Psychology Statistic Note
Psychology Statistic Note
Table of Contents
1.0 Basic concepts .............................. 1
1.1 ...............................Why study statistics?
..................................................................................... 1
2
1.0 Basic concepts
1
3) SPSS syntax editor 1.12 Data type
Command, recommended (.sps)
1) Raw data
1.9 Data preparation 2) Frequency distribution
3) Grouped frequency distribution
1) Define variable
SPSS data editor 1.13 Median deviation (MD)
The average of the absolute differences between the
2) Enter data
data points and their mean.
Data view
How dispersed the data
Range
3) Run frequency
Variance: squared of MD
Check wrong data
SD: measure of deviation from mean
Dispersion is normal if 3 SD above or below mean
4) Data editing
Clean up data
1.14 Hypothesis
5) Realiability test Types of error
Testing internal consistency of instrument (alpha
cronbach) Type I Type II
Reject H0 when its true Fail to reject H0 when its
6) Data transformation false
Compute (create new var)
Recode (categorised data & change value system) Reject H0
Reject H0 at 0.1, sure reject at 0.5
7) Exploratory data analysis Reject H0 at two tailed, sure reject one tailed
Normality
Sig value of kolmogrov-smirnov @ shapiro wilk ≥ Reject H0 at N=20, sure reject N=50
α or skewness between -2 - +2
Normal distribution
1.10 Measures of center in a numerical data set
1) Right skewed distribution
1) Mode: the most frequent value in a data set. Not Most of the data falls to the left of the mean
influence by extreme value Mode<median<mean
2) Median: the middle value in a data set. Not influence 2) Left skewed distribution
by extreme value Most of the data falls to the right of the mean
Mean<median<mode
3) Mean: the average value of data set. Much affected
by the presence of extreme value -z > + = smaller portion
-z > - = smaller portion
-z < - = larger
1.11 Scale of measurement
-z > - = larger
2
2.0 T-test Decision
𝛴𝑦 Effect size
ȳ=(𝑛) <0.2 – trivial
0.2 – small
0.5 – medium
(𝛴𝑦)2 0.8 – large
𝛴𝑦2−
Sy = √ 𝑛
𝑛−1
tcal
ȳ − µ0
t= 𝑠𝑦
√𝑛
tcritical
𝛼
t , df (n-1) (two tailed)
2
∝
t , df (n-2) (one tailed)
3
2.2 Paired sample t-test
Conclusion
Purpose/requirement
Use to compare diff btw pre & post test Reject Ho: if there is significant mean difference between
Scores of the pre & post meet the assumption of pre & post test at α level of sig
normality (parametric) (sig value of kolmogrov-
smirnov @ shapiro wilk ≥ ∝ @ skewness btw -2 – 2+ Fail to reject Ho: if there is no significant mean difference
between pre & post test at α level of sig
Variables
DV – I/R
t-critical IV – N/0
-sda
Hypothesis
Decision
CI Ho: µ1 = µ2
Reject Ho: 0 outside CI
+
đ− tα,df (sȳ)(only relevant for two tailed) Ha: µ1 ≠ µ2
:µ1 > µ2
:µ1 < u2
Effect size (stats)
-sda-
Calculation
Eta-square (practical)
𝑡2 Fcal
𝑡 2 +𝑛−1
𝑠2 𝑙𝑎𝑟𝑔𝑒
F=
𝑠2 𝑠𝑚𝑎𝑙𝑙
𝑛𝐿−1
Fcritical = F
𝑛𝑆−1
tcal
4
unequal variance Eta-square (practical)
ý1−ý2
t= 2 2 𝑡2
√ 𝑠1 + 𝑠2 2
𝑛1 𝑛2
n = 𝑡 2 =𝑛1=𝑛2−2
equal variance
Decision
Manual
SPSS
Manual
SPSS
Reject Ho : sig-t ≤ α
CI
+
đ − tα,df, (sx̄1- x̄s2)
+ 𝑠12 𝑠22
đ − tα,df, √ 𝑛1 + 𝑛2
𝑛1+𝑛2
d=√ 𝑛1𝑛2
5
3.0 ANNOVA
Decision
Purpose / requirement
Manual: Reject Ho if Fcal > Fcritical
To compare differences btw more than two group SPSS: reject Ho if sig-F ≤ α
(normally in exam 3)
Assumptions must be:
Parametric
𝑀𝑆𝑊
Sig value of lavene statistics > α
𝑘
T/f, data distribution has met the assumption of qk,dfw,α√ 1
𝛴𝑛
homogeneity of variance
Involves ΣY and its total
Ȳ (use if post hoc apply) I J MD
n&N 1 2 Ȳ1-Ȳ2
sum of squares (SST, SSB, SSW) 1 3 Ȳ1-Ȳ3
df (dfT, dfB, dfW) 2 3 Ȳ2-Ȳ3
Mean squares (MSB, MSW)
Conclusion
Summary of Anova table
Reject Ho if there is a sig diff in DV among groups at
Source SS df MS F α level of sig
Btw SSB K-1 MSB Fail to reject Ho if there is no sig diff in DV among
Within SSW N-K MSW groups at α level of sig
Total SST N-1 Post Hoc: there is sig dif btw gp? & gp? And btw gp?
And gp?, but there is no sig diff btw gp? And gp? At
α level of sig
Eta-square: about…..% variance in DV is explained by
Variables groups @ the effect size of the diff is considered to
DV – I/R be small S/M/L
IV – N/O
Calculation
Fcal:
𝑀𝑆𝐵
F=
𝑀𝑆𝑊
Fcritical:
𝑑𝑓𝑏
F ,α
𝑑𝑓𝑤
Eta square:
<0.10 – trivia
0.10 – small
0.25 – moderate
0.40 – large
6
4.0 CHI-SQUARE
Use Guilford’s rule of thumb
Purpose / requirement
< 0.2 negligible relationship
1. Goodness of fit 0.2 – 0.4 – Low
2. Test of independence (final exam) 0.4 – 0.7 – moderate
3. Test of homogeneity 0.7 – 0.9 – high
> 0.9 – very high
For test of independence:
2 variables (IV & DV) Conclusion
To test relationship btw two categorical variables
and determine the strength of the relationship Reject Ho: DV is significantly dependent on IV at α
Involve contingency tables: level of sig
(Column (y), Rows (x)) Fail to reject Ho: DV is not significantly dependent
on IV at α level of sig
X Y RT
Symmetric measures (Measure of association):
Low Mod High
A O O O
1. Phi Coefficient (2x2 contingency table)
B O O O
2. Contingency coefficient (>2)
CT GT
3. Cramer V coefficient (>2)
(𝑅𝑇)(𝐶𝑇)
E=
𝐺𝑇
(𝑂 − 𝐸)2
O E (O-E) (O-E)2
𝐸
Variables
DV – N/O
IV – N/O
Hypotheses
Ho: DV is independent of IV
Ha: DV is dependent on IV
Calculation
X2Cal
2
2 (𝑂−𝐸)
X= , (always (+))
𝐸
X2 critical
(Refer table X2 distribution)
df = (R-1) (C-1)
Decision
Variables
DV – O
IV – O
Or
DV – I/R
IV – I/R
Hypotheses
Ho: Ps = 0
Ha: Ps ≠ 0 (two-tailed final exam)
Calculation
Descriptive part
6𝛴𝑑 2
rs = 1 - 𝑛(𝑛2 −1)
dpr rs (either (-) or (+) and
refer to Guildford
describe nature of relationship =
positivity/negative & n/i/m/h/vh relationship
between IV & DV
8
6.0 PEARSON PRODUCT-MOMENT CORRELATION Inferential:
tcal:
Purpose / requirement 𝑟−𝑝
t= 2
√1−𝑟
1. to determine relationship btw two variables (IV 𝑛−2
& DV)
2. Parametric (assumption of normality)
3. Both IV/DV meet the assumption of normality tcritical: (refer t-table, α bahagi dua)
where sig value of kolmogrov @ Shapiro ≥ α @ 𝛼
skewness btw -2 to +2
t 2 , 𝑑𝑓 (𝑛 − 2)
4. Range -1 to +1
Summary: Decision
Ho: p = 0
Ha: p ≠ 0
P>0
P<0
Descriptive part:
(𝛴𝑋)(𝛴𝑌)
𝛴𝑋𝑌− 𝑛
r= 2 2
√(𝛴𝑋 2 −(𝛴𝑋) )(𝛴𝑌 2 − (𝛴𝑌) )
𝑛 𝑛
9
7.0 SIMPLE LINEAR REGRESSION 𝑑𝑓𝑅
F 𝑑𝑓𝐸 , α
Purpose / requirement
Slope, (t)
1. To determine relationship btw two variables (IV &
DV)
t cal
2. To make prediction of DV based on IV
𝑏1−𝐵1
3. Extension of Pearson correlation t=
4. Must meet assumption where the data is parametric 𝑀𝑆𝐸
√
(normally distributed) 𝑆𝑆𝑋
𝑆𝑆𝑅
Calculation
R2 = 𝑆𝑆𝑇
Regression model 2. Multiple correlation coefficient (0-1), R = √𝑅 2
F cal
𝑀𝑆𝑅
F = 𝑀𝑆𝐸
F critical
10
Conclusion
11