Vous êtes sur la page 1sur 6

Article Review Assignment #2

Cheryl Cheong

1. The variables are Perceived Organizational Support (POS), Perceived


Supervisor Support (PSS), organization size, employee tenure, supervisors
perceived organizational status, and voluntary employee turnover.
POS is the global perceptions formed by employees regarding the
organizations valuation of their contribution and concern for their welfare.
PSS is the general perceptions developed by employees regarding the
degree to which supervisors value their contributions and care about their
well-being.
Organization size is the number of employees in each organization
Employee tenure is the length of time an employee has worked for the
organization.
Supervisors perceived organizational status is the degree to which
employees identify the supervisor with the organization.
Voluntary employee turnover is when an employee self-willingly leaves the
organization. This could be a result of a better job offer, staff conflict, or the
lack of opportunities in career development in the current organization.
2. Organizational Support Theory says that employees form a general
perception regarding the extent that the organization values their contributions
and cares about their well-being (Eisenberger et al, 1986). This is done so as to
determine the organizations willingness to reward increased work effort and
meet employees socioemotional needs (Rhoades & Eisenberger, 2002).
This theory is based on the Social Exchange Theory, which posits that
employee and employer relationship are based on reciprocity. If one treats
another well, the reciprocity norm obliges the return of favorable treatment. Put
into the context of employment, effort and loyalty of employees will be
reciprocated if the employer gives tangible benefits and social rewards
accordingly (Rhoades & Eisenberger, 2002).
3. All the studies are nonexperiments.

4. The three criteria for inferring causation include:

1) Independent variable (X) is related to the dependent variable (Y).


2) X precedes Y.
3) Confounds that could account for the relationship between X and Y
must be eliminated.
Eisenberger et al. (2002) attempt to address the first criteria by calculating the
intercorrelations between PSS and POS in Table 1. PSS at time 1 has a
correlation value of 0.47 with POS at time 1. PSS at time 2 has a correlation
value of 0.55 with POS at time 2. This shows that PSS is related to POS,
although the values are only moderately strong. Also, in figure 1, the regression
coefficient between PSS and POS is 0.54.
Eisenberger addresses the second criteria by using a longitudinal design to see if
PSS at time 1 correlates with POS at time 2. In table 1, PSS at time 1 correlates
with POS at time 2 with a value of 0.47. This shows that PSS precedes POS and
the inference for causality can be made.
Eisenberger addresses the third criteria by using multiple regression, a statistical
technique to assess more than 2 variables at once while controlling the effects of
tenure and organizational size. The pathway is shown in figure 1. Tenure was not
significantly related to Time 2 PSS (=0.5) or Time 2 POS (=0.01).
Organizational size was not significantly related to Time 2 PSS (=0.01) or Time
2 POS(=-0.04). This effectively rules out tenure and organizational size as
confounds. However, Eisenbergers treatment is limited as there are still other
potential confounds that have been left out of the study. They are union
negotiations, governmental health and safety regulations, minimum wage law,
and labour laws. (Rhoades & Eisenberger, 2002).

5. Moderator hypothesis: the relationship between PSS and POS depends on the
supervisors perceived informal organizational status.

PSS

POS

Supervisors
perceived
informal
organizational
Mediator hypothesis: the effect of PSS on voluntary employee turnover operates
through POS
PSS

POS

Voluntary
employee
turnover

6. The first hypothesis is that Withdrawal Behavior (WDB) is negatively related to


POS. The second hypothesis is that extra-role behaviors should be positively
related to POS.
7a. There was a statistical control used for organization size and employee
tenure. Standardized regression coefficients, regression lines, standard
hierarchical regression analysis, and hierarchical logistic regression are
examples of the statistical controls used in the three studies.
7b. Employee tenure.
8a. PSS and POS have an intercorrelation of 0.60, given that p<0.001. PSS and
POS hence have a strong, positive relationship.
8b. The Type I Error rate for the conclusion is 0.1%.
8c. Statistical power is the probability of detecting true effects correctly leading
to the rejection of a false null hypothesis. The sample size of study 2 can be
increased by administering the survey to employees in other organizations, not
just retail sales. Another way to achieve a larger sample size would be to conduct
meta-analysis: combining the data across many samples from a wide variety of
organizations.

9a. There is convergent validity for the Supervisors Perceived Organizational


Status. In table 2, the factor loadings of the three types of items (VAL, INF and
AUT) overlapped in magnitude and each item loaded highly on the single factor
(0.81). There is evidence for discriminant validity as well in table 3. The

intercorrelation of Supervisors perceived organizational status with PSS and


POS is 0.47 and 0.69 respectively. Since these estimated correlation values
between the factors are less than 1.0, there is discriminant validity.
9b. Eisenberger et al. measured POS and PSS across all 3 studies using
surveys based on SPOS, followed by statistical analyses. There was no control
group and no random assignment. The nonrandom assignment is a threat to
internal validity as not all participants are given an equal chance of receiving the
survey. Also, this is not an ideal measure as survey methodology raises issues
with internal and construct validity. Surveys are vulnerable to the history threat as
events outside the study could have influenced the measurement. Also, surveys
are standardized and administered via mail, there is no one to clarify possible
misinterpretations of questions by participants, leading to inaccurate results.
Also, due to time constraints, participants might have given superficial answers.
This would result in inaccurate statistical analyses. Hence, this method of
measurement is less-than-ideal.
10. Yes they are subject to common method bias. In study 1, the surveys were
self-reports that were mailed out to participants at two different timings. There
could have been percept-percept inflation the estimates for perceived
supervisor support at time 1 (or two) are higher than what it actually should be. In
study 2, the surveys are biased. The PSS survey was modified by substituting
the word organization with supervisor in all the questions used in the POS
survey. Also the supervisors organizational status was measured using a 12point questionnaire. There is a source of bias as all the constructs are being
measured by the same self-reports (very little difference). Participants might be
influenced by testing, where memory of earlier responses would affect their
responses for the 2 later surveys. Hence, their reponses could be biased. The
self-reports could be influenced by mood and fatigue. In study 3, the surveys
were administered to the same employees again. The results could have been
influenced by aroused suspicion that the survey was studying their opinions of
their supervisors and could affect their jobs. Hence, participants might have
reported higher ratings that would lead to an overestimated result.
There are a few ways to deal with this threat to construct validity. The first would
be to use other ratings or archival measures of POS or PSS. This would avoid
same-source bias. In the example of supervisors perceived status, it would be
helpful to use the organizational ratings as employees might feel that supervisors
identify with the organization (perhaps because the supervisors were nicer to
them) whereas the organization might feel that the supervisors values do not
fully align with the organizations goals. The other recommendation would be to
use different surveys for PSS and POS- both should not be based on SPOS.
11a. Reliability is the consistency of the measurement and is free from random
error. It consists of internal consistency reliability, test-retest reliability, and
interrater reliability.

11b. Cronbachs alpha reliability values for Study 1:


POS: 0.74 (time 1) and 0.75 (time 2)
PSS: 0.81 (time 1) and 0.82 (time 2)
Study 2:
POS: 0.88
PSS: 0.88
Supervisor Status: 0.89
Study 3:
POS: 0.87
PSS: 0.90
These reliability estimates are adequate. All the values are above 0.70 and this is
acceptable. There is relatively high internal consistency for all the items. Also,
factory analyses were used, which allowed for the identification of uni- or multidimensionality. This helps to maintain the assumption of unidimensionality
existing in the sample of test items, which is embodied by the concept of
reliability. If this assumption was violated, the reliability values would have been
an underestimation. Since factor analysis was used, an underestimation has
been prevented. However, while internal consistency is necessary, is not
sufficient and it would be better if the study had included test-retest reliability
measures for studies 2 and 3 too.
11c. The reliability values increase with the length of the scales.

POS
PSS

12. The external validity of the inferences from this study is very low. There was
no random sampling of participants. In the study 1, the sample was chosen by
picking a specific Belgium university, then choosing the alumni that graduated
between 1997 and 1998. The survey was then mailed to all the alumni and the

final sample would simply consist of those that replied. There was a low
response rate of 54%. Study 1 was trying to achieve diversity in the sample but
they should have used stratified random sampling instead. The inference that
PSS is positively related to temporal change in POS cannot be generalized to the
rest of the population.
In study 2, the sample consisted of 313 retail sales employees who worked for a
chain of large discount electronics and appliance stores in northeastern U.S. The
survey was distributed to all the employees and the final sample consisted of
those that responded, of which there was a response rate of 98%. Again, there is
no nonrandom sampling. While there is a high response rate, this could have
been because the surveys were administered during the employees regular
working hours and physically collected by researchers. The inference that PSS
and POS is positively related to the supervisors perceived organizational status
has low external validity.
In study 3, the sample consisted of retail sales employees who worked for the
same organization investigated in study 2. An additional 226 employees who
received the same PSS and POS scales given in study 2 but did not receive the
scale items in study 2 were used to assess the supervisors organizational status.
There is some semblance of a control group but the entire sample was mixed
and used to calculate the correlations. The response rate of study is 98%.
Turnover data was obtained from organizational records 6 months after the
survey administration. It is hard to conclude high external validity study 3s
inference that POS is a mediator of the PSS-employee turnover relationship due
to the lack of random sampling.
References
Rhoades, L., & Eisenberger, R. (2002). Perceived organizational support: A
review of the literature. Journal of Applied Psychology, 87(4), 698-714. doi:
10.1047//0021-9010.87.4.698

Vous aimerez peut-être aussi