PSY 6450 Unit 7 Goal Setting Schedules of Reinforcement 1

advertisement
PSY 6450 Unit 7
Goal Setting
Schedules of Reinforcement
1
Schedule
Exam (27 points), Monday, 11/12
Exercise (8 points), Wednesday, 11/07
2
SO1: Locke/Latham maintain that difficult
goals lead to higher levels of performance
Goals should be realistic and challenging but not too difficult
From a behavioral perspective. Why? There are 3 problems.
Diagrams and analyses will be provided in lecture
3
SO2A: Specific goals are better than general goals
(Locke). Why from a behavioral perspective?
Goals affect performance only because of the consequences that
follow behaviors that lead to goal attainment.
 When goals are specific
 They specify the response requirements
 The criterion for reinforcement/reward
 Thus, both employees and managers can easily discriminate
successful from unsuccessful performance
 Goals function like task clarification in the sense that the
employee knows exactly what good performance consists of
 They also provide an explicit “evaluative” component which may
be necessary for feedback to function effectively (in fact, I have
recently come to the conclusion that some type of evaluative
component is a necessary condition for feedback to work, in
most situations - exceptions, self-competition or strong
generalized reinforcement for signs of achievement and being
the “best”)
4
(material is from an analysis by Fellner & Sulzer-Azaroff, 1984, jOBM. Time to update the literature review, do your best goals next; evaluation
Component does NOT have to be goals - could be achieved a number of ways, but goals “work” )
SO2B. What are the problems with “do your best”
goals?
 What about “do your best goals?”
They preclude objective assessment because no
performance criteria are stated
Employees may set lower goals than the supervisor
and anticipate rewards that they then don’t receive
 Remember, most employees evaluate themselves better
than their supervisor evaluates them
5
(material is from an analysis by Fellner & Sulzer-Azaroff, 1984, jOBM. Time to update the literature review)
SO3A: Translation of “desire and
intention” to to attain the goal
 Behavioral translation, where the control of behavior is
based on the past reinforcement contingencies:
Goals will control behavior if, in the past, when a goal
has been set, reaching a goal has been rewarded and/or
not reaching a goal has been punished or criticized
 Emphasis is on what happened in the past
 Emphasis is on the consequences of behaviors that have led
(or not led) to goal attainment, not the goal itself as is
emphasized by Locke (which is an antecedent)
(note cognitive way of talking about goals and effects: “rational” theory in the sense that we set goals and then in advance direct our behavior - the
control of the behavior is in the “here and now,” not in the past consequences)
6
SO3B: We do talk to ourselves about goals.
Thus, rule-governed behavior plays a role.
 What rule is a goal likely to evoke? (according to Malott)
If I don’t get to work on this, I will not meet the goal and I will
look bad
 Once that rule is evoked, how does it control
behavior? (according to Malott)
The rule sets up noncompliance with the rule as a learned
aversive condition, and compliance with the rule (getting to
work) immediately decreases that aversiveness
(must recognize that we do talk to ourselves about goals before we engage in behaviors that will lead to goal attainment, and it would not be
realistic or good to ignore that - too simplistic an analysis - 3C on next slide)
7
SO3C: Goal as an MO
How could a goal function as an MO?
That is, what effects would it have
Reinforcer establishing?
Evocative?
8
SO4: Assigned vs participative and selfassigned goals
 At the present time, the research indicates that
assigned and participative goals are equally
effective
There are no consistent data whatsoever indicating that
participative goals influence behavior more effectively
that goals established by management
 The key issue appears to be not how a goal is
set, but whether a goal is set (of course the goal must
be realistic and attainable)
9
(issue came up at ABA last year, with a student -from another program-arguing and maintaining that participative goals were better)
NFE: Feedback and goals
 Not in SOs, but we do know that goals combined with
feedback are more effective than either alone
 Feedback enhances the effectiveness of goals
 Goals enhance the effectiveness of feedback
 Whenever possible, the following combination should
be used:
 Graphic feedback that displays performance over time
 Goals
 Some type of performance consequence
10
SO5: Daniels vs. Dickinson
 Daniels maintains that if you set a goal and if
performance meets but does not exceed that
goal, the contingency is a negative rather than
positive reinforcement contingency
Also maintains that negative reinforcement contingencies
are bad contingencies because they represent aversive
control
 In order for negative reinforcement to work there must be a
pre-existing aversive stimulus that the behavior terminates
or avoids
 Is this a correct analysis?
(answer is not in sos or on your ppt)
11
Dickinson’s position
Analysis will be provided in lecture
12
Main point repeated
Analysis will be provided in lecture
13
Example:
 Union National Bank
Baseline: 1,065 items per machine hour
Feedback: 1,800 items per machine hour
Incentive, top incentive rate was for 2,500 items per
machine hour: 2,700 items per machine hour
Incentive 2, top incentive rate was for 3,500 items per
machine hour: 3,500 items per machine hour
 During the first incentive phase, proof operators
met but did not exceed the goal (except to a level than
insured they met the goal)
 Yet during the second incentive phase when
additional incentives (reinforcement) was
provided, they increased their performance (but
again, only to a level that met the goal)
14
SO6: Most common mistake re goals
Answers will be provided in lecture
What is the most common mistake that business people
make after implementing a goal setting program for
employees?
Why is that a problem?
What are employees going to do?
15
(mgrs loose their common sense when they become managers. social isolation and criticism)
What about successively increasing goals? NFE
 Daniels recommends that you first set goals low so people
can meet them, then gradually increase them
 Wilk & Redmon used successively increasing goals
 Sulzer-Azaroff used successively increasing goals
 Proceed with caution
You may be able to successively increase goals if rewards are not
tangible, but with tangible rewards, particularly with incentives, you
should never increase the goal level without increasing the reward
level
Tiered reward systems work well
 Union National Bank - increased incentive rate
 Pampino et al. (U2) - an additional lottery ticket
 Performance matrix - more points for higher levels of performance
16
Schedules of Reinforcement
 The basic schedules of reinforcement are
emphasized way too much in OBM. They are not
very relevant. I’ll come back to this in a moment
 Muchinsky characterizes “reinforcement theory”
almost entirely in terms of schedules of
reinforcement and their manipulation
 Provides definitions and examples of basic
schedules
 I correct his definitions in SO7 (NFE)
17
SO8: Muchinsky states that hourly
pay is an example of a FI schedule
Is it?
Why or why not?
Will be discussed in class
18
Schedules of Reinforcement
 Back to Dickinson’s point: The basic schedules of
reinforcement are emphasized way too much in OBM.
They are not very relevant.
 SO9: Hantula’s conclusions after reviewing the effects of
schedules of reinforcement on organizational behavior review covered 1971-1994
 Reinforcement schedules (in comparison to hourly pay) are an
effective way to manage work, however
 The parameters of the schedule did not result in consistent
differences in performance. Rather, the presence of a
contingent relationship between performance and rewards was
the critical factor with respect to improving performance
 Bucklin & Dickinson found the same thing in a review of monetary
incentives
19
SO9: What does this mean?
 Performance contingent rewards do increase work
performance,
 But different schedules of reinforcement (e.g., FR
vs VR schedules, FR1 vs FR4, FR1 vs VR2, VR2 vs
VR4) do not affect performance differently in work
settings
(ABA presentation set up incentives for staff in human service setting - very nice study - spent many, many hours deciding what
reinforcement schedule to use - wasted hours).
20
SO10: Why are these results different
than the results of research on basic schedules?
 In the operant laboratory, different schedules of reinforcement
do generate different response rates and patterns of
performance. So, what may account for the differences seen in
the laboratory and in applied settings?
 Before answering, why does anyone care? Why is this analysis
important?
 Our basic principles of behavior have been called into question
(particularly by expectancy theorists in I/O) because humans do
not show the same response patterns as nonhumans
 That is, they claim this proves that our basic principles are
incorrect
 So, we have to be prepared to answer these criticisms and
concerns
21
Two reasons why humans do not usually display the typical
performance patterns displayed by nonhumans in an
operant laboratory setting
1.
Although schedules used in applied settings are
indeed schedules of reinforcement, they are rarely, if
ever the same schedules examined in the laboratory,
even though they are called the same thing (e.g., FR1,
FR3, etc.). Given that they are not the same, we
should not expect the performance patterns to be the
same
o Example, FR3 for riding a college campus
bus. Every third student was given a token
that could be traded for merchandise at local
stores
o What’s wrong with “this picture?”
22
Two reasons why humans do not usually display the typical
performance patterns displayed by nonhumans in an
operant laboratory setting
2.
Adult humans tend to describe contingencies to
themselves and then their behavior is controlled by
their self-stated rules
o FI: Slow responding is reinforced
o FR: Fast responding is reinforced
o Fergus Lowe’s (Welsh behavioral
psychologist) study with infants, 2-3 year olds
and 5-year olds
23
Wilk & Redmon article
 Study was conducted as Dr. Braksick’s doctoral
dissertation while she was at WMU
 Excellent model of how to do research in the real
world; few better examples
 Follow-up of a study conducted at WMU in our
admissions and orientation office
Pam Liberacki, Director of Admissions and Orientation
 Leslie was hired as a consultant to implement
the program at U of M based on the success of
the program here
24
(not going to go over many of the SOs)
SO11: Why was the efficiency
measure used? Provide the formula.
Participants were 16 clerical workers at
UM
DVs
Number of tasks completed
Performance efficiency
Employee satisfaction
25
SO11: Why was the efficiency
measure used? Provide the formula.
 Performance efficiency formula:
Total number of tasks completed by all participants
Total number of hours worked by all participants
 Why is this an important measure - why not just
use the total number of tasks completed?
 The total number of hours worked by the employees differed from
week to week
 If you only looked at the total number of tasks completed, you
wouldn’t know whether workers were completing more tasks
because they were working more hours or whether they were
completing the more tasks in the same amount of time
 If workers completed more tasks but also worked more
hours, then you have not increased performance
26
(asking you to learn the formula to make sure you understand it)
Skipping to SO16: What procedure was used to
verify that the supervisor actually delivered the
feedback?
 After feedback was given during the week day,
the employee placed a check mark on the next
entry on their data sheet
If you use a graphic feedback display, have employees
initial the graphic feedback display
If you post a graph, have employees initial the posted
graph
More modern technology: send the graph or feedback
via email with verification that the email has been
opened by the recipient (not as good - employees could
conceivably open the email and not look at the
feedback, but better than nothing)
(I am pointing this out because it is an excellent procedure - it’s simple, doesn’t require any extra effort on the part of the researcher, yet does confirm
that feedback was provided as it was supposed to be provided - fidelity of implementation of the IV - a lot of our students at WMU have used this or
something similar in their studies )
27
Base
Filing
983
GS GS
Fdbk Graph
1703
Mail 5077 8822
Room
4188
13389
Credit 685
Eval
861
1049
Data 582
Entry
994
1243
28
(Results!)
SO19: Most importantly, what does this study reveal?
Base
Filing
983
Mail 5077
Room
GS
GS
Fdbk Graph
1703
4188
8822 13389
Credit
Eval
685
861
1049
Data
Entry
582
994
1243
The important role that
graphic feedback plays
in improving performance
29
(click highlight: go back to preceding slide - abrupt immediate increase even over previous phase of GS and verbal fdbk)
Questions over Wilk and Redmon?
30
SO20: Sulzer-Azaroff et al. (NFE)
 Purpose of the study
To determine whether targeting behaviors
(rather than accidents/injuries) would lead to a
decrease in accidents/injuries
 First BBS study to focus on behaviors and prove
a link between that focus and reduction in
accidents
 Prior study in a university chemistry lab, but too
few accidents/injuries to document the link to a
reduction in accidents/injuries
31
SO21: Target behaviors/conditions or
accidents and injuries?
 21A Some behavior analysts feel very strongly
that it is inappropriate to target accidents and
injuries rather than behaviors/conditions
Employees will not report accidents and injuries if you
target those and reward low accident/injury rates - of
course, that is a very bad thing
If you target low accident and injury rates, supervisors
are more likely to use aversive control (when an
accident/injury occurs, they will punish/criticize
workers)
32
(paper company - lottery based on low accidents/injury rates)
SO21: Target behaviors/conditions or
accidents and injuries?
 21B But, what is the danger of targeting only
behaviors/conditions
The ultimate goal is to reduce accident/injury rate. If
you don’t at least measure those, you won’t know
whether you have really been successful
 What if you targeted the wrong behaviors/conditions?
33
(paper company - lottery based on low accidents/injury rates)
SO23: Determining where to start in an
organization
How were the departments selected, and
why were departments selected on this
basis?
Records were analyzed to determine which
departments had the highest accident and injury
rates and the initial program was implemented
in those departments
Focusing on these “hot spots” would give the
greatest initial payoff
(we like big pips! Improvements will help convince others in the organization that the program works and is worth the time and effort to
Implement. Champions within the organization who are “on board” and enthusiastic. Roll-out the program to other departments)
34
SO25: What was the cost of one lost
time accident/injury?
$17,000 in compensation costs alone
Annual savings estimate of $55,000
Why is this important?
Safety programs make good sense
economically
Conflict between operations and safety
(students working in a local paper company, behavior based safety assessment - death. Operations killed it - 1/2 of 1% of operating budget
went to worker’s compensation expenses. Always cost out the expenses involved in accidents/injuries)
35
Questions over Sulzer-Azaroff et al.?
36
Parsons et al. article
 This is the best study I have seen about a large
scale OBM intervention in a human service
setting
 The study was conducted in five group homes for
the developmentally disabled
 In the study objectives, I point out some very
useful procedures that could be implemented in
any human service setting although clearly some
of the details of the procedures would have to be
modified
 Implemented a total system intervention package
37
Parsons intro, cont.
There are two studies
I only have one SO over E1 because I
wanted to focus on the intervention, but
part of the beauty of this work is having
the normative data from E1 when
analyzing the results from E2
38
Overview of Experiments 1 & 2
 Experiment 1
Benchmarking study on treatment and services
22 living units in six state residential facilities
18 were certified as intermediate care facilities under
Medicaid (which means services can be reimbursed
through Medicaid)
 Experiment 2
Purpose was to develop and implement a comprehensive
management system to improve treatment services in five
group homes
Group homes were Medicaid certified
 Medicaid had reviewed services and the facilities had been
given a time-limited mandate to improve services or face
de-certification. Improvement was critical - “critical business
39
issue.
SO26: Results of E1, the benchmarking study
 24A On average, what percentage of resident
behavior was off-task?
When developmentally disabled clients are in group
homes, 2/3 of their time is spent in activity that that
appears to have no habilitative value. This suggests
that residential facilities are not fulfilling their active
treatment obligations
 24B On average, what percentage of resident
behavior was active treatment?
40
General point (NFE)
 These data actually confirm earlier benchmarking
studies
Iwata et al. (1976) found staff spent 45% of their time offtask and only 4% engaged in active training
 Behavior analysts have become very skilled at
developing treatment programs, but the problem
is getting the direct care staff to implement those
programs
Need to train human services professionals in PM
Most find themselves doing staff management (and
systems management), yet many do not take PM courses
 Confirmed by several graduates of the BA program
41
(Very few academics focus on OBM in human service settings)
Organizational structure, staff, and residents
Faculty Program Director
PhD, Experimenter
Area Director
BA in Education
1. Group Home Supervisor
Intermediate
Home Supv
Intermediate
Home Supv
2. Group Home Supervisor
Intermediate
Home Supv
Intermediate
Home Supv
3. Group Home Supervisor
Intermediate
Home Supv
Intermediate
Home Supv
4. Group Home Supervisor
Intermediate
Home Supv
Intermediate
Home Supv
5. Group Home Supervisor
Intermediate
Home Supervisor
Intermediate
Home Supv
110 Direct care staff
165 Residents
42
Intervention: Four basic components (NFE)
Structure (scheduling) and reassignment
of staff
Structure and scheduling is a recurrent
intervention in human service settings
Task clarification
Individual accountability
Staff training
Monitoring of performance
Supervisory feedback
One of my purposes with the SOs is to point out the systems aspects of the program - they implemented monitoring and feedback systems
43
for individuals at EACH level of the organization - we often intervene at the direct care staff level, but who provides PM to the group
home supervisors, and to the supervisor of the group home supervisors? We forget to do that, yet are often surprised our interventions don’t last
SO29 (skipping 27-28)
 29A How often did each supervisor or assistant
supervisor observe each staff person?
Once a week
 29B What procedure was used to verify that the
supervisor observed and gave feedback to the
staff member immediately after the observation?
Each staff member initialed the checklist
I am pointing this out because this is basically the same procedure
used by Wilk & Redmon and it permits the assessment of the integrity
of the intervention without observers. Remember this procedure!
44
SO30: Now we have supv. monitoring and giving
feedback to staff with verification
 What procedure was put in place so area
supervisors knew whether the supervisors were
indeed observing and giving feedback to staff?
Each week the observation forms were given to the
area director who supervised all group home supervisors
who reviewed them
(watch wording for SO30, too close to SO29; interestingly, the authors don’t indicate whether the area director summarized and gave
feedback to the group home supervisors re conducting the observations, but… next slide )
45
SO31: Now we have the area director monitoring the
performance of group home supervisors - who gives
feedback to the area director? (NFE)
 The data on resident behavior collected by researchers
(independent of the preceding measures on staff observations) were
summarized and graphed, and sent to the program director weekly.
 The program director sent the graphs along with comments to the
area director, who then sent the appropriate graphs to each group
home supervisor
 Note two separate and independent measurement systems
 Were supervisors observing and giving feedback to the direct care staff
 How was the supervisory system affecting resident behavior - was
decreasing resident off-task behavior and increasing active training
 Also note that the resident behavior data were collected by:
 8 staff members
 Student interns (number wasn’t specified)
 Extremely labor intensive
46
SOs 33& 34: Back to why the normative
data from E1 was so important
SO33: What very nice contribution does the
normative data provide when analyzing the
results of the study?
 Most studies would have reported the
improvement in resident behavior in comparison
to baseline
During baseline off-task behavior averaged 64%,
which decreased to 41% during the PM intervention
That looks like a nice decrease (23% decrease) but
residents were still off-task 41% of the time
47
(cont. on next slide)
SO33, cont.
With the normative data they could also report
Their baseline average was similar to the average
off-task behavior in the 22 other group homes (18
of which were Medicaid certified): 64% and 67%,
respectively (so maybe they weren’t doing that
badly to begin with!)
Not only did off-task resident decrease
considerably, but it is now well below the normative
average, so…
48
SO33, cont.
Not only could the administrators and
researchers show that these group homes had
improved considerably, they could also show
that they were doing considerably better than
other state residential facilities
49
SO 34: Why is it important to collect
normative data from a staff perspective?
 Basically, so you know realistically, what good
performance is given typical staff-to-resident ratios
 The residents were profoundly developmentally disabled,
typically nonverbal, and required assistance in self-care
routines
 The agency can only hire a certain number of direct care
staff due to budgetary constraints - and usually these
type of organizations are understaffed
 It is simply unrealistic to assume that it is possible to
have 0% off-task resident behavior - so back to the
original question - what is good performance?
50
SO37: What is the potential disadvantage of
targeting staff behavior in contrast to resident
behavior?
 As the authors note, and I mentioned briefly earlier, while
group home supervisors observed the behaviors of staff
and gave feedback to them weekly immediately after the
observations, neither staff behavior nor supervisor
observation behavior were graphed and fed back to
supervisors or staff
 Rather, the feedback that was given was feedback on
the % of off-task resident behavior and % of time
residents were involved in active treatment
 To truly determine a functional relationship between staff and
supervisor behavior and resident behavior, you would have to
measure both (however, I admit I am convinced by the data)
51
SO37, cont.
 The authors make the point, however, that there is a
disadvantage of monitoring staff behavior
 Maintain that staff frequently do not like to be observed and often
react negatively - from mild nervousness to out right hostility
 But, they do not react as negatively when resident behavior is
monitored and reported
 Thus, this may have made it more likely that supervisors would
continue to use the system
 It’s an interesting point - but I don’t know how valid it is
 I looked at the reference given, but it was to a book written by
Reid et al. for practitioners, and no data were provided
 It would be an interesting (but difficult) study to conduct
52
Highly recommended book:
Reid, Dennis H., & Parsons, Marsha B
(1995)
Motivating human service staff
Habilitative Management Consultants
PO Box 2295
Morganton, North Carolina 28680
53
Questions on the Parsons et al. article?
54
In-class exercise
Was the continuous reinforcement (CRF)
schedule used in the Latham & Dossett
article a true CRF schedule?
Was the variable ratio 4 (VR4) schedule
used in the Latham & Dossett article a true
VR4 schedule?
55
Download