Variability, uncertainty and inferences Flashcards
data collected from humans is varied and this variability presents the challenge that any sample taken from the population is going to vary. What is this called?
sampling variation
the value used to represent the natural variation between participant scores and sample means =
standard deviation
what is standard deviation? (use words standardised and deviation)
it is the standardised measure of the amount of deviation we expect (the variation from the standard measure)
the SD accounts for the _______ variance and the mean difference between experimental conditions accounts for the _______ variance
unsystematic, systematic
what do we assess the effect (systematic variance) against?
how much of the total variance is unsystematic
what gives a graphical representation of variation in scores?
error bars
what do error bar graphs with 95% confidence interval provide?
an estimate of the effects we should find in a population e.g. we are 95% confident that the mean in the population will fall between the upper and lower CI
what do we look for in error bar graphs to be 95% certain?
there is no overlap between groups
what are the reasons that the sample mean will unlikely be the same as the population mean?
sampling error and unsystematic variation
estimate of the deviation between our sample mean and population mean = ?
standard error
what does a large standard error value indicate?
there is lots of variation
unsystematic variance = ? systematic variance = ?
error, effect
when running an inferential test we are looking at how much ____ and ____ there is
error (unsystematic variation), effect (systematic variation)
total amount of variance =
total unsystematic V (due to ppts differing) + total systematic V (due to experimental manipulation)
if there is more unsystematic variation what is the likely effect size and is there an effect?
unlikely to be an effect in pop > small effect size
uncertainty = ______ / ______
effect / error (mean difference/standard error) THIS PROVIDES THE T VALUE
what does the t test compare?
estimate of error against the effect (effect/error=t value)
from what value can we calculate the p value?
from the t value we can calculate the p value
whether we are likely to see an effect in the sample if there is no effect in the population =
p value
what does a p value of less than 0.05 mean?
there is a very small chance of finding an effect in the sample if there is no effect in the population > this means there is a significant effect!!
why do we conduct inferential statistics?
we need to generalise the effect to the population and make inferences about the effect in the population
why do we not test qualitative data using inferential stats?
the focus is on subjective experiences
why can inferential stats result in flawed estimates of error?
we haven’t tested the whole population so we don’t know the population mean and SD, and small samples won’t be representative of the population
size of effect =
mean difference
what are the 3 measures of effect?
statistical significance, 95% confidence interval, effect size
standardised measure of how important the effect is that you find in the sample (magnitude of effect) = ?
effect size
what do the 3 measures of effect individually measure?
p value = whether or not there is an effect, CI = estimate of effect you may have in the pop, effect size = compares the size of the effect
what is the specific effect size for t tests?
cohen’s D
lots of overlap = small or large cohen’s d?
small value > not much difference > no big effect
low overlap of cohen’s d?
high value > more difference > more of an effect
in an ANOVA what are independent variables called?
factors (and groups are called levels)
differences found due to the fact that ppts did different things in each of the groups =
the treatment effect
1 IV with 2 conditions = ? 1 IV with 3 conditions = ?
independent t test, independent 1 way ANOVA
differences between groups assumed due to the manipulation we have applied to the groups =
between group variance
how is the grand mean calculated?
the mean of each group mean
the between group variance is the difference between the ______ mean and each _____ group
grand, group
how do you calculate between group differences?
take away each group mean from the grand mean, add these values together, square this value, multiple by n, divide by df
what do between group differences include?
effect (differences due to manipulation), error (individual differences, sampling error)
how is within group variance different from between group variance?
between = systematic variance (effect) compared to within = unsystematic variance > how much error we have in our effect. this error is due to individual differences and sampling error
between groups variance (effect) / within groups variance (error) gives a value that represents the effect. what is this value called?
F ratio
value that provides an indication of the size of our effect taking the error into account =
F ratio
if F is ______ than 1 you have more effect than error
greater
if F is ______ than 1 you have more error than effect
smaller
the np2 value is converted into a ______
proportion e.g. 0.45 > 45%
what does np2 tell us?
what proportion of our total variance is due to the treatment effect
variability in data results in ______
uncertainty
to compare effect sizes what do we need to use?
standardised measures of effect size
what does a 1 way ANOVA tell you?
whether there are differences between your group (but not which groups differ or by how much)
what does it mean the ANOVA outcome is not significant?
there are no differences between groups