Please enable JavaScript.
Coggle requires JavaScript to display documents.
INTRO TO SPSS & DATA ANALYSIS I (Intro to SPSS (Data Inspection…
INTRO TO SPSS & DATA ANALYSIS I
Intro to SPSS
Data Inspection
Descriptive statistics helps identify potential problems
Data Preparation
Inter-item Reliability
Reliability
: degree to which an instrument (e.g. a multi-item scale) measures the same construct across items
Cronbach's alpha is used to asses reliability
--> for loyalty items: scores over
0.6
have high internal consistency (if > 0.6, can create new composite variable)
--> if < 0.6, leave out an item and check agn, if need, leave out another item etc)
Data Description
Graphs
easier to interpret than tables
can be used to check the distribution of responses
can spot outliers
Boxplots
observations should fall within the whiskers if not they are potentially outliers.
generally, observations indicated by a * are outliers since they are 3xIQR from the mean
Data Analysis I
Hypothesis Testing
One sample t-test
Used to determine if population mean differs from a specific value
How?
Compare sample mean to a test value which does not have a distribution ard it
if p-value < 0.05, reject H0
Two-tailed vs One-tailed t-tests
Paired sample t-test
Used to compare two means that are from the same indiv, object, or related units
e.g. do the means differ? H0: the means are the same
Since p > 0.05, don't reject H0 i.e. means are not significantly different
Independent sample t-test
Used to compare the mean on one variable for 2 different (independent) grps of ppl
2 test results:
--> equal variances assumed
--> equal variances not assumed
Levene's Test: do the grps being compared have the same population variance?
--> Null hypothesis (H0): The population variance is the same across grps
-if p-value for Levene's test is more than equals to 0.05, don't reject null hypothesis and use "equal variances assumed" row to interpret t-test results
Basic Concepts
Why do hypothesis testing?
A
hypothesis test
enables us to evaluate 2 mutually exclusive statements (hypotheses) abt a population to determine which one is best supported by the sample data
Essential in interpreting the data analysis results, bc it helps us determine if our results are statistically significant
Hypothesis:
an assumption made abt a population parameter such as mean
In hypothesis testing, the null hypothesis (H0) and alternative hypothesis (H1 or H2)
Evaluate evidence to determine if can reject null hypothesis
--> evidence strong enough --> favour alternative hypothesis
--> evidence not strong enough --> favour null hypothesis
--> p-value < 0.5: reject H0
--> p-value > 0.5: DON'T reject H0
Correlation Analysis
Used to explore r/s b/w variables
R/s are bivariate i.e. only 2 ariables are considered at a time
Typically hope to see
--> significant r/s b/w X variables and Y variable
--> no very strong r/s b/w the X variables
In correlational analysis:
-->
Null hypothesis
: there is no association b/w 2 variables i.e. the correlation coefficient (r) = 0
Look at p-value an r for the correlations b/w the IVs and DV
--> significant?
--> correlation size? [small: r < 0.3], [medium: r = 0.3 < r > 0.5], [large: r > 0.5]
Then look at the p-value and r for the correlations b/w the IVs
Syntax
Used to run analysis, leads to same correlation analysis being performed