Lecture 4 Flashcards
what does prediction in RQs in psychology research has involve?
using knowledge about one or more constructs (where people stand in those constructs) to indicate people’s standings on another constructs.
aka:
how strongly is one construct RELATED to other construct?
is prediction the same as explanation or cause????
prediction is used as INDICATION.
how well can we use people’s scores on IV to indicate DV without thinking that iV causes DV.
eg: cow’s utter size may indicate the amount of milk produced. but this may not be the cause (IV). maybe the breed???
eg: barometer needle movement is not caused by the weather outside directly. but we the needle indicates!!!
‘explain’ / ‘account for’ can be used. but in a descriptive sense, not causal sense
characteristics of RQ for prediction
same 4 as ASSOCIATION RQ:
- question mark
- all constructs and variables
- population
- driving verb –> predict
additional characteristics
- constructs have different role and function
- meaning and focus of RQ depends on which variable is IV and which is DV
what are the different role and function of constructs
- one construct is being predicted (DV) –> Y variable
- other constructs are predictors (IV) –> X variable
what is the difference between RQ association and RQ prediction?
in association RQ, the wording and structure (if A is associated with B = B is associated with A)
in prediction RQ, the structure and wording matters
variation
total amount of variability in a distribution. measured by sum of squared deviation scores (SS), dependent of sample size (larger as sample size gets larger)
SS is like a deviation score
variance
average of variation. hence average of sum of squared deviation scores. variance would be independent of sample size
standard deviation
basically the deviation of raw score to its mean. eg: in IQ —> 115 - 100 = 15 (deviation is 15)
square root of variance. expressed in same metric as means, or observed scores
what is the main differences between variation, variance and standard deviation
standard deviation is square root of variance. variance is average of variation. and variation is total sum of variability in distribution.
why squared deviation (variance and variation) is utilised in analysis of data?
implication in terms of area (in graphs) of variation from mean.
correlation vs regression distinguishing feature
correlation defines symmetric relationship and regression defines asymmetric relationship
what is symmetric relationship
one where the variables have the same ROLE
what is asymmetric relationship
one where the variables have a different role
correlation calculation
STANDARDISED measure of the strength and direction of the association between scores on 2 variables (involves z score)
involves CROSS PRODUCT of z values
each of the cross product (of each individual) is summed up and divided by n-1
covariance calculation
unstandardised measure of strength and direction of the association between scores on 2 variables (does NOT involve z score).
involves CROSS PRODUCT deviation score (observed score subtracted by mean score)
what is the difference between correlation and covariance
correlation –> standardised measure. cross product involving z values. covariance divided by the all magnitude information is thrown away (since it is basically covariance divided by standard deviation)
covariance –> unstandardised measure. cross product involving deviation values. has magnitude, and size of relationship
how does correlation and covariance relate to each other?
correlation is covariance of x and y variables (multiply) divided by standard deviation of x AND y variables (multiply)
how is a correlation related to regression line
regression line is LINE OF BEST FIT. it is a product of correlation of 2 variables and their standard deviations (see formula)
in what kind of case would regression line be the same as correlation?
when the standard deviation is the same in the 2 variables.
could you have the same correlation in 2 variables but different regression?
yes. this is because regression takes into account the standard deviation in each of the variables.
what does slope mean in a linear function
it is the RATIO of differences between y variables over x variables
how to interpret slope?
slope represents the result change on y axis (DV) for one unit increase in the x axis (IV)
or one unit decrease in DV x axis result change on the IV y axis
full regression equation vs regression model equation difference
full regression equation involves residual (error) scores. regression model equation involves predicted scores
what is residual scores
the difference between the observed scores and the predicted scores. unexplained by equation
what does it mean to have small residual scores
means that error is lesser and hence predicted scores would be larger.
hence the predicted scores are closer to the observed scores which is desirable
how do values of intercept and slope in regression equation relate to residual?
in regression equation,
the intercept and slope in regression model suggests the predicted value of Y (DV), in which when we subtract the observed data with this predicted value would give us the residual value in FULL REGRESSION EQUATION.
what is Ordinary Least Squares (OLS) estimator for?
a way to obtain, given sample data on DV, we can find the value for intercept and slope in IV (regression line) such that it will MINIMISE the sum of squared residuals
