mass of one teaspoon of sodium chloride, in grams, correct to two decimal places”

OR

“The mass of one teaspoon of sodium chloride was recorded, using an electronic balance, in grams, correct to two decimal places”

 

A procedural recount should be written impersonally. You should not use terms such as I and we.

EXAMPLE:             “Water was poured to the beaker”

NOT

“I poured water into the beaker” or “Poured water into the beaker”

 

  1. Drawing a scientific diagram.

To draw an appropriate scientific diagram you should:

  • use a sharp pencil

  • use a ruler

  • draw using unbroken lines

  • draw it in 2D

  • make sure the diagram is an appropriate size (approximately 1/3 of a page)

  • label each piece of equipment

  • draw each piece of equipment in proportion

  • draw the equipment as it is set up, not separately

  • not include equipment such as electronic balances

 

  1. Presenting your results.

The results you record should be observations only. No inferences should be made.

  • If you are collecting numerical data, use a table to present your results (following the SCEGGS guidelines).

  • If your collected data allows it, a graph should be plotted (following the SCEGGS guidelines).

  • If no numerical data was collected, then write out your observations as clearly as possible (a table might still be appropriate here).

 

G.    Include a discussion.

A discussion explains your experiment to another person. It should include:

  • an explanation of your results.

  • the identification of errors in your experiment.

  • an assessment of the reliability of the experiment (repeat and average).

  • an assessment of the validity of the experiment (comment on; how well the variables are controlled, if the most appropriate equipment was used, if there was a suitable range of measurements).

  • any suggestions of improvements that could be made to increase the accuracy of the experiment.

 

H.    Writing a conclusion.

The conclusion should be one sentence that answers the aim specifically. It should be a general statement that does not include any results or personal pronouns.

EXAMPLE: As the temperature of the water increased, the time taken for the sugar to dissolve decreased.