1 Flashcards
parametric vs non-parametric
Parametric tests involves assumptions about the sample and data, whereas non-parametric tests are ‘assumption free’.
parametric tests are more powerful which means they can detect even a small difference between the IV and DV.
non-parametric tests are simplier and easier to perform.
an example of non-parametric tests include wilcoxon and chi-square
an example of parametric tests include unrelated t-test or ANOVA.
between group design
between group design include 3 or more groups of participants, particpating in different conditions.
an advantage of using a between group design includes decreases the effects of demand characteristcs, this means participants are less likely to guess the hypothesis and change their behaviour according to this. another advantage is there is no order effects as the participants only complete one condition
a disadvantage is the need for more participants.
when using a between participant design, when using a parametric test we must use a unrelated t-test or ANOVA to anaylse the data
within group design
within group design involves each participant being exposed to the same form of manipulation
an advantage of this is it requires less participants,
however disadvantage includes increased demand characteristics as Ps are more likely to guess the hypothesis as they’re involved in all manipulations. another disadvantage includes order effect which means results can be influenced by the order in which participants participate in manipulations
using related t-test ANOVA
type 1 vs type 2 error
type 1 error is when we reject the null hypothesis but it should have been accepted (alpha)
type 2 error is when we accept the null hypothsis but it should have been rejected (beta)
we can control for type 1 error by selecting an approproate p value.
statistical power influences type 2 errror, the higher the statistical power, the lower the chance of a type 2 error, the lower the statistical power, the higher the chance of a type 2 error
one way of managing power and therefore reducing chance of type 2 error is by having a large sample size
correlations
study the relationship between 2 varibles
there is no manipulation, it studies the relationship between 2 natrually occuring events
a disadvantage of correlations is they dont show a causation between the 2 varibles
the results of a correlation can show a positive relationship which demonstrates as one varible increases so does the other, or a negative relationship.
an example of a statistical test involving correlations is the pearon R
post hoc comparisons
‘post’ meaing after refers to a test done after the significance test.
this test should only be done if the statistical test shows a significant difference between the 2 variables.
post hoc tests differ from signifcance tests because they demonstrate where the difference is, rather than if there is just a difference.
popular post hoc tests include binforroni and tukey. binforroni is using for related data and tukey is using for unrelated dataa
levels of measurement
there are 4 levels of measurement; ordinal, interval, ratio, and nominal
nominal data includes data for labels, e.g. sex or age
interval data includes data with a fixed scale e.g. temperature
ratio data is similar to interval data as it is a fixed scale, but it also has an absolute zero which means it doesnt include negative numbers
ordinal data is ranked data, e.g. customer satisfactory scale
1 tailed VS 2 tailed
1 tailed test, also refered to as directional, states the direction of the predicted results, for example coffee will increase heart rate
2 tailed test, also refered to as non-directional, does not state the direction, for example coffee will affect heart rate
1 tailed tests are carried on when there is extentive past research, therefore we can predict the direction of the results.
with 2 tailed test the 5% significant is split into 2.5% which is placed either side of the 2 tails of the critical area. with 1 tailed tests, the 5% significant is place at the predicted tail end of the distribution
significance testing
significance testing determines whether the differences between the IV and DV is significant or due to chance
in significant testing we use a critical value of .05, this means the results have a 5/100 probabily of being significant and a 95/100 probabilty of being due to chance.
anything lower than .05 is signifcant, anything above is not. if it is signifcant we reject the null hypothesis
we use statistical tests to judge the significance of data, using a specific test depends on a varity of circumstances e.g. the type of design used
statistical power
Power is the probability of making a correct decision (to reject the null hypothesis) when the null hypothesis is false.
power can be increased by sample size
effect size also increases sample size
the higher the power, the less chance of a type 2 error, the lower the power, the higher chance of making a type 2 error.
mathmatically, power is 1-beta
effect size
effect size is how much effect the IV has on the DV
the greater the effect size, the greater the power of detecting a signficant difference
effect size differs from significance as it tells you how much of an effect, rather than just stating it has an effect
effect size is important for applied research, for example drug studies wanting to investigate how much an effect the drug has on an illnesses
a large effect size does not mean it is signifcant,.
order effects
Order effects refer to differences in research participants’ responses that result from the order (e.g., first, second, third) in which the experimental materials are presented to them
varience
Variance is a measure of how much values in a data set differ from the mean.
distribution
Normal Distribution– many distributions of naturally occurring variables correspond to this shape.
Normal distribution is symmetrical about each side of central axis, the mean, median, and mode all coincide
However, a lot of data collected in behavioural science are not normally distributed.
Negatively and positively skewed distributions-
Negatively-skewed – the ‘tail’ is to the left
Mean and median are smaller than the mode positively skewed.
Positvely skewed- Tail’ is to the right
Mean and median are larger than the mode
outliers
standardised score
also known as Z scores
tells us where any particular score lies in relation to the mean of a distribution
Indicates whether a particular score is above or below the mean
Also shows how far above or below the mean the score is
E.g. we may wish to know whether a student’s score is better on Test A than on Test B, or how much better or worse the score is compared with other students
