Who takes care of grandma?€¦ · Respondent-Driven Sampling Chain referrals/snowball sampling: Purposive selection of ‘seeds’ ‘Seeds’ then recruit respondents from the target

Background Implementing RDS Preliminary Findings Discussion

Who takes care of grandma?Insights from a survey using RDS on the living and working

conditions of 24-hour Polish care workers

Lena Hipp1,2, Ulrich Kohler2, Sandra Leumann1

1Berlin Social Science Center (WZB) 2University of Potsdam

“Analytical Sociology: Theory and Empirical Applications”Venice, 19. November 2019

1 / 29


Structure

Background

Implementing RDS

Preliminary Findings

Discussion

2 / 29


Motivation• Increasing demand for paid care work• “Estimations” of 100.000–800.000 informal care workers in

private households (ZQP: 2016)• But: de facto no reliable knowledge about those workers

.5

1

1.5

2

1999 2001 2003 2005 2007 2009 2011 2013 2015 2017

Relatives Relatives & Care Service Stationary

Data: Statistisches Bundesamt (2019)

People in need by year and type (in Mio)

3 / 29


Motivation

4 / 29


Project Goals

• Standardized survey of 24h-care workers in Berlin• Demographics• Working conditions• Prevalence of illicit employment• . . .

• A valid (in a statistical sense) description of targetpopulation (“representativeness”)• Implementing “Respondent-Driven Sampling” (RDS)

5 / 29


Respondent-Driven Sampling

• Chain referrals/snowball sampling:• Purposive selection of ‘seeds’• ‘Seeds’ then recruit respondents from the target population• Respondents recruit further respondents• Details

• Preconditions for success• Incentives (primary and secondary)• Reciprocity/trust• Monitoring fieldwork• Collection of data on the network size

• Weighting procedures

6 / 29


Statistical requirements for RDS

Assumptions

• „Small-world“-characteristics in the target population• Accurate reporting of network size• Random peer recruiting• Recruitment via „1st-Order-Markov-Chain“

→ selection probability as a function of individuals’ network size→ estimation of unbiased parameters in target population

7 / 29


Formative assessment

• Identifying target population• “live-ins” (originally: any type of care work)• currently working in Berlin (originally: Berlin & Brandenburg)• Polish origin (originally: any nationality/migration

background)• Designing the questionnaire• Pretesting• Selecting the interview site• Setting up logistics, e.g.,https://www.pflegestudie-berlin.de/

• Sample size calcuations (# of coupons & # seeds)

8 / 29

https://www.pflegestudie-berlin.de/


Designing coupons

9 / 29


Description of the sample

●●●

●●

● ●● ●

●

●●

●●

●

●●

●

●●●

● ● ●● ● ● ●

●

●●

●●

●●

● ●●

●

● ●●●●●

●●●

●

●●

●●●

●

●●

● ●●

●●●●

●

●●

●

●●●

●●

●

●●

●

● ● ● ●●

●

● ●●●

●

●●

●

●●

●●

●●

●●●●●

● ●●

●●

●●

●●●

●

●

●● ●

●●

●

●● ●

●

●●

●●●

●●

●●●

●●

●

●●●●

●●

●●●

●

●●

● ●

●●

●●

●●● ●

●

●

●● ● ●

●●

●●

●

●

●●

●

●●

● ● ●●

●● ● ●●●

●●●

●●

●●●

●●●

●

●●

01

02

03

0407

08

09

05

1006

● ● ● ● ● ● ● ● ● ●

N: 200

3 12 3 162 1 1 11 4 2 1

10 / 29


Date, day of the week, and time of the survey

0

50

100

150

200

Par

ticip

ants

in to

tal

0

2

4

6

8

10

Par

ticip

ants

per

wee

k

08-2018 10-2018 12-2018 02-2019 04-2019 06-2019 08-2019Date of the interview

Monday

Tuesday

Wednesday

Thursday

Friday

Saturday

Sunday0 20 40 60

Number of respondents

1213

14

15

16

1718

7

8

9

10

11

11 / 29


Where do our respondents come from?

GdańskGdynia

Szczecin

PoznańWarszawa

Lublin

Łódź

Kraków

Bydgoszcz

Radom

Katowice

Częstochowa

BiałystokToruń

KielceWroclaw

Warszawa

(21,34](17,21](15,17](13,15](10,13](8,10](5,8](4,5][4,4]

12 / 29


Who are the 24-h live-ins?Gender, age, education, and qualifications

21179

15112651

378459

13196680

MaleFemale

20-3940-5960-7980-85

< A-levelsA-levels (PL: Matura)

Higher education

Nursing qualification1 year minimumTraining course

None0 20 40 60 80 100

% of participants

Naiv RDS I RDS II

13 / 29


How and with whom do they live?Living situation (Berlin/Poland) and family situation

16633

1481414

28634933

116209

Own roomShared room

Outside the house

Homeowner in PLRenting in PL

Living with relatives/friends in PL

Living alone in PL2-person household in PL3-person household in PL

4+-person household in PL

No children in PL1 child in PL

2+ children in PL0 20 40 60 80 100

% of participants

Naiv RDS I RDS II

14 / 29


Who are their “clients”?Age, care needs, and family situation of carees

12427764

2410190736

1953253

13645

60-6970-7980-89

90-101

Confined to bedDementia

WheelchairBlind

None of these

SeriousAverage

MildDon't know

Living aloneLiving with partner

0 20 40 60 80% of respondents

Naiv RDS I RDS II

15 / 29


What work do they do?Housework duties

95

126

133

134

139

169

174

182

191

Doing the dishes (by hand)

Ironing

Putting the dishes into/out of the dishwasher

Dusting

Vacuuming/Wiping the floor

Shopping and/or running errands

Doing the laundry

Taking out the bins

Preparing meals0 20 40 60 80 100

% of participants

Naiv RDS I RDS II

16 / 29


What work do they do?Care-related duties

8

13

17

49

55

59

67

67

68

115

160

161

None of these

Cleaning and maintenance of catheters

Breathing excercises

Changing the patient's position

Excercises to help mobilize muscles/joints

Cutting/filing nails

Memory training

Assistance at night

Oral/dental hygiene

Toilet assistance

Body hygiene

Getting dressed0 20 40 60 80 100

% of participants

Naiv RDS I RDS II

17 / 29


What work do they do?Medical care

13335510131720334479126

Medical port maintenanceEnemas

Changing catheters/bladder irrigationClearing upper airways

InjectionsOstomy care

Treating bedsoresInhalations

Measuring the blood sugarWound care/changing bandages

Putting on/taking off compression stockingsNone of these

Taking the blood pressurePreparing/administering medicine

0 20 40 60 80 100% of participants

Naiv RDS I RDS II

18 / 29


Other duties

11

20

21

64

76

77

85

115

155

None of these

Going to church

Cinema, theater, museum, etc.

Reading aloud

Going to restaurants/cafés

Escorting to acquaintances/relatives

Escorting to doctor's appointments

Flipping through photo albums

Taking for a walk0 20 40 60 80 100

% of participants

Naiv RDS I RDS II

19 / 29


How does their work supplement care by others?Cooperation with care services and relatives

1686431005

442927651116

No outpatient careOnce a weekTwice a week

Three times a weekFour times a weekFive times a weekSix times a week

Every day

Never/No relativesAlmost daily

A few times a weekOnce a week

Every two weeksOnce a month


Naiv RDS I RDS II

20 / 29


What are their working hours?Free time during the work assignment

26489029

10357189

5623713

No breaks last work dayA <2h break last work day

A 2-3h break last work dayA >3h break last work day

No days offOne day off

Two days offMore than two days off

No longer breaks (≥4h)One longer break (≥4h)

Two longer breaks (≥4h)More than two longer breaks(≥4h)

0 20 40 60 80% of participants

Naiv RDS I RDS II

21 / 29


How do they find their jobs?Means of finding work

74

96

26

101

66

25

First job through an agency

First job through a contact

First job through the media

Current job through an agency

Current job through a contact

Current through the media0 20 40 60 80 100

% of participants

Naiv RDS I RDS II

22 / 29


How are they employed?Form of contract and contractual parties

146

29

21

105

14

23

79

24

Written contract

No written contract

Explicit refusal

Contract with the agency

Contract with the client

Contract with the relatives

Instructions from the agency

No work instructions0 20 40 60 80 100

% of participants

Naiv RDS I RDS II

23 / 29


How much do they earn?Income situation

638118

186101

169253

1293632

800-999 EUR1000-1299 EUR

1300+ EUR

Free accommodationPartial reimbursement

No charge

Free diningPartial reimbursement

No charge

Reimbursement for travelPartial reimbursement

No charge0 20 40 60 80 100

% of participants

Naiv RDS I RDS II

24 / 29


Are their indications of illicit employment?Social security

711108

444635

5212419

124302221

German health insuranceNo German health insurance

Don't know

A1 certificateNo A1 certificate

Don't know

German social insurance numberNo German social insurance number

Don't know

No business registrationBusiness registered in Poland

Business registered in GermanyDon't know


Naiv RDS I RDS II

25 / 29


How happy are they?Life satisfaction in general

6060491211

53834767

576658102

Very satisfiedRather satisfied

ModerateRather dissatisfied

Very dissatisfied



Very dissatisfied



Very dissatisfied0 15 30 45 60

Family life

Life satisfaction

Standard of living

Naiv RDS I RDS II

% of participants

26 / 29


How happy are they?Satisfaction with the work situation

37

99

47

5

2

64

89

28

2

1

Very satisfied

Rather satisfied

Moderate

Rather dissatisfied

Very dissatisfied

Very satisfied

Rather satisfied

Moderate

Rather dissatisfied

Very dissatisfied0 15 30 45 60

Payment

Relationship with client

Naiv RDS I RDS II

% of participants

27 / 29


Implications of preliminary findings based on RDSsurvey

• Substantial findings (Important: Polish live-ins in Berlin!)• Live-ins as an alternative to care services?• Attractive for the clients?• Attractive for the workers?• Social inequalities in care work & old age?• Social security?

• Implementation of RDS• Considerable time restrictions of the respondents• Large primary incentives necessary• Unresolved problem of non-monetary incentives• Unresolved problem of verifying whether the respondents

belong to the target population• High time flexibility in organization of fieldwork (staff!)• Financial flexibility essential• Patience<

28 / 29


Thank you for your attention

Hipp, Lena, Ulrich Kohler und Sandra Leumann (2019)How to Implement Respondent-Driven Sampling inPractice: Insights from Surveying 24-Hour Migrant HomeCare Workers. Survey Methods: Insights from the Fields,1–13. DOI: 10.13094/SMIF-2019-00009

29 / 29

https://doi.org/10.13094/SMIF-2019-00009

Appendix

Appendix

1. RDS design Design

2. Derivation of the number of cases Case numbers

3. Simulation of the expected sample size Simulation

4. Further reading Literature

30 / 29

Appendix

Sample design

Seed Survey?

PI

Recruit?

Resp. Survey?

PI

SI

Stop Stop

StopYes

Yes

Yes

No No

No

go back

31 / 29

Appendix

Desired number of cases

Number of cases is chosen in a way that the estimatorπ̂ does not deviate (with probability of 95%) more than dfrom the population parameter.

n = Deff. ·Z 2

1−α · π(1− π)d2 (1)

with Z 21−α ≈ 1.96 and Deff ≈ 2 (Design effect).

at π = 0.5 (worst-case scenario) and precision of d = 0.05,

n = 2 · 1.962 · 0.52

0.052 = 768 (2)

With 768 cases, the estimates are likely less than 5 percentagepoints off.

go back

32 / 29

Appendix

Expected realized sample size

With s seeds, c coupons issued, and a recruitment suc-cess rate of r , the number of cases after W recruitmentwaves is:

nW ,s,c,r =W∑

w=1

(s − 1.6) · (c · r)w−1 (3)

where the number of unsuccessful seeds has been setto 1.6 (average of RDS samples performed so far; WHO2013: 70)

go back → Simulation results on the next slide

33 / 29

Appendix

Simulation

768

768

768

768

2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10

5 Seeds, 2 Coupons 5 Seeds, 3 Coupons 5 Seeds, 4 Coupons 5 Seeds, 5 Coupons




60%

50%

40%

33%

Recruitment success

Num

ber o

f res

pond

ents

in re

alis

ed s

ampl

e

# WaveIn case of acceptance of 1.6 unsuccessful Seeds

Respondents after ... waves

go back34 / 29

Documents

Who takes care of grandma?€¦ · Respondent-Driven Sampling Chain referrals/snowball sampling: Purposive selection of ‘seeds’ ‘Seeds’ then recruit respondents from the target