Assignment 5: ANCOVA & Factorial Designs
(Output Interpretation Explained)
The Effects of Goal Orientation on Job Search
and Reemployment
Question 1
The researchers were interested in the effect of different workshops (Group) on job search
behavior – measured before the workshop (T0activities) as a covariate in the statistical
analysis
Why did the researchers use job sear behavior at T0 as a covariate in the statistical analysis?
1) To reduce the error variance within the group => the covariate variable accounts for
some the unexplained variance
2) To correct the effect in the model for the influence of the covariate
If assumptions are not violated => likely that adding a covariate is truly controlling for its
effect on the dependent variable (i.e., eliminates confounds)
The main research questions are:
(1) “Are unemployed job seekers who were trained in setting learning goals for their job
– more likely than others to put continued effort into job seeking, resulting in more
job search behavior?”
(2) “Is there a relationship between the type of training and finding a job – 8 weeks after
the training?”
First – differences between the different workshop were examined – with respect to job
search behavior when controlling for job search behavior at T0
Second –the proportion of people from the different workshops that had a particular
employment status at T3 was examined
,Question 2
Describe the dependent and independent variables for the above research questions:
Independent Variables
Type of training => control, learning goal, and performance goal workshops
- Between-subject factor => different participants randomly assigned to different
categories of the IV; a participant is assigned to and experiences only one of the
three experimental conditions
- Nominal/Categorical variable => the variable has three categories, without any
intrinsic ordering of the categories
Job Search Behavior at T0 (covariate) => examined before training begins
- Between-subject factor => participants are measured on job search behavior
prior to training (only once) and the results are compared between participants of
different groups
- Measured using 8-item index of job search activities => participants had to
indicate how much time they had spent on such activities (response options: 1=no
time at all to 5=very much time)
- Interval variable => the intervals between the values of the numerical variable are
equally spaced
Age (covariate)
- Between-subject factor => each participant’s age was measured only once and the
ages were compared between participant
- Measured on an interval/continuous scale => we know if one participant is
older/younger than another; we can tell by how much one is older/younger than
the other
Trait Learning Goal Orientation (covariate)
- Between-subjects variable
- Measured with a two-dimensional measure (response options: 1 = strongly
disagree to 5 = strongly agree)
- Interval variable => the intervals between the values of the numerical variable are
equally spaced
Trait Performance Goal Orientation (covariate)
- Between-subjects variable
, - Measured with a two-dimensional measure (response options: 1 = strongly
disagree to 5 = strongly agree)
- Interval variable => the intervals between the values of the numerical variable are
equally spaced
Dependent Variables
Job Search Behavior at T2 => examined 2 weeks following training via phone
interview
- Between-subject factor => participants are measured on job search behavior two
weeks after training (only once) and the results are compared between participants
of different groups
- Measured using 8-item index of job search activities => participants had to
indicate how much time they had spent on such activities (response options: 1=no
time at all to 5=very much time)
- Interval variable => the intervals between the values of the numerical variable are
equally spaced
Employment Status (T3job) => measured 8 weeks after training via phone interview
- Between-subject factor => participants are measured on employment status eight
weeks after training (only once) and the results are compared between participants
of different groups
- Participants were asked whether they found a job (1 = yes, 0 = no)
- Measured on a nominal scale => there is no order and there is no distance between
yes and no
ANCOVA Test and Cross-Table
These analyses will be shown and the output interpretation will be explained following the
data screening and outlier detection.
1) ANCOVA => testing the effectiveness of the workshops with respect to job search
behavior when controlling for job search behavior at T0
2) Cross-Table – including type of training and employment status
, Question 3
The minimum statistic for T2activities => should be 1 at the lowest, since response options
range from 1 to 5
Kurtosis (the degree of presence of outliers in the distribution) => should be 3 for normal
distribution
- Low kurtosis < 3 => light tails; lack of outliers; most data points are present in
high proximity with the mean; flat distribution
- High kurtosis > 3 => heavy tails; presence of outliers; peaked distribution
If positive => heavy-tailed; if negative => light tailed
Skewness (measure of (lack of) symmetry) => should be 0 for normal distribution
- Positive value => too many low scores; right-skewed
- Negative value => too many high scores; left-skewed
Remember:
Standardized residuals > 3 are cause for concern
Standardized residuals > 2.5 for more than 1% of the data are cause for concern
Case numbers 46 and 60 have a std. residuals > 2.5 and this is 2.2% of the data => cause for
concern – and Case number 47 has a std. residual > 3
(Output Interpretation Explained)
The Effects of Goal Orientation on Job Search
and Reemployment
Question 1
The researchers were interested in the effect of different workshops (Group) on job search
behavior – measured before the workshop (T0activities) as a covariate in the statistical
analysis
Why did the researchers use job sear behavior at T0 as a covariate in the statistical analysis?
1) To reduce the error variance within the group => the covariate variable accounts for
some the unexplained variance
2) To correct the effect in the model for the influence of the covariate
If assumptions are not violated => likely that adding a covariate is truly controlling for its
effect on the dependent variable (i.e., eliminates confounds)
The main research questions are:
(1) “Are unemployed job seekers who were trained in setting learning goals for their job
– more likely than others to put continued effort into job seeking, resulting in more
job search behavior?”
(2) “Is there a relationship between the type of training and finding a job – 8 weeks after
the training?”
First – differences between the different workshop were examined – with respect to job
search behavior when controlling for job search behavior at T0
Second –the proportion of people from the different workshops that had a particular
employment status at T3 was examined
,Question 2
Describe the dependent and independent variables for the above research questions:
Independent Variables
Type of training => control, learning goal, and performance goal workshops
- Between-subject factor => different participants randomly assigned to different
categories of the IV; a participant is assigned to and experiences only one of the
three experimental conditions
- Nominal/Categorical variable => the variable has three categories, without any
intrinsic ordering of the categories
Job Search Behavior at T0 (covariate) => examined before training begins
- Between-subject factor => participants are measured on job search behavior
prior to training (only once) and the results are compared between participants of
different groups
- Measured using 8-item index of job search activities => participants had to
indicate how much time they had spent on such activities (response options: 1=no
time at all to 5=very much time)
- Interval variable => the intervals between the values of the numerical variable are
equally spaced
Age (covariate)
- Between-subject factor => each participant’s age was measured only once and the
ages were compared between participant
- Measured on an interval/continuous scale => we know if one participant is
older/younger than another; we can tell by how much one is older/younger than
the other
Trait Learning Goal Orientation (covariate)
- Between-subjects variable
- Measured with a two-dimensional measure (response options: 1 = strongly
disagree to 5 = strongly agree)
- Interval variable => the intervals between the values of the numerical variable are
equally spaced
Trait Performance Goal Orientation (covariate)
- Between-subjects variable
, - Measured with a two-dimensional measure (response options: 1 = strongly
disagree to 5 = strongly agree)
- Interval variable => the intervals between the values of the numerical variable are
equally spaced
Dependent Variables
Job Search Behavior at T2 => examined 2 weeks following training via phone
interview
- Between-subject factor => participants are measured on job search behavior two
weeks after training (only once) and the results are compared between participants
of different groups
- Measured using 8-item index of job search activities => participants had to
indicate how much time they had spent on such activities (response options: 1=no
time at all to 5=very much time)
- Interval variable => the intervals between the values of the numerical variable are
equally spaced
Employment Status (T3job) => measured 8 weeks after training via phone interview
- Between-subject factor => participants are measured on employment status eight
weeks after training (only once) and the results are compared between participants
of different groups
- Participants were asked whether they found a job (1 = yes, 0 = no)
- Measured on a nominal scale => there is no order and there is no distance between
yes and no
ANCOVA Test and Cross-Table
These analyses will be shown and the output interpretation will be explained following the
data screening and outlier detection.
1) ANCOVA => testing the effectiveness of the workshops with respect to job search
behavior when controlling for job search behavior at T0
2) Cross-Table – including type of training and employment status
, Question 3
The minimum statistic for T2activities => should be 1 at the lowest, since response options
range from 1 to 5
Kurtosis (the degree of presence of outliers in the distribution) => should be 3 for normal
distribution
- Low kurtosis < 3 => light tails; lack of outliers; most data points are present in
high proximity with the mean; flat distribution
- High kurtosis > 3 => heavy tails; presence of outliers; peaked distribution
If positive => heavy-tailed; if negative => light tailed
Skewness (measure of (lack of) symmetry) => should be 0 for normal distribution
- Positive value => too many low scores; right-skewed
- Negative value => too many high scores; left-skewed
Remember:
Standardized residuals > 3 are cause for concern
Standardized residuals > 2.5 for more than 1% of the data are cause for concern
Case numbers 46 and 60 have a std. residuals > 2.5 and this is 2.2% of the data => cause for
concern – and Case number 47 has a std. residual > 3
