Anomalies can be identified. Vice versa
to check if a result is anomalous
REPEAT experiment
More reliable mean
check
results
concordant
(similar)
Minimise effects of variation in data
Large sample size
Representative
Allows statistical tests to be conducted
data for last 3 tubes more reliable than for 1st 3. Practice in
judging when product appears/approximate time of curd
appearance known, so able to watch more carefully at that time
Accuracy: how close
mean is to true value
Reliable: little variation in results,
approx same. Repeatable- if
experiment repeated, similar results
CONFOUNDING (CONTROL/KEEP SAME) VARIABLES
TEMPERATURE
If not controlled in
investigation, how to find
how it could've influenced
results
Monitor room temp thru out/ at
beginning & end of experiment
Keeping it constant
Water bath: more temp variation in air than water/it
keeps test tubes at constant temp
Temp of water bath as reliable as possible:
measure temp of bath at beginning & end of
reaction period (or more) to assess effect of
any temp changes during reaction/to show
there was no/little variation in temperature
EQUILIBRATION
Leave tube 1 in water bath for 5 mins before adding enzyme. To
equilibrate /for solution to reach the temp of bath
Monitor with
thermometer
use data-logger to
record temp
Adjust with hot/cold water
pH
Keep constant: use buffer
If 2 enzymes with diff
optimum pH, then use
2 buffers
Other C
variables/precautions/standardisations
Volume/mass
Concentration
Standardise type/category
Time: between readings;duration;
mixing
All controlled apart from 1-independent variable
Allows comparison of results/reproduce-able (diff techniques but
same results)/repeatable (exact same technique) by someone else
Investigating 1 variable only
No./rate of repeats
Read meniscus at eye level
Same end point each time
LIMITATIONS (UNCONTROLLABLE
CONFOUND'N VARIABLES)
Difficult to time/measure accurately
Diffi to keep temp constant
Diffi to decide when product present /formed--subjective
Everyone's judgement diff
Theres many variations (creeping in..) in 1 'type'
CONTROL EXPERIMENT (TO COMPARE with effect of independent)
YES! shows independent variable causing result , not any
other variable.
NO! experiment to determine
whether result due to independent,
nothing else
To conduct treat everything same
but eliminate/replace independent
used in experiment e.g. replace
alcohol with H2O
%change/increase in mass/volume/ ratios: COMPARISON as initial diff
Analyse class results, student calculates % increase in
pulse rate/student. YES: result comparison (as resting pulse
maybe diff so accounting all results); NO: exercise not standardised so can't
compare
RANDOM
TESTING
Divide into 2
equal groups
keep control & experimental group
'same' for 1 factor e.g. age, sex,
mass
Or divide randomly
which avoid bias/allows statistical analysis
STANDARD DEVIATION
BARS suggest about difference..?
If SD bars don't overlap difference
significant/more variation in one than the other
Overlapping SD results due to chance
WHAT it shows?
SPREAD of measurements around mean/accounts all
measurements
how reliable
measurements
are/mean is
Even without same sample size for each set of measurements, can calculate SD
Large SD: greater spread of data. Smaller SD: bigger central tendency & data concentrated around mean.
ACCURACY
measuring time for larger
volume drain using larger
syringe: MORE ACCURATE as
time to drain longer / less effect
from stopping / starting timer /
smaller percentage error;
using digital timer that measures to 0.0001 sec
No/little difference) as effect of stopping /
starting timer still has big effect
Judging when all solution
has drained subjective / not as
accurate as timer
Measuring pulse for shorter than 20 secs=>unreliable
results. incomplete beats/cycles; Small difference in
measurements can produce large difference in pulse rate
when multiplied;
Length no
shorter than
instructed
Errors have relatively larger
effect/length maybe not long
enough to give measurable
change
GRAPHS
Join points on graph
straight lines NOT curve; no
INTERMEDIATE values; can't predict
rate between plotted values
Student said optimum pH for
lactase was pH 6.0. Not valid as
peak could be anywhere pH 5.5 -
7.5 / should've repeated
experiment at appropriate pH
Should curve start from 0? NO: Curve shouldn't extend beyond data
collected. YES at origin no product formation (in other cases control
necessary to deduce this)