PSYC3001 – Tips for Making up Data 心理学 assignment代写
100%原创包过,高质代写&免费提供Turnitin报告--24小时客服QQ&微信：120591129
PSYC3001 – Tips for Making up Data 心理学 assignment代写
UNSW PSYC3001 – Tips for Making up Data for Assignment 2 – Dr Melanie Gleitzman
1
Tips for making up data for PSYC3001 Assignment 2 2017
You have been asked to make up data for a 3 x 4 design and carry out a planned contrasts analysis in PSY.
However, rather than enter your data directly into PSY, it will save you time to use either SPSS or Excel to
create your data set, because these programs give you greater control over changing the characteristics
of your data to suit your assignment. Whichever you use you will need to import your data, including
group coding, into PSY.
The SPSS instructions below are for a 2 x 2 design with n = 5 Ps per cell.
In order to conveying the impact on data of changing between cells variability and/or within cells
variability, the discussion below refers to whether data reflect A, B and AB effects. In this case,
the SPSS ANOVA summary table is commensurate with PSY output for A, B and AB contrasts for 2
x 2 design. [NOTE: Your assignment asks for planned contrasts and not overall tests.]
Generating DATA:
Step 1: Once you have chosen your factors and levels (and DV), think about the story you want your data
to tell. A good place to start with this is to think of what sort AB interaction effect you want your data to
show.
PSYC3001 – Tips for Making up Data 心理学 assignment代写
Step 2: Think of a pattern of cell means that will convey your AB interaction effect.
Step 3: In SPSS (or Excel), create the variables A, B, GROUP, MEAN, ERROR and input appropriate values.
ERROR = within cell individual difference scores (the values above are a ‘quick and easy’ way of injecting
individual difference into a data set).
Use COMPUTE to create DV = MEAN + ERROR.
A = levels of factor A (1,2).
B = levels of factor B (1,2).
Note the order of these
values indicates which
rows refer to which cells in
the design. eg A = 1, B = 1
indicates cell a1b1; A = 1,
B = 2 indicates cell a1b2,
and so on.
GROUP = 1, 2, 3 and 4,
corresponding to the 4
cells: a1b1, a1b2, a2b1,
a2b2, respectively.
MEAN = cell mean (you
input whatever values you
want) corresponding to
the 4 cells: a1b1, a1b2,
UNSW PSYC3001 – Tips for Making up Data for Assignment 2 – Dr Melanie Gleitzman
2
For the above data, the Two‐Way ANOVA Summary table indicates B and AB are significant, but not A:
Tests of Between-Subjects Effects
Dependent Variable: DV
Source Sum of Squares df Mean Square F Sig.
A 5.000 1 5.000 2.000 .176
B 45.000 1 45.000 18.000 .001
A * B 125.000 1 125.000 50.000 .000
Error 40.000 16 2.500
Corrected Total 215.000 19
Step 4: You may need to modify your data if you do not get the significant effects that you are after.
What if your data do not generate the desired significant effects?
Now suppose instead of the above cell means, the MEAN values were as below (one‐third the size of
those above), with the ERROR values the same as above:
The Summary Table shows AB significant, but not A or B.
Tests of Between-Subjects Effects
Dependent Variable: DV
Source Sum of Squares df Mean Square F Sig.
A .556 1 .556 .222 .644
B 5.000 1 5.000 2.000 .176
A * B 13.889 1 13.889 5.556 .031
Error 40.000 16 2.500
Corrected Total 59.444 19
UNSW PSYC3001 – Tips for Making up Data for Assignment 2 – Dr Melanie Gleitzman
3
Note that the SSE and MSE is same as first example above. Do you understand why?
In this case, the amount of within‐cells individual difference is too large for the between‐cells variation,
OR another way of saying this is that the metric of the DV (where cell means vary between 6 and 8.67) is
not appropriate for the metric of the ERROR scores. To inject more between‐cells variation into the data,
the pattern of means can be ‘expanded’ as per example 1 above, or the ERROR scores can be contracted
(eg halve the ERROR scores].
Halving the ERROR scores (ie values of 1, .5, 0, ‐.5, ‐1 instead of 2, 1, 0 ‐1, ‐2) generates the following
summary table:
Tests of Between-Subjects Effects
Dependent Variable: DV
Source
Type III Sum of
Squares df Mean Square F Sig.
A .556 1 .556 .889 .360
B 5.000 1 5.000 8.000 .012
A * B 13.889 1 13.889 22.222 .000
Error 10.000 16 .625
Corrected Total 29.444 19
Note that halving the magnitude of the ERROR scores decreases SSE from 40 to 10. The smaller MSE
leads to significant Fs for B and AB.
What if your data generate ANOVA Fs that are too large (>100)?
The same principles apply as for the above cases, but in the opposite way. Rather than wanting to
increase the spread of cell means or decrease the within‐cells variability you want to do the opposite.
If your ANOVA F is too large, this means your ERROR scores are not variable enough for your pattern of
cell means OR your pattern of cell means are too spread out given the within‐cells variability.
Either increase your ERROR scores (make them more discrepant from 0, eg 4, 2, 0, ‐2, ‐4), OR decrease
the range of your cell means.
To import your data into PSY
Data must be ordered Group 1 through 4. You can use ‘save as’, and select variables Group and DV, and
save file as .dat. Then copy and paste .dat file into PSY, below heading [DATA], and add your contrasts.
Or, copy and past Group and DV columns directly from SPSS into PSY.
For J x K design
You can use the above method to give you an indication of whether your data reflect A, B and AB effects.
Of course, you will need to run your planned contrasts in PSY to know whether your contrasts are
significant or not. However, if you find you do need to modify your data (and most students will need to
do so), it will be easier to do the modification in SPSS (or Excel), than in PSY.
PSYC3001 – Tips for Making up Data 心理学 assignment代写
