SMA TỔNG HỢP.xlsx - Table Summary

You might also like

You are on page 1of 7

QUESTION KEYWORDS PURPOSE HOW TO INTEPRET (GENERAL)

Interpreting IV and DV are the same:


Use the reliability - Cronbach's Alpha: it is the best in the range [0.5;0.95]
test, specifically the - If there is no item deleted having higher Crobach's Alpha -> KEEP ALL QUESTIONS FOR
How can you be Cronbach’s alpha for THIS VARIABLE
sure that your determining the
Test reliability, how
Cronbach’s groups of reliability or - If item deleted has higher Cronbach's Alpha -> Delete this item to see the new result
Reliability test focus of your
alpha questions internal consistency + If new result has item deleted having higher Cronbach's Alpha -> Continue delete and see
question
measure the same of a set of questions. the result
constructs? (Test how focus of + The new Cronbach's Alpha is the highest -> STOP
your questions) + STOP if only 2 questions left
- If the highest Cronbach's Alpha < 0.5 -> KEEP ALL QUESTIONS FOR THIS VARIABLE

- Interpret IV:
KMO & Bartlett's Test table: KMO from 0.5 to 1, sig < 5% -> it is useful for us to run factor
analysis.

Total Variance Explained table: the maximum variables are abc but SPSS suggests only xyz
variables representing all variables. After component xyz the number of questions that can be
represented is lower than 1. In conclusion, there will be xyz factors, which means that they can
explain at least 1 variable. The cumulative % = ??% -> the model can explain ??% DV
How did you
reduce your data A technique used for Rotated Component Matrix: 1 column = 1 factor. If < 0.5 -> do not care. Name each factor by
Reduce from a big
from 20+ data reducing and variable has greater power (nhiều variable + số lớn). If a variable is represented by more than 1
Factor set of variables to
variables into a summarizing. from factor (or no factor) -> delete and run again.
analysis smaller and useful
smaller and more big observations tin
one
useful set of to smaller observation - Interpret DV:
variables? KMO & Bartlett's Test table: KMO from 0.5 to 1, sig < 5% -> it is useful for us to run factor
analysis.

Total variance explained table: SPSS suggests that only 1 factor can represent a dependent
variable, which is good.

Component matrix table: With 1 factor, the relationship between SI1, SI2 is good

Between 2 levels
of nonmetric
variable, who Look at Independent sample test, sig of Levene's test
have higher TH1: Sig < 0.05 => look at sig T- test Equal variance not assumed
INDEPENDENT - Sig T-test < 0.05 => there is significant difference in the mean between 2 independent variable
VARIABLE --> Look at 1st table: higher mean =higher IV
toward DV - Sig T-test > 0.05 => there is no significant difference in the mean between 2 independent
(Ex: Between The mean of 2 Comparing the means variable
Independent income lower different samples of 1 variable between
than 2M and but 1 variable 2 different samples TH2: Sig >= 0.05 => look at sig T- test Equal variance assumed
between 2-5M, - Sig T-test < 0.05 => there is significant difference in the mean between 2 independent variable
who have higher --> Look at 1st table: higher mean =higher IV
attitude toward - Sig T-test > 0.05 => there is no significant difference in the mean between 2 independent
the subscription variable
of movie
streaming
T-test services)
T-test QUESTION KEYWORDS PURPOSE HOW TO INTEPRET (GENERAL)
Which is higher
between 2
Independent
variables
(Ex: Which is Do not look at the first table.
higher, Attitudes Never look at the second table
towards the Comparing the mean Look at the 3rd table Paired Samples test:
means of 2 different
subscription of of 2 different Compare sig in 3rd table (Paired samples test table)
Paired independent
movie streaming variables (in 1 - Sig < 0.05 → good sig → different perception on 2 variables (1)
variables
services and sample) - Sig > 0.05 → no good sig → equal perception on 2 variables --> END
Involvement in
streaming media Look at the 1st table: Compare mean, which is higher has higher perception
service?
OR Which is
higher, Moral and
Risk?)
H0: No association between 2 variables
EX: Is there
H1: There is an association between 2 variables
relationship
between income Relationship
Relationship between Do not look at 1st table
and university between nonmetric
Cross nonmetric variables - Do not look at 2nd table because it is just about your sample not a whole population
you choose? variables -
Tabulation gender/income/occup
People living in gender/income/occu
ation Look at 3rd table: look at asymptotic significance of Pearson Chi-square
different places pation
sig <= 5% → reject H0
have different
sig > 5% → accept H0 → there is no association between variables
income?

ANOVA table
We care sig of between group (bc it reflects the relationship between income&facility)
"The relationship + Between group:
One-way ANOVA is
of X (1
used for examining
nonmetric) and Y 1/ Sig of between group <0.05 → good sig (income really leads to facility)
the differences in the
(1 metric)" or 1/ Sig of between group >0.05 → X don't affect/ no relationship Y; --> whatever the age the
relationship mean values of the
"Does income customer is, satisfaction are the same --> DỪNG INTERPRET
between 1 dependent variable
affect/lead to 2/ lấy sum of squares của between group/ total * 100 → % of non-metric affect metric
ANOVA One-way nonmetric variable (metric) associated
facility"?
and 1 metric with 1 nonmetric
+ Within group: sum of squares lớn bc millions thing affect/lead to perception in facility outside
variable (more than 3
"Which age group of income.
samples/values)
have highest
independent
consumer POSTHOC TEST table (want to compare each level of (non-metric) lead to (metric))
variable.
satisfaction" + Big sig (>0.05) → no difference in IV between level 1 and 2 OR have same perception
between level 1 and level 2.
+ Low sig (<0.05) → have a difference in IV between level 1 and level 2.
QUESTION KEYWORDS PURPOSE HOW TO INTEPRET (GENERAL)

BẢNG ĐẦU TIÊN KHÔNG NHÌN


"Between X1
TEST OF BETWEEN-SUBJECT EFFECT Table
(nonmetric 1) and
X2 (nonmetric 2),
Two-way ANOVA is + Sig of interacting power: multiplied 2 nonmetric variables < 5% → interacting power between
which lead/ have
used for examining 2 nonmetric variables → END INTERPRET
affection more to
- more affected or the differences in the
Y"
lead mean values of 1 + Sig of interacting power: multiplied 2 nonmetric variables > 5% → No interactive power
ANOVA Two-way - mean values of 1 metric dependent between 2 nonmetric variables → talk about 2 nonmetric variables individually
"Y is more
metric and 2 variable associated → If X1 changes then the relationship between X2 and Y doesn't change.
affected by X1
nonmetric variables with 2 nonmetric
than X2?"
independent + Sig of each non-metric:
Attitude is more
variables. <5% → have affect on Y
affected by
>5% → don't have effect on Y/ don't have relationship with Y
geographic than
income?
+ Cả 2 sig của non-metric đều <5% → Compare Type III Sum of squares → higher → stronger
effect (NẾU CẢ 2 SIG CÙNG NHỎ HƠN 5%)

BẢNG ĐẦU TIỀN KHÔNG NHÌN

TEST OF BETWEEN-SUBJECT EFFECT table

Control variable TH1 : + Compare sig of nonmetric vs sig of metric: control variable (chưa chạy ANOVA
Used to remove
helps controlling two-way trước), compare sig of interactive, metric and nonmetric
"I believe Y2 also extraneous variation
the model
affect to Y1" from the dependent
ANCOVA TH2 : + Compare new sig of geographic*income, geographic, income vs old sig in 2-way
(Y2: control variable, because the
More affected ANOVA
variable) effects of the factors
(nonmetric vs
are of major concern.
metric) > Sig càng lớn: Adding controlled variable, Geographic*Income, Geographic, Income less affects
on metric var

> Sig nhỏ hơn: càng thêm controlled variable, Geographic*Income, Geographic, Income the more
affects on metric var

2 approaches to interpret the results:


Agglomeration schedule:
Stage là bước thực hiện
Cluster combined: 2 obvervations có distance gần nhất kết được thành 1 nhóm
Coeffection: tương quan giữa hai 2 observation ở trên --> difference between the coeffection
between rows show the distance of 2 groups of observation
How many parts/
Stage Cluster first appear: biến xuất hiện đầu tiên ở bước số mấy - Next stage: bước tiếp theo mà
segmentations
Hierachical SPSS suggests biến đó sẽ xuất hiện
should we divide
Chọn step mà có khoảng cách lớn nhất --> đếm số bước còn lại đến bước cuối cùng, bao nhiêu
the market into?
bước là bao nhiêu nhóm cluster
CLUSTER Use Dendrogram
Draw vertical line through vertical lines → only care the longest horizontal lines
→ Count how many lines you have in the longest horizontal lines zone: X lines → the best
number to segment is X clusters
QUESTION KEYWORDS PURPOSE HOW TO INTEPRET (GENERAL)
What
characteristics are Based on the Final Cluster centre: We ranked the points of each factor from 1 to n to a new
according to you, Define the
those clusters and table(1 for the highest point, n the lowest point)
K-means how many segment chararacteristic of
which aspect is From the new table, we define the characteristic for each cluster. In the case, the different is
should be divided these segmentation
the key driving unclear -->increase the number of cluster and run the k means again
them
Regression

Look At Coefficients Table:


Can 1 IV leads to Analyze the
Sig of independent < 5% → There is a relationship between IV and DV or IV can lead to DV or
DV? relationship between
IV has effect on DV
(Ex: Can Attitude one (metric)
Single Unstandardized B = … → Independent variable increases 1 lead to dependent var increases …
leads to independent
Subscription variable and metric
Look At Model Summary Table:
intention?) dependent variable
Adjusted R square = … → DV is explained … % by this model (independent var)

Look at Coefficients Table:


Among other
- Sig of IV:
factors, perceived Analyze associative
+ Sig of independent variable < 5% → There is a relationship between IV 1 and DV or IV 1
risk is the most relationships
can lead to DV or IV 1 has effect on DV
important to between a metric
+ Sig of independent variable > 5% → There is no relationship between IV 1 and DV
improve Among other dependent variable
+ Continue with IV 2, IV 3
subscribing factors (... is the and many
Multiple intention! Prove most important to independent variables
- Compare beta (standardized coefficients beta) of each IV → the highest beta → have the highest
that I am right! Dependent and determine which
effect on DV → Can improve DV by focusing on the IV having highest beta
(Among other variable) independent
factors, 1 IV is variable has the
Look At Model Summary Table:
the most most effect on the
Adjusted R square = … → Dependent variable is explained … % by this model (independent
important to dependent variable.
variables)
DV)
How strongly are Look at the Correlation table
sales related to - The Pearson value: how near 2 metric variable are to each other
advertising relationship between
expenditures? relationship 2 metric variables - Check all Pearson Correlation between 2 IV: r <0.9 is good
Correlation Is there an between 2 metric (thuong 2 If r = 1 -> the same variable
association variables (IV) Independent r = 0 -> no relationship between 2 variables
between market variables) r = -1 -> the same variable but ask in opposite way
share and size of Note: If the Pearson Corelation value > 0.90 --> There is a multicollinearity !!! Run factor
the sales force? analysis again
Are consumers’
perceptions of sig < 0.05 there is a correlation bt @ and @.
quality related to sig >0.05 there is no correlation bt @ and @.
their perceptions
of prices?
QUESTION KEYWORDS PURPOSE HOW TO INTEPRET (GENERAL)
Discriminant Two-group If subscribing The first table is useless talking about sample size, second table also too talking about sample size
intention DV If 2 LEVEL in again but divide into 2 level of satisfaction.
only has 2 levels: your model
low and high, and affect the most / Discriminant In Eigenvalues table, Eigenvalues = 0.XXX (eg: 0.346), function 1 explains 100% of variance
attitude is the the most analysis: A technique Canonical correlation = 0. XYZ (eg: 0.507 → 0.507 *0.507= 0.257 → 25.7% of dependent
most important important for analyzing variable explained by independent variable
mong the others, marketing research
is that right? data when the In the Wilk’s lambda table: IF
dependent variable + sig = 0.000 < 0.05, so it less than 5% → SPSS CAN CREATE DISCRIMINANT, the
is nonmetric and discriminant function is significant
independent variables + sig = XYX > 5% --> SPSS CAN NOT CREATE DISCRIMINANT
are metric
IVs: metric → DV: In Standardized Canonical Discrimimant Function Coefficients table: This table shows the beta
nonmetric: of the independent variable.
Discriminant Comparing the beta, we can see that IV with the highest beta affect DV the most. IV with the
second highest beta affect DV the second most
==> Thus, it is correct/ incorrect with your question that IV is the most important mong the
others
Look at classification result table XY% of crossvalidated grouped cases correctly classified >
HIGHER THAN 50% --> WHICH IS GOOD
Discriminant Three-group If subscribing 3 LEVEL Discriminant
intention (DV) analysis: A technique Eigenvalues Table:
measured by 3 for analyzing We have 2 functions because we have 3 levels (Bad behavior, Neutral Behavior and Good
levels: low, marketing research behavior)
medium and high, data when the Look at the % of variance column:
a three group dependent variable Function 1 can explain xx% of variance.
discriminant is nonmetric and Function 2 can explain xx% of variance.
independent
variables are metric Wilks’ Lambda table
Wilk’s Lambda = xx
Sig. of Function 1 through 2=0.000, which is lower than 5%. While sig. of Function 2 =0.638,
which is higher than 5%
→ We should use both 2 functions, and don’t use the Function 2 alone. (CHỌN CÁI <5%)
Nếu cả 2 cùng < 5% → we can explain model with 2 functions or with function 2 only. However,
with function 1 only, we don’t know
To test
relationshi
Standardized Canonical Discriminant table
p of your
Betas of var x and y are ___ and ___, which are the two highest betas in Function 1 → Function 1
DV, is a
has good relationships with var x and var y.
regression
Betas of var x and y are ___ and ___, which are the two highest betas in Function 2 → Function 2
but before
has good relationships with var x and var y.
doing the
regression
Which function should we choose? Look at the plot (Canonical Discriminant Functions)
you use the
table:

Look at Group Centroid


We want to improve DV → Group 3 is a group of high Satisfy → want number of them to
increase
Group 3 has a good relationship with function 1
→ Want to improve DV → function 1, improve group 3 → Improve 2 highest betas in function 1
→ Want to improve function 2, improve group 2

Classification Results table


XY% of crossvalidated grouped cases correctly classified > HIGHER THAN 33.3% --> WHICH
IS GOOD
QUESTION KEYWORDS PURPOSE HOW TO INTEPRET (GENERAL)

Do not care about the block 0


In block 1: The omnibus Tests of Model Coefficients table in Model row: (sig) < 0.05
→ Block 1 model is a significant improvement to the block 0 model.

Model summary: Nagelkerke R squared = 0.XYZ(eg: 0.347)→ DV is explained XY.Z% by this


model.
12. If subscribing Logistic regression is
intention only has used to predict the
2 levels: low and relationship between
The Classification table:
high, frequency NOT important at independent
Look at Overall percentage: A% (eg:70.4%) are correctly classified which is good because >
Logistic of past behavior all among other variables and the
50%
is not important factors dependent variable
at all among where the dependent
Variable in the Equation
other factors, is variable is binary.
If sig. of coi những cái nhỏ hơn 5% để nói chỉ có nó affect lên DV thôi vd \sth{EG: sig. of
that right?
attitude (0.00) and sig. of risk (0.01)}\ are lower than 5%, which means that just it/them{attitude
and risk} AFFECT subscribing intention.
+ SIG of cái đề hỏi < 5% --> GOOD SIG --> WRONG STATEMENT
+ SIG of cái đề hỏi > 5% --> RIGHT STATEMENT
Therefore, the statement that D eg:frequency of past behavior is not important is right.

A technique for
positioning, which
means making a
spatial map, visual Care Object Points map: cannot talk about characteristics, talk about competitors:
display of these direct/indirect competitors
brands that we put on + Direct competitors: which one nearer to our brand
the brand competitors + Indirect competitors: which one very far from our brand
to this map, with the
Who are your
purpose of comparing To define which one is more direct in competing with our brand, we look at the distances {At the
direct and indirect
Direct and indirect and knowing about Output: click Distances}
competitors, what
Internal competitors, the positioning of The smallest distance between our brand and other brands [ eg: Keeng Movies (0.682) is
are the strengths
strength current brands on nearest to Netflix] → It is the most direct competitor.
of you and your
Multidimensio these dimensions.
competitors?
nal Scaling *Vector of strength:
A technique for Look at construct (không phải bảng SPSS) → which brand is strongest in each construct
positioning through a Start from (0,0) →
spatial map to [EG: + ATT: WeTV is the best → draw a straight-line through WeTV named ATT
determine direct and + STRE: VieOn is the best → draw a vector through VieOn: named STRE
indirect competitor
and the strength of
them
Between you and The technique can see
your competitors, how far each brand to Object Common Table: which point is nearest to the Perfect point is Netflix.
who is the nearest Nearest, perfect be the perfect, also Distance Table: which brand has shortest/ smallest distance with Perfect Brand.
External
to be the perfect competitor which brand is the
competitor in this nearest to the perfect --> Brand X is the nearest to be the perfect competitor in the market
market? brand compared to
the other
QUESTION KEYWORDS PURPOSE HOW TO INTEPRET (GENERAL)
New product, Conjoint is used to - Orthogonal table:
collect preference, create a number of → A Plan is successfully generated to 8 cards
test new product new products based You only need 8 new products
Brainstorm for
on a certain number
ideas of a new
of attributes and each - Conjoint analysis:
product with 3
level. Then Conjoint Warning: No reversals occurred → good to run
attributes, 2 levels
tests estimates which Model Description: each attributes have 2 levels
Conjoint each. Collect
attribute is the most
preferences of all
and the least - Overall Statistics:
members in your
important that leads Utility estimate of attribute: brings xx% to respondents’ preferences.
own group to test
to respondents’ Utility estimate indicates which option is the best for your new product: highest utility estimate
the new product.
preference
- Important value: highest value → the most important factor that lead to people preference,
lowest value → the least important factor

You might also like