Test 3:Main Points

Question 1

Generally, conducting a correlation with nominal (categorical) data will produce a meaningless correlation.
However, a correlation can be meaningful when using categorical variables if?

Answer 1

Normally, categorical correlations are meaningless and results reflect the way in which the data was arbitrarily coded into categories ad, not an actual association between two variables.

However, a correlation can be meaningful if the categorical variable is dichotomous (can only be one of two values) and correlated with a continuous variable.

In this instance only will finding a correlation be meaningful with nominal data.
. r (218) = .16, p = .018. what does this mean?

The positive correlation is directed towards females because they were arbitrarily coded as having a relatively higher standard than males 1 vs. 0, and this indicates women have higher levels of anxiety than males.

In contrast, a negative correlation is directed towards males and indicates that males have higher levels of anxiety than females.

Question 2

What are the two key pieces of information in a correlation equation?

Answer 2

(A) Pearsons r 
     coefficient:
     Tells us the strength 
     and direction of the 
     association between 
     X and Y.

(B) P-value:
     to determine if the 
     correlation is 
     statistically 
     significant.

Question 3

What are the 5 key pieces of information in a regression equation?

Answer 3

x-values= the actual 
       observed values that 
       participants generate 
       from the survey on 
       the (IV) the predictor 
       variable.
y-values= the actual 
      observed value that 
      participants 
      generated in their 
      survey for the (DV)   
      or 
      outcome variable. 
Constant = y-intercept 
     i.e. where the slope 
     cuts the y-axis. 
     Changing the  
     constant value will 
     move the slope up or 
     down the y-axis!
b1 = the unstandardised 
        regression 
        coefficient or b. 
        Is the slope of the 
        regression line and 
        thus, tells us the 
        strength and 
        direction of the 
        correlation.
e = error or residual (the 
       difference in value 
       between the 
       predicted y value (regression line)
       and the actual observed y-value (diamond on the 
       graph)

The equation calculates our best prediction of y-value (DV) by multiplying the b value by every IV score for that particular individual, add the constant value to find the best estimate of the predicted y-value for that individual.

Question 4

what is the unstandardized estimate?

Answer 4

Unstandardized regression coefficient

> The Unstandardized regression coefficient +
Standard error is how you calculate the beta weight
(the standardised regression coefficient).

Question 5

Regression Equation:

Answer 5

y= constant - b1 (x) + e

In other words, we multiply b1 by every score of X (IV) in the dataset. Then we add the value of the constant. This value produces the best estimate of the predicted y value (for that participant)

Question 6

why is asking what is a “good” test-retest reliability score, overtime, a stupid question?

Answer 6

because it depends on the measure. different variables have different levels of stability and this will determine if we as researchers want to find stability in our scale.

For example,
Demographic Information is highly stable over time and we would expect to see a test-retest correlation of .90.

In constant, physiological measures has intermediate stability over time and a good test-retest reliability score would be .70

Finally, mood measures are notoriously unstable over time and we would expect a test-retest reliability score of around .50.

Question 7

If test-retest reliability is poor this may indicate that either:

Answer 7

the measurement is psychometrically poor or that the variable is inherently unstable.

Question 8

*Three Key Characteristics of Regressions: (differences from correlations)

Answer 8

(A) Single headed arrows i.e. directional
(B) Prediction goes from left to right i.e. IV[s] predicts
DV
(C) There is a single DV (constraint on this type of
analysis)

Question 9

What does a regression equation mean?

Answer 9

o A regression equation is the best estimate of how to predict y-values (DV) from x-values (IV).
o In context this equation tells us that we multiply -1.03 (b1) by every observed score for the predictor variable (x i.e. rumination) then we add the “constant” i.e. 7.00 = this equation provides us with the best estimate of the y-value (i.e. subjective happiness) for that individual.

o Note: this is an estimate. Therefore, the value obtained has a degree of error to it- where it will not suit every participant.