Anova Flashcards
What are the principles of ANOVA.
To test he hypothesis that all means are equal. Compare the variances between the conditions and within the conditions to produce the test statistic known as F (F ratio) If the P value associated with F is significant then it is unlikely that the difference between the means was due to sampling error. Further tests are needed to see if ther is a sig dif between the other conditions. Ie 1and 4
A type 1 error is…?
When the null hypothesis is rejected when it shouldn’t have been. Ie the Means ARE equal after all. Relax people!!
A type 2 error is when ….?
The hull hypothesis is not rejected when it should have been. Ie. There means are actually unequal. There is actually a significant variance in the means of the Dvs for the different levels of the IV
What is the logic of inferential statistics?
To work out the probability of obtaining an effect due to sampling error when the null hypothesis is true.
If you obtain a one tailed p value of 0.02, the equivalent two tailed p value is …?
0.04
What are the pros and cons of using repeated measures?
cons, subjects can get better at a task, or figure out the point of the study and then try to please the researcher. this can be It also might not be possible if for eg, the condition is smoking or sex.
use counter balancing
Pros, Fewer participants are needed, and, it takes acount of individual differnces as a confounding factor, this has improtant implications for Error variance ( ie there is less of it!)
Variance is……?
variance is the average squared deviation (difference) of a data point from the distribution mean
To do this you take the distance of each data point from the mean, square each distance, add them together and then find the average.
The sum of squares is…..?
when you just literally add up all those squared distances from the mean deviation …
The sum of squares is the variance without finding the average bit at the end ….. you just literally add up all those squared distances from the mean deviation …
What are the assumptions about the data that you must make before perfroming an ANOVA?
1) any dependant variables should be measured on at least the INTERVAL SCALE, or the RATIO SCALE
2) The populatioin should be NORMALLY DISTRIBUTED
3) The variances of the populations should be aprox EQUAL (ie, HOMOGENEITY OF VARANCES
4) NO EXTREME SCORES
Partial eta Squared shows…?
shows the percentage of varience that can be accounted for by the different back ground conditions.
written as 0.23 = 23%
or… the variance in the DV accounted for/explained by the IV factor
The Error Variance is…..?
is found within groups and is a result of individual differences and sampling error, NOT a result of the IV!!!!!!!
The IV has a “treatment effect’ BETWEEN groups and on the measurement of their DV
Sphericity.
Sphericity. It is a measure of the correlation (a relationship) between scores in different conditions. This is again because of the fact that the same participant is taking all the conditions available in the study. Therefore, there is a chance that their performance/score in one condition is related to another/others for the factors/IVs tested.
box plot? scatter plot? line graph? histogram?
what are they good for?
Box plot. - -It shows the means as well as the percentile variation around them.
Histogram is better for focusing on the distribution even though the means can be presented.
Scatterplot is for exploring the relationship between two interval/scale variable.
A line graph can show the means, but not spread, but useful for ‘interactions’ (see later learning materials in the module).
when is an independant samples (unrelated) t-test used
for between participants design
what non parametric test can be used if t-test is not? Mann-Whitney!!
Mann-Whitney is…?
Mann-Whitney is a non parametric alternative to the independant t-test
With an interaction:……
One variable behaves differently in each condition of the other variable
A correlation gives the ********** + ************ of the association between two variables. A regression allows one to ******* ***** ***** *** ******* (the one set as the criterion/outcome variable) from scores on another (the one set as the explanatory/predictor variable).
If applied to data obtained from an experiment, it can be used to ‘suggest’ that the scores on one variable might influence the scores on another; i.e., ‘infer’ a causal relationship
direction and strength
predict scores on one variable
What are the two basic conditions or assumptions that should be satisfied for one to use a bivariate linear regression?
First ,the criterion/outcome variable should be measured on an interval/scale. Secondly, the relationship between the two variables should be linear (can be represented by a straight line); otherwise there cannot be a best-fit line, which is how the regression model works; predictions are generated from this line.
In linear regression, the differences between the predicted scores on the best- fit line and the actual scores obtained and plotted are called: _____________
Residuals They are important in evaluating how good (fitted) an regression model is that is based on the best-fit in SPSS.
A regression allows one to ,,,,,,
predict scores on one variable
(the one set as the criterion/outcome variable)
from scores on another
(the one set as the explanatory/predictor variable).
Variance in the Criterion or DV
explained by the regression model is shown as
a percentage by …..
R squared
ie the residual squared, often somethingk like .234
within group variance is caused by….?
invididual differences and sampling error
NOT the IV
Between group varience is caused by …..
Individual differences , sampling error,
AND the TREATMENT EFFECT,
ie the IV
Anova compared…..
to obtain waht stat…
the variance between groups with the variance within groups
Epistemology concerns the theory or
nature of ‘knowledge’ (as in what it is) —whether it is already ‘out there’ to be found and measured (a stance taken by the ‘quantitative’ camp), or to be ‘constructed’ by human meaning-making (the ‘qualitative’ stance).
The positivist stance, which upholds that
‘knowledge’ should be readily and directly observable and thereby measurable in making up the ‘objective’ facts. Therefore, for things that are not readily observable or hard to quantify (points of views, emotions, etc.)—often common interests of psychology—this branch at least tends to gauge aspects of them that Can be measured and objectified (such as behavioural measures or self-reporting scores) to infer about them.
A small sample is typical of research using IPA. Even one participant, as single ‘case study’, is favoured. What theoretical influence underpins this practice?
Idiography refers to the focus on what is ‘distinct’ in individuals (versus what is a general trend among a whole group)
In Qualitative research, If a sample (more than one) is drawn, what should the sample be like?
Homogeneous means participants should share certain significant experience, characteristics or context in which the researcher is interested.
As s/he samples with this purpose in mind, this is also ‘purposive’, as the sample is ‘purposefully homogeneous’.
IPA is concerned with understanding the ‘lived experience’ of a phenomenon (as realist). Discourse analysis is concerned with how language is used to ‘construct’ psychological/ social reality and as such is… (complete with epistemology)
Social constructionist as it is concerned with how people use language to socially ‘construct’ their reality (also ‘relativist’ and ‘critical realist’).
What are the two main strands of Discourse analysis approaches?
State one of the key similarities and one key difference between these strands.
Discursive psychology and Foucauldian discourse analysis (FDA). The former is concerned with how persons use discursive resources to achieve interpersonal objectives in social interactions;
the latter the kinds of resources/ understanding people draw on to make sense of the world and position themselves within.
As both concern the role of language, both tend to use method with ‘texts’, and most commonly semi-structured interviews. Yet while the purpose of DP means that as much detail as possible is transcribed of ‘naturally occurring talk’ (which includes intonation, pauses, interruptions, hesitations, etc.), FDA focuses more on the content of conversations (but both attend to interviewer and interviewee).
When applying one of the DA or FDA approaches in analysis, the researcher pays attention to how people manage their stake and interest in conversations as ‘rhetorical’ strategies on a micro/local level. Name two of such strategies:
The ‘rhetorical’ strategies used by DP can include any of (but not exhaustive to)
construction (how people construct their accounts out of pre-existing linguistic resources as ‘interpretative repertoires’),
variation (how people are inconsistent in conversations),
context (how people vary with context),
function (how people manage stake, interest, blame and their position and ward of accusations, etc.).
Do you want the levene’s test for equality of variences to be significant , ie the independent samples t test.?
no you dont want it to be significant , iie there is NOT a significant difference in the two samples, so you can go a head with the ANOVA,
pHEWWWYY