Analisis Data Lapangan

Data Inferensi
 Choosing the right statistics

1. Once we have type our data into SPSS and produced descriptive statistics, we usually wish to analyze the data to
test certain hypotheses.

2. Statistics courses teach you how to calculate a correlation coefficient or perform a t-test, but they may not spend
too much time helping you to learn how to choose appropriate approach to address particular research questions.

3. Depending on the question you are addressing and the nature of your data, you may use a variety of different
statistical tests or techniques.

4. Statistics texts and journal articles are good references for you to understand the basic techniques and principles
underlying them.

5. A number of statistical tests will examine differences between groups: they will compare samples to infer whether
the samples come from the same population or not (e.g. t tests).

6. Other tests examine the association/relationship between variables, such as a correlation (e.g. Pearson’s
correlation) or tests of independence (as in the chi-square test).
Association/Relationship
between variables
 Correlation

Structural
Partial
Equation
correlation
Modelling

Relationship
between
variables
Canonical Multiple
correlation regression

Discriminant
Factor
function
analysis
analysis
Correlation

1. Used when you want to explore the

strength of the relationship between Partial correlation
two continuous variables.
2. Gives you an indication of both the
1. An extension of Pearson
direction (positive or negative) and
the strength of the relationship. correlation — allows you to control
for the possible effects of another
3. A positive correlation indicates that confounding variable.
as one variable increases, so does the
other. A negative correlation 2. Partial correlation ‘removes’ the
indicates that as one variable effect of the confounding variable
increases, the other decreases. (e.g. socially desirable
responding), allowing you to get a
more accurate picture of the
relationship between your two
variables of interest.
Multiple regression

1. A more sophisticated extension of

correlation. Exploratory Factor analysis
2. Is used when you want to explore the
predictive ability of a set of
independent variables on one 1. Allows you to condense a large set of
continuous dependent measure. variables or scale items down to a
smaller, more manageable number of
3. Different types of multiple regression dimensions or factors.
allow you to compare the predictive
ability of particular independent 2. Summarizing the underlying
variables and to find the best set of patterns of correlation and looking
variables to predict a dependent for ‘clumps’ or groups of closely
variable. related items.
3. Often used when developing scales
and measures, to identify the
underlying structure.
Differences between groups
Analysis of
covariance t-test
Differences between
groups

Multivariate One-way
analysis of analysis of
variance variance
Two-way
analysis of
variance
T-tests One-way analysis of variance
(ANOVA)
1. Used when you have two groups (e.g. 1. Similar to a t-test, but is used when you have two or more
males and females) or two sets of data groups and you wish to compare their mean scores on a
(before and after), and you wish to continuous variable.
compare the mean score on some
2. It is called one-way because you are looking at the impact of
continuous variable. only one independent variable on your dependent variable.
2. Two main types of t-tests. Paired sample 3. A one-way analysis of variance (ANOVA) will let you know
t-tests (also called repeated measures) whether your groups differ, but it won’t tell you where the
are used when you are interested in significant difference is (gp1/gp3, gp2/gp3 etc.). You can
changes in scores for participants tested conduct post-hoc comparisons to find out which groups are
at Time 1, and then again at Time 2 (often significantly different from one another.
after some intervention or event).
4. You could also choose to test differences between specific
3. Independent sample t-tests are used groups, rather than comparing all the groups, by using
planned comparisons.
when you have two different
(independent) groups of people (males 5. Two types of one-way ANOVAs: repeated measures ANOVA
and females), and you are interested in (same people on more than two occasions), and between-
comparing their scores. groups (or independent samples) ANOVA, where you are
comparing the mean scores of two or more different groups of
people.
 Two-way analysis of variance Multivariate Analysis of Variance
(ANOVA) (MANOVA)
1. Used when you want to compare your groups on a number of
1. Test the impact of two independent variables on one different, but related, dependent variables; for example,
dependent variable. comparing the effects of different treatments on a variety of
outcome measures (e.g. anxiety, depression).
2. The advantage of using a two-way ANOVA is that it allows you
to test for an interaction effect—that is, when the effect of one
independent variable is influenced by another; for example,
when you suspect that optimism increases with age, but only
for males.

3. It also tests for ‘main effects’—that is, the overall effect of each
independent variable (e.g. sex, age). Analysis of covariance (ANCOVA)
4. Two different two-way ANOVAs: between-groups ANOVA 1. Used when you want to statistically control for the possible
(when the groups are different) and repeated measures ANOVA effects of an additional confounding variable (covariate).
(when the same people are tested on more than one occasion).
2. This is useful when you suspect that your groups differ on some
variable that may influence the effect that your independent
variables have on your dependent variable.

3. To be sure that it is the independent variable that is doing the

influencing, ANCOVA statistically removes the effect of the
covariate.
 Demographic anaysis

Descriptive Data assumption analysis

No hypothesis

Status of variables/Level
Association between 2
variables (IV and DV)

Statistics
Correlation
Association between
more than 2 IV on 1 DV
Association

Prediction Regression
Inferential
statistics

Hypothesis
ANOVA
Difference between Comparing 1 score
more than 2 groups Independent between 2 groups
Difference
sample
T-test
Paired sample
Difference between 2 (repeated
Comparing 2 set of
groups or 2 sets of score measure) scores from 1 group
Things to Consider before Analyzing
Your Data

1. We use information from our samples to estimate population values

(called parameters).
2. That is, we will use sample means and standard deviations to
estimate the mean and standard deviation of the population the
samples are drawn from.
3. Thus a number of tests are referred to as parametric tests.
4. In order to undertake parametric tests appropriately there are a few
things we need to consider.
The sample properly represents the population.
Normally, we argue that the sample should be randomly
selected from the population. If the sample is
biased then the sample values will not be good
estimators of the population values.

1. Is your sample unbiased?

Parametric tests require at least interval data – where the
consecutive numbers on the measuring scale are at equal
intervals.

We are also assuming the scales are continuous: that is, there
are no gaps or breaks within them.

For example, time, distance and temperature are all interval

scales.

Without interval data we cannot meaningfully calculate

statistics such as the mean and standard deviation using the
raw scores.

2. What measurement scale is your data?

Regardless of ambiguities and disagreements, researchers generally treat Likert-type as an interval level of
measurement. Particularly for aggregated scores obtained from multi-item scales, researchers assume that
scores are “reasonably” interval-level. (Furr, 2011)

In practice we may generally assume that the score scales of psychological and educational
tests are not interval scales. Nevertheless researchers frequently act as if the score scale is an interval
scale. One might say that no harm is done as long as the predictions from this way of interpreting test
results are useful. When difference scores are used as an indication of a
learning result or an improvement and these scores are related to other variables, certainly the
interval property is invoked. (Gruijter & Kamp, 2002)

If it can be demonstrated that the scores provide more

useful information for placement or prediction when they
are treated as interval data, they should be used as such.
(Lord & Novick, 1968)

There are pros and cons in using the Likert scale as an

interval scale, but the controversy can be handled by
increasing the number of points. (Wu & Leung, 2017)
A number of statistical tests require us to make this
inference.

The statistical test assumes it to be the case and if it is not

then the result may underestimate or overestimate the value of
the statistic.

This can be checked by plotting your data on a histogram or

boxplot or by more precisely by conducting a one sample
Kolmogorov–Smirnov test which statistically tests the normality
of the data.

3. Are the scores from each sample drawn from normally distributed
populations?
Research topic: Stress at work among lecturer.
RO1: To determine the levels of stress among lecturers  descriptive statistics
RO2: To differentiate the levels of stress between male and female lecturers. 
independent sample t-test
RO3: To differentiate the level of stress between lecturers between three faculties  ANOVA
RO4: To measure the relationship between levels of stress and work performance among
lecturers  Pearson correlation.
RO5: To measure the prediction of coping style, social environment, and physical
environment onto stress among lecturers.  multiple regression
RO6: To determine the best predictor to the stress among lecturers.  multiple regression
RO7: To measure the effect of positive social environment on the stress among lecturers. 
paired sample t-test/repeated measure t-test – experimental