Presentation is loading. Please wait.

Presentation is loading. Please wait.

Reliability & Validity

Similar presentations


Presentation on theme: "Reliability & Validity"— Presentation transcript:

1 Reliability & Validity
How Science Works Reliability & Validity © SSER Ltd.

2 Reliability & Validity
Data should be reliable and valid. Reliable data is from a trusted source and is reproducible (repeatable under the same conditions). Valid data has been collected, recorded and interpreted correctly and relates directly to the original hypothesis. Valid data must be reliable, because the steps you take to ensure validity will also ensure reliability. However, reliable data is not necessarily valid... If you correctly measure an irrelevant variable your data could be considered reliable but certainly not valid as it does not relate to the hypothesis! Interpretation is the meaning and conclusions we derive from the data.

3 Improving Reliability & Validity
Only valid data can be used as evidence to support or refute a hypothesis. To be sure that the data is valid, you must pay close attention to: Selecting and setting-up apparatus The experimental method Cross-checking your results Interpreting your results appropriately

4 Selecting and Setting-up Apparatus

5 Selecting and Setting-up Apparatus
When selecting and setting-up apparatus, you should ask yourself questions, such as: Is the apparatus connected up properly and safe to use? If fluids (gases and liquids) are involved, are there any leaks? Does the apparatus allow the dependent variable to be correctly measured? Have the measuring instruments been calibrated prior to use? Have you selected measuring instruments that are accurate enough for your needs?

6 Range of the Independent Variable
What is wrong with the data from this experiment?

7 Range of the Independent Variable
The selected range for the independent variable was inappropriate as the water boiled in 3 minutes - most of the data is worthless. The range of the independent variable should be chosen so that the experiment produces valid results. This decision is usually made after the pilot test. For the previous experiment, you may have to reduce the heat setting or reduce the range of values of the independent variable from 0 to 3 minutes.

8 Increment of the Independent Variable
The increment is the increase from one value of the independent variable to the next. Smaller increments lead to more readings. Please start to show the animation...

9 The Experimental Method

10 The Experimental Method
Before starting the experiment, you should ask yourself questions, such as: Did you start with a clearly stated hypothesis and prediction? Have you identified the independent and dependent variables? Have you made sure that the changes in the dependent variable are only due to changes in the independent variable, and to no other variables, i.e. have you identified and controlled all the control variables? Have you ensured a fair test?

11 The Experimental Method
During and after the experiment, you should ask yourself questions, such as: Have you selected an appropriate range for the independent variable? Have you selected appropriate increments of the independent variable within the range? Have you used the apparatus correctly and safely during the experiment? Does your apparatus correctly measure the dependent variable? Has the data been collated, analysed and displayed correctly? Are your conclusions justified from the data?

12 Cross-Checking with Secondary Data
Reliability can be increased by referring to secondary data sources. Secondary data has already been collected by other scientists, doing the same or similar experiment.

13 Cross-Checking with Secondary Data
There are many sources of secondary data, such as: Another team of students within your class The Internet Government statistical reports Scientific articles in newspapers and magazines The reliability of your own data is increased if it agrees with secondary data, since it shows your data is reproducible.

14 Cross-Checking with Secondary Data
However secondary data must be treated carefully, because: The way the data was collected and analysed may not be known, so it may not be possible to guarantee that the data is unbiased. The data may not be exactly what is required: The experimental setting could be different from your own experiment, e.g. an experiment performed in Alaska to determine the effectiveness of insulation, does not necessarily support data collected from an experiment performed in the UK The data may be out of date

15 Cross-Checking Through Repeated Measuring
Reliability can be increased by using other team members to take the same measurement, or by using another more precise measuring instrument as a cross-check. Example: If you are measuring the temperature of a liquid you could gather results using: An analogue thermometer containing mercury An analogue thermometer containing alcohol A digital thermometer If the three readings are the same, then you can be more confident that the measurements collected are reliable.

16 Anomalous Results “An anomalous result is one which appears to be inconsistent with the rest of the data - it stands out.” Anomalous results can occur for a number of reasons, such as equipment malfunction or human error. Consider the set of readings: 5 7 4 13 6 The 13 stands out as very different from the rest of the readings. The piece of data ‘13’ should be measured again to check if it is a correct reading. Don’t immediately assume that an anomaly is an error - an anomaly led Fleming to the discovery of Penicillin!

17 Human Error

18 Human Error

19 Human Error “No matter how good a measuring instrument is, if it is not used carefully then errors in readings will occur.” There are two types of human error: Random Systematic Both random and systematic errors lead to incorrect readings.

20 Random Error “Random errors occur ‘out of the blue’ from an inconsistent application of a technique.” Random errors are caused by: A momentary loss of concentration Distraction Such a reading would be easily recognised in a set of repeated readings, as it would stand out from the others. Random errors lead to unreliable data.

21 Systematic Error “Systematic error results from a consistent misapplication of a technique.” Examples of systematic errors: A balance was not calibrated properly at the start of an investigation Readings on an electronic balance were repeatedly misread A ruler was used from its end, not the zero line Systematic errors lead to invalid data.

22 Human Error: Random or Systematic?
For each of the errors identified below, decide if it is random or systematic. Complete each question, then click ‘check answer’...

23 Summary Questions

24 End of Show Copyright © 2006 SSER Ltd. and its licensors.
Images are for viewing purposes only. All rights reserved.


Download ppt "Reliability & Validity"

Similar presentations


Ads by Google