Marketing Research (SPSS)
bingjieyu9
_____06SPSSdatapreparationforhypothesestesting13.pptx6. SPSS Data Preparation for Hypotheses Testing
Dr. Boonghee Yoo
RMI Distinguished Professor in Business and
Professor of Marketing & International Business
2
Before data downloading the data, check if the codes are correct in Qualtrics
An example of wrong codes
Recode the incorrect codes manually.
Click the setting button
left to the question.
Before downloading the data… Write for the Question Labels in Qualtrics.
Give a very short variable label.
Download data into SPSS and open it in SPSS
Add a footer
6
VDI to use SPSS
PrideDesktop, a virtual desktop (VDI) application, can be run on almost any device running macOS, Windows, ChromeOS, iOS (iPhone, iPad), and Android OS. More information can be found at www.hofstra.edu/PrideDesktop
Familiarize yourself by playing with SPSS.
Press “Help” buttons in SPSS, which are designed to work as a manual.
Check out SPSS books from the library and eBrary.
Watch YouTube video tutorials.
See the SPSS tutor at Calkin’s Lab.
Ask the instructor, me.
SPSS Questions?
The SPSS helper at Hofstra
If you need a SPSS help, see Rose Tirotta at the Calkins Lab. You first need to email her at [email protected] to set up an appointment.
Name the variables (One word; Give no space)
Label the variables in a few words.
If that’s too long, change it in Qualtrics.
Values
Label the values of ordinal- or nominal-scaled variables.
Do not label the values of interval- or ratio-scaled variables.
Measures
Nominal, ordinal, or scale (= interval and ratio together)
SPSS Variable View: Define Variables
The Downloaded SPSS Data (Variable View)
Match the names and labels by retyping “Question numbers”
Define “Values” and check if the “Measure” is correct.
Delete invalid responses in rows in the SPSS data.
Were all questions answered?
Eliminate the rows (respondents) with too many non-responses
Was a reasonable time spent to complete the survey?
Eliminate the surveys completed in hurry. Time Spent = End Time – Start Time
Were the answers consistent among themselves (consistency = answering the similar-content questions in a similar way)?
Eliminate the surveys with contradictory answers to similar questions.
Did the answers show a reasonable amount of variation for the questions which are different in content to one another?
Eliminate the surveys showing too small overall standard deviation of many and related questions: For example, compute Stan_Dev = SD(v10 to v35).
After eliminating responses, is the sample size satisfactory?
If not, survey more.
Also, delete non-variables in columns.
Create x1 and x2 based on the scenarios presented if xs are manipulated.
Scenario 1 = x1 (hi), x2 (hi) x1 = 2, x2 = 2
Scenario 2 = x1 (hi), x2 (low) x1 = 2, x2 = 1
Scenario 3 = x1 (low), x2 (hi) x1 = 1, x2 = 2
Scenario 4 = x1 (low), x2 (low) x1 = 1, x2 = 1
Add x1 and x2 in the “variable view” in SPSS.
Sort by the scenarios to make it easy to type codes (1 = low and 2 = high) into x1 and x2.
x1 and x2 now have codes (1 or 2).
Scenario 1 = x1 (hi), x2 (hi) x1 = 2, x2 = 2
Scenario 2 = x1 (hi), x2 (low) x1 = 2, x2 = 1
Scenario 3 = x1 (low), x2 (hi) x1 = 1, x2 = 2
Scenario 4 = x1 (low), x2 (low) x1 = 1, x2 = 1
Scenario I 2 3 4
Hypotheses Testing
15
p-value is the probability that the test statistic takes place if Ho is correct.
Ho typically asserts no relationship or no difference.
alpha (a) is Type I error to reject Ho if Ho is correct.
Reject Ho if p-value ≤ a
(i.e., Ha is supported)
Fail to reject Ho if p-value > a (i.e., Ha is not supported)
Measures (items of each measure retained by factor analysis)
Reliability of the measures
If fail, defend why and suggest how to redo the study.
Use the correct technique.
Know what to discuss about the procedure and the result in text.
Create the right tables summarizing the result.
Variables and their measures
p-value < alpha
Conclusion:
Hypotheses are supported or not
Statistical technique
