Transcript 2b variables

Unit 3- Investigative Biology
Topic 2- Experimentation
b) Variables
Variables what you need to know
• Controlling and or monitoring confounding
variables, including randomised block
design.
• Variables can be discrete and continuous
variables that can give rise to qualitative,
quantitative or ranked data.
• Due to the complexities of biological
systems, other variables besides the
independent variable may affect the
dependent variable.
Variables what you need to know
• These confounding variables must be held constant
if possible, or at least monitored so that their
effect on the results can be accounted for in the
analysis.
• In cases where confounding variables cannot easily
be controlled, blocks of experimental and control
groups can be distributed in such a way that the
influence of any confounding variable is likely to be
the same across the experimental and control
groups.
• The type of variable being investigated has
consequences for any graphical display or
statistical tests that may be used.
Variables
•
Any factor in an experiment can be a variable.
•
However, in a perfect experiment all possible factors are kept constant
and tightly controlled apart from the independent variable.
– The only factor that is changed is the independent variable.
– This allows cause and effect to be established.
•
Biology is not always that simple – other variables outside the
independent variable may potentially affect the independent variable.
•
Such variables are called confounding variables.
Variables
Confounding variables
•
Must be kept constant if possible, or at least monitored so that their
effect on the results can be accounted for in the analysis.
•
Any change in a confounding variable may affect the validity of any
observed change in the dependent variable.
•
Anomalous experimental results are often dismissed as ‘the experiment
hasn’t worked’.
•
There are no wrong results just results that don’t yet have an explanation.
•
A better explanation is that the results may just tell you something you
weren’t expecting or that confounding variables were not taken into
account.
Variables
• In a simple experiment on the effect of temperature on enzyme
activity confounding variables could include:
–
–
–
–
–
enzyme concentration
enzyme source
enzyme surface area
concentration of substrate
time base for measurement.
All of these can and should be easily controlled by careful
measurement and sourcing of experimental materials.
Variables
•
A more complex data set could be encountered when comparing groups of
individuals.
•
If we wish to survey a group to see if music has an effect on ability to
complete a mathematical task there are a myriad of potential variables to
control.
•
If the independent variable is simply music or no music we could
standardise the music type by picking one piece and playing it at constant
volume.
•
Confounding variables within the sample group could include:
– age
– gender
– time of day tested
– mathematical ability.
•
These variables are difficult to control but we could block the groups.
Randomised block design
• Blocking is a practice whereby if confounding variables cannot be
controlled their effects are minimised by selecting control and
experimental groups in which the effects of confounding variables
are equal.
• In our example both the control and experimental groups should
have the same ranges of ages, gender and mathematical abilities.
• If it is not practical to test the control and experimental groups at
the same time of day and day of week the variance of testing times
and numbers of participants should be the same in both the control
and experimental groups.
• Such studies are made even more problematic by subjective
reactions to different music genres.
Variables
A true independent variable?
• We said earlier that simple laboratory-based experiments or even complex
studies of people can have easily defined independent variables, but what if
there is no true independent variable?
•
It may appear that studying the effect of gender on a factor such as
mathematical ability or creativity will have an independent variable in
gender. This is not quite the case.
•
The experimenter may find a statistically significant difference between
gender and the dependent variable but this may simply show correlation,
not causation.
•
The component of gender that may be causative is not being tested
directly: a correlation is detected but causation is not confirmed.
Variables
One or more independent variables?
• This seems to be contrary to what has gone before: how can an experiment
have more than one independent variable? Take, for instance, the effect of
drugs on human physiology.
• Many drugs alter their effect when combined with other therapies. While
the effect of one drug on its own may provide a single independent variable
this is less useful if the drug is usually used in combination with one or more
other drugs.
• Each drug in this case is a factor. When factors are studied in combination
this is called a multifactorial study.
• The challenge to the experimenter is to generate enough test groups to test
every combination of drug factors and doses.
• Simply testing one drug on a laboratory model, ie in vitro, may be elegant and
easily controlled but its relevance may be limited in vivo.
Variables
Different classes of variable exist.
• Discrete variables
Variables with discrete points as possible values.
– Graphed as bar charts defining the variability of a factor amongst
discrete groups, eg the data on the effect of gender on mathematical
ability would consist of two discrete male and female bars with a numerical
value of average mathematical ability on the y-axis.
• Continuous variables
Variables where the scale is continuous and not made up of discrete steps.
– Graphed as line plots, x-y scatter plots or frequency histograms.
– The analysis of the effect of age on hearing range would consist of age as
a scaled x-axis and hearing range in Hz on the y-axis.
In any graphing of experimental data the convention is to graph the
independent variable on the x-axis and the dependent variable on the y-axis.
Data
•
Qualitative data
– The values collected do not imply a numerical order or range of effect.
– The classic iodine solution test for starch is an example of the
generation of qualitative data: starch either is present or it is not
present. No estimate of the quantity of starch is possible without
further experimentation .
•
Quantitative data
– The values collected indicate a numerical value, eg 10 cm3 of O2 is
liberated per minute with 0.1% catalase whereas 1 cm3 is liberated per
minute with 0.01% catalase. The values could be plotted on a line plot
with enzyme concentration on the x-axis and volume of O2 released on
the y-axis. This gives an indication of how much one factor varies with
the other.
•
Ranked data
– This is where a numerical rank, eg 1st, 2nd, 3rd, is applied to a test
situation, eg a group of subjects are asked to rate their preference
for an item or situation on a point scale.
Variables what you need to know
• Controlling and or monitoring confounding
variables, including randomised block
design.
• Variables can be discrete and continuous
variables that can give rise to qualitative,
quantitative or ranked data.
• Due to the complexities of biological
systems, other variables besides the
independent variable may affect the
dependent variable.
Variables what you need to know
• These confounding variables must be held constant
if possible, or at least monitored so that their
effect on the results can be accounted for in the
analysis.
• In cases where confounding variables cannot easily
be controlled, blocks of experimental and control
groups can be distributed in such a way that the
influence of any confounding variable is likely to be
the same across the experimental and control
groups.
• The type of variable being investigated has
consequences for any graphical display or
statistical tests that may be used.