Primrose Student.

ZQMS-ARC-REC-002

ASSIGNMENT COVER

REGION: Harare

PROGRAMME: INTAKE 27

FULL NAME OF STUDENT: Chinodakufa Zorodzai PIN: P1398041W

MAILING ADDRESS:[email protected]

CONTACT TELEPHONE/CELL: 0773553579 ID. NO-77.046869.W.77

COURSE NAME:Survey Techniques and Experimental Design COURSE CODE: BSAM 306

ASSIGNMENT NO. e.g. 1 or 2: 2 DUE DATE: 05.04.19

ASSIGNMENT TITLE: Question 1 - A colleague suggests to you that only true experiments

can provide useful results in Science .What would you offer as defense for the role of quasi

experimental and non - experimental designs in the context of social science and define

characteristics of each design type.

-Quasi experimental design lack random assignment .but they test casual hypothesis .

-They identify a comparison group that is as similar as possible to the treatment group

in terms of baseline characteristics .

-There are different techniques for creating a valid comparison group such as
Regression Discontinuity Design (RDD), Propensity Score Matching (PSM ) and

Difference In Differences (DID )

*A - PSM -Propensity Score Matching

-A design whereby every single observation characteristics but the likelihood that the

individual participate in the intervention .

-Propensity Score Matching ensures that the average characteristics of the treatment

and comparison group are similar , are deemed sufficiently to get an unbiased impact

estimate

*How to apply it

1-ensure representative sample survey eligible participants and non participants in the

intervention

-baseline data are preferred in this type of technique for calculating propensity score .

-the method is also used with endline data with variables not affected by the

intervention

2- Estimating propensity score

-it is constructed using participation equation

-it is a logit or probit regression with programme participation as the dependent

variable

Participants in the programme = 1

Participants not in the programme = 0

-the characteristics deemed to affect participation should be well considered and

exhausted .

-there is need to exclude characteristics that may have been affected by the

intervention

-so it is best to use baseline data where available to estimate propensity score

3-Selecting a match algorithm

-each member of the treatment group is then !matched to one or more members in the

comparison group

-it is done by matching each participant to the nearest neighbour non participant

-the mean of the closest five neighbours is commonly used

-a single individual in the comparison group may be matched to several different

individual treatment groups

-for validity it is important to compare observed values for participants and non

participants with the same range of traits

-if observation in propensity score is lower than the lowest observed value in the

treatment group is discarded

-the same as if observation in propensity score is higher than the highest observed

value in the treatment group is discarded

-what remains after is the region of common support.

*4-Check for balance

-Characteristics of treatment group and comparison group are compared to test for

balance
- there will be no significant different in average observable characteristics between the

two groups

-the treatment and the comparison group are similar on observable groups

characteristics then the variance in the incidence of eg child diarrhoea between

treatment and comparison group can be compared to difference such as access to

clean water .

*5-Estimating programme effects and interpret results

-this is done on either single or double difference

-firstly calculate the difference between indicator of individual treatment and average

value for the matched comparison in individual , average out of all ithese differences .

Advantages

1- they are always feasible if data is available

2 - they can be done after an invention is finished

3- recalling is used for construction since baseline data is excluded for prey

intervention characteristics

Disadvantages

1 -matching individuals on the basis of observable characteristics linked to predicted

likelihood of participation
2 -there is bias if there are observed characteristics that affect participation and which

range over time .

B - Regression Discontinuity Deign -RDD

-it is an acceptance method for estimating the effects of a programme or policy .

-there is use of a threshold criteria whereby they must be met before people can

participate in the intervention being evaluated .

-threshold rule determines the eligibility for participants in the programme

-it is usually based on continuous variable assessment for all potential eligible for eg

free breast cancer screening

- older women tend to be affected by breast cancer and older ones are the ones

selecteed for free screening

How To Apply
-determine the margin around the threshold

-it is done using an illiterate approach

-for a start a small margin can be set up so resulting treatment and comparison groups

can be tested their balance or similar

-then if the matching is good the margin can be widened a little and recheck the

balance again

-the process is to be repeatedly until samples start to become dissimilar .

-balance is based on observable characteristics -no imbalances among non -observable

characteristics

-one the sample is established than a regression line is fitted

-it is a line drawn through the data points that represents the best fit between the

variables being studied

-it just summarises the relationship between the selected variables

-when the line slopes down a positive or direct relationship is indicated

Advantages

1-It deals with non -observable characteristics more convincing than other quasi

experimental design approaches .

2- it also utilises administrative data to a large extend

3- it reduces the need of data collection when including administrative data

4 -mpact is valid to those close to threshold

Disadvantages

-selection criteria is not always clear

2- it is more difficult when programmes do not contain information on individuals

refused entry to the programme .

3- it needs a large sample for making an analysis

4- it yields a local area treatment effect

5- impact is invalid to those far from threshold

C DID - Difference In Differences

-it is Aldo known as double difference method

-it compares changes in outcome over time between treatment and comparison groups

to estimate an impact

-it gives a stronger impact estimate than a single difference

-when applying it removes the difference in the outcome between treatment and

comparison groups at baseline level .

-also it uses other matching methods such ad RDD and PSM but without matching

-researchers need to test parallel trends assumptions that is the the trends resulting

between treatment group and comparison areas if they were similar before the

intervention .

How To Apply It

-identity the indicators of interest to be measured relevant to the interest that is being

evaluated .

-the difference in indicators values from before and after the intervention for the

treatment group are compared with some difference values for the comparison group .

-eg in order to identity the effects of a free food scheme on a nutritional status of

children , the mean difference for both treatment group and comparison group would

be calculated between the two examined .

-children who participated in the intervention compared to those who did not

-then the intervention and comparison group will have been matched on key
characteristics using PSM approach to ensure similarities as possible .

*Advantages

1- A good approach to calculate quantitative impact estimate

2- it gives a stronger impact than a single difference

3- it uses other matching methods such as RDD and PSM .

4 - no bias estimate of the impact

5- it is simple

*Disadvantages

1- it is used without matching to take care of selection bias to ensure alikeness

2- it takes long since researchers need to test parallel trends assumptions if they were

similar before .

3 approach alone is not usually enough to address selection bias

4- applying it removes the difference in the outcome between treatment and

comparison group at baseline level .

Qn 2 - Outline the strength and weakness of the following techniques of data collection

in research ----a)- survey (7)

b)-questionnaire (6) c)- interview (6). d )-observation (6)

Aswers

1-A survey
-a survey is usually an official examination done by a tasked trained group of people in

order to ascertain the condition , quantity and quality of something eg national crop
assessment

survey .

Techniques of data collection

Survey Strength Weakness

-A survey is usually an official Formal surveys

examination done by trained -should provide reliable data
personnel in order to -an interview of random used as primary and
ascertain the condition , chosen respondents secondary data .
quantity and quality of
something eg crops -highly represented -come after designing infomal
they may be formal or Surveys eg crop and livestock surveys (boring ).
informal assessments are done at
ward level ---->district -----
>province ------>head office --- -enumerators are paid after
-----> outcomes. job done so are costly .

-Seriousness is needed to
carry out formal surveys
-good statistical significance
since are done by trained **Informal ones
enumerators .
-no standard type of
questions

-precise results -are pre- testing hypothesis

there is greater precision in -need to formulate questions

terms of measuring the such that formal surveys are
collected data (quantitative carried out
data ).
-failure to formulate questions
may affect formal surveys as
in terms of quality of data
**Informal surveys

-are carried out such that

 formal ones are carried out

-helps to narrow down the

data to be collected in the
formal surveys

-helps to sharpen questions in

the formal surveys

Questionnaire -contain uniform set of -dishonest answers

-a written instrument /tool questions a person being interviewed
with questions used to may lie for sake of boredom
interview respondends on a -have relative large number of usually are for the whole year
particular research . responds . using the same person

-contains both quantity and -some of questions remain

quality data to be collected unanswered
difference in understand and
-they are cheap to use interpretation
following the designed
questions - respondents may have a
hidden agenda
-are ease to analyse they may think after are given
making analysis from something to benefit their
questionnaires bring the well being
whole idea of what to be
done -lack personalisation
usually may divert from that
individual to other people
being represented
Interview - there is higher response -translation problems may
-are conversation between shown between the two create distorted data
interviewer and interviewee parties
where questions are asked -it is difficult to get or
and answered that is face to -it is possible to recheck and impression / sense on
face though are more types . have follow ups if there are information since some
any other omitted points respondents may be shy to
embarrass their selves.
-allows respondents to share
what they know in their own -presence of the interviewer
 words and thinking tends to modify the behavior
of respondents
Observation -method is simple to carry out -some of the occurrences may
-an act of noting and -useful for designing not be suitable for open
recording something hypothesis -come to know observations.
or an event about the likes and dislikes of
people - not all occurrences lead
-greater accuracy observer themselves to observation
can directly check the study eg love , affection
accuracy from the observed ,feeling , and emotions .
ones
- social phenomenon can not
-a universal method that is be controlled or used for
used in all science , whether laboratory experiments .
physical , social shows greater
universally of practice - faulty perception

- independent participation ie a high technical job so one

one can contribute to give may not or never be sure that
correct answers freely on own what he is observing is the
thinking . same as it appears to his eyes
.

Sources

Qn1 - http://www.nber.org./papers/.w14723.PDF? new window = 1

Qn 2 a- 2d page 71-81

Survey techniques and experimental design module by J.Fransis

and R.M.Jingura

MARKER’S COMMENTS: ______________________________________________________

______________________________________________________________________________

______________________________________________________________________________

______________________________________________________________________________
______________________________________________________________________________

______________________________________________________________________________

_________________________________________________________________________________
_________________________________________________________________________________
_________________________________________________________________________________
_________________________________________________________________________________
_________________________________________________________________

OVERALL MARK: _____________ MARKER’S NAME: ________________________

MARKER’S SIGNATURE:_______________________________ DATE: ___________

Issue Date: 3 October 2013 Revision 0