Chapter 13 Flashcards
Explain the difference of variables vs constructs:
Variables:
- Quantities that can be measured directly.
e.g., Age, income
Constructs:
- A concept that cannot be measured directly but rather indirectly using multiple items/questions.
e.g., happiness, loyalty
What are the four types of variables?
Nominal, ordinal, interval or ratio.
The type of variable dictates the statistical technique that can be used.
Explain Exploratory Factor Analysis (EFA):
- An EFA is a data reduction technique that identifies constructs (hidden/underlying dimensions).
- (EFA) tests validity of constructs in a
questionnaire.
An EFA answers 2 questions:
1. How many factors (constructs) are there?
Look at:
- Cumulative percentage explained by the factors >= 60%
- Eigenvalues > 1
- Significant decline in scree plot
Factor loadings >=0.4 are meaningful
- What are they?
Can we conduct an EFA? Look at:
1) The Kaiser-Meyer-Olkin measure of sampling adequacy value (KMO value)
2) Communalities of individual items
Explain the Kaiser-Meyer-Olkin measure of sampling adequacy value:
*Indicates how strong the correlation between items (questions) are.
*Ranges from 0 to 1.
*>0.5 – strong enough correlation and EFA can be conducted
*<0.5 = too weak to conduct EFA
Explain Communalities of individual
items:
*Extent to which an individual item
associates with the other items.
*Values near 1 = high correlation.
Values < 0.2 should be reconsidered.
How do you calculate the sample size of an EFA?
Sample size: Levels (5-point scale or >) X TOTAL number of questions =
minimum responses
E.g. Your questionnaire has 2 dimensions (price & quality) with 5 statements (10 statements/items in total) each which are measured on a 7-point scale.
Calculate the sample size:
7-point sale x 10 questions = 70 responses
Define Validity:
Validity is the degree to which a construct measures what it was designed to measure. An overall scale should consist of the
correct constructs.
Define reliability:
Reliability is the consistency of a set of measurements of a measuring
instrument. The reliability of a construct determines whether the measurements of the same construct give, or are likely to give, the same values.
Explain what item analysis is:
Item analysis produces Cronbach’s Alpha value (α) which provides a measure of reliability of the tested construct.
Cronbach’s Alpha value interpretation:
α > 0.8 = good reliability
α > 0.6 and <0.8 = acceptable reliability
α < 0.6 = unacceptable reliability
How do you determine the sample size when testing reliability?
Sample size: Levels (5 point scale or >) X number of questions (in construct) =
minimum responses.
E.g. If you’re using a 7-point scale, and there are 5 questions/items testing your
constructs, what is the minimum number of responses you need to test reliability?
7-point scale x 5 questions = 35 responses
NB! Construct with the GREATEST number of statements must be used to
calculate the sample size needed to test the reliability of all the constructs in
the questionnaire.
Explain Descriptive Statistics:
Variable measurement type dictates available statistics:
Nominal & Ordinal = frequencies & percentages of categories
E.g. gender, level of education
Continuous or Interval = measures of distribution & dispersion
E.g. age, satisfaction, sales
What does a frequency distribution indicate?
How data is distributed over various categories.
They use Percentages – Shows relative importance of figures more clearly than original data.
Which measures of central tendency are usually used in marketing research?
Measures of central tendency – shows most probable/appropriate response to question
- Mode – most frequent value (e.g. most frequent gender)
- Median – middle value
- Arithmetic mean (average)– sum of all values divided by number of values (e.g.
average age)
What are the seven steps of hypothesis testing?
- Formulate the null & alternative hypotheses.
- Select the appropriate statistical test. (Chapter 14 content)
- Specify the level of significance required.
- Determine the value of the test statistic.
- Determine the critical value of the test statistic.
- Compare the value of the test statistic with the critical value.
- Conclusion
Explain the measures of dispersion:
- They reflect how the data is spread around the measures of central tendency.
The most common measures are:
1. Range- Difference between the highest and lowest value in the dispersion.
2. Variance and Standard Deviation- Based on deviations around the mean of the observations.
3. Coefficient of variation - Used to compare the dispersion of two or more series of data.
Explain Step 1 in hypothesis testing:
Step 1: Formulate Null & Alternative
Hypotheses
Null hypothesis (H0) – Current situation
E.g. There is no relationship between price and sales
Alternative hypothesis (H1) – Research hypothesis
E.g. There is a negative relationship between price and sales
Explain step 3 in hypothesis testing:
Step 3: Specify significance level
How much risk is the researcher willing to take?
α = 0.05 level of significance – 5% of being wrong (normally used)
α = 0.01 – 1% of being wrong
Type 1 error = null hypothesis is correct and rejected
Type 2 error = null hypothesis is false and not rejected
Explain Step 4 in hypothesis testing:
Step 4: Determine the value of the test statistic
p-value (probability value) created by SPSS indicates statistical significance (whether there is enough evidence to reject the null hypothesis).
Explain Step 5 in hypothesis testing:
Step 5: Determine the critical value of the test statistic
Depends on α value (level of significance) set in previous step (e.g., 0.05 or 0.01).
Explain Step 6 in hypothesis testing:
Step 6: Compare the value of the test statistic with the critical value
E.g. if α was set to 0.05 and the p-value < 0.05 = reject H0 hypothesis (H1 is accepted)
