The challenges of sampling in Africa

Transcription

1 The challenges of sampling in Africa Prepared by: Dr AC Richards Ask Afrika (Pty) Ltd Head Office: Tele Fax: Mobile Phone: Web Portal: credible conversations

2 Points for discussion Sampling principles Steps in developing a sampling plan Sampling methods Factors influencing sampling methods Key considerations

3 Sampling principles

4 Culture Sampling principles Sampling is use to do estimations/ inferences on a specific topic on a smaller population, avoiding having to review the entire population Sampling is used when it is not administratively feasible to review every sampling unit in the target universe

5 Sampling principles Target population Sample Target population is the population of ultimate interest (e.g. all cellular users) The sample is a subset of the target population (e.g cellular users: MTN, Vodacom, Cell C, Virgin and 8.ta) From the characteristics of the sample, one can infer/generalise characteristics to the target population, if the sample is representative of the population Who was studied impacts to whom the results can be applied

6 Sampling principles Target population WEIGHTING Sample All males and females in Gauteng n=100,000 Females = 60,000 (60%) Males = 40,000 (40%) Sample of males and females in Gauteng n=200 Females = 100 (50%) Males = 100 (50%) Need to make accurate inferences/generalise characteristics of the sampled to the overall population, you need to weigh the data to reflect the true proportions of the target market Up-weigh results of females and down-weigh results of males Add weighting factor in data

7 Steps in developing a sampling plan

8 Steps in developing a sampling plan Execute sampling plan Step 6 Step 1 Define population of interests National Provincial Urban/Rural Demographics Usage of a specific product etc. Choose data collection method Determine sample size The sample design should be such that the standard errors of estimates become as small as possible, since the smaller the standard errors the more precise the estimates. Step 5 Step 2 Face-To-Face: CAPI PAPI CAWI Telephonic Central location Online Select sampling method Step 4 Step 3 Probability (random selection, known probability for inclustion) Non-Probability Choose sampling frame Random digit dialing (self generated numbers) Lists Data base Census Directory lists Household survey

9 Steps in developing a sampling plan (possible challenges) Execute sampling plan Untrained fieldworkers Unfamiliarity of terrain Language differences Culture differences Incorrect listing Determine sample size Representivity Inferences Objectives Reporting domains Step 5 Step 6 Step 1 Define population of interests Step 2 Choose data collection method Timing Budget Infrastructure Level of education Select sampling method Representivity Inferences Objectives Step 4 Step 3 Choose sampling frame Sampling frame availability Quality of sampling frames Updated sampling frames Spatial data - Mapping survey sample areas

10 Sampling methods

11 Sampling methods Sampling Methods Probability Non- Probability Probability sampling is a sampling process that utilizes some form of random selection. In probability sampling, each unit is drawn with known probability, or has a nonzero chance of being selected in the sample Nonprobability sampling depends on subjective judgment. In nonprobability sampling, often, the interviewer selects a sample according to convenience. Nonprobability sampling is well suited for exploratory research intended to generate new ideas Most preferred and robust sampling method

12 Sampling methods Sampling Methods Probability Non- Probability Systematic Stratified Convenience Snowball Cluster Simple Random Judgement Quota

13 Sampling methods Sampling Methods Probability Systematic Cluster Simple Random Stratified Arranging the target population according to some ordering scheme and then selecting elements at regular intervals through the ordered list. Systematic sampling involves a random start and then proceeds with the selection of every kth element from then onwards i) Divide population into clusters ii) Randomly sample clusters iii)measure all units within sampled clusters (This is done usually when sampling a population that's distributed across a wide geographic region) Samples which are drawn from populations that contain a finite number of N units. To select n units out of N, such that each number of combinations has an equal chance of being selected Where the population embraces a number of distinct categories, the frame can be organized by these categories into separate "strata." Each stratum is then sampled as an independent sub-population, out of which individual elements can be randomly selected(can be done proportional, disproportional, equal sample sizes)

14 Sampling methods Sampling Methods Non- Probability Convenience Judgement Quota Traditional "man on the street The problem with all of these types of samples is that we have no evidence that they are representative of the populations we're interested in generalizing to Sampling done with a purpose in mind (females years of age) Need to be representative of a specific target (40% males / 60% females) interview until quota is filled It is proportional, however not random Snowball Identify someone who meets the criteria and ask to recommend/refer other who they may know also meet the criteria (hard to find/small market share)

15 Sampling methods Sampling Methods Probability Pro s Con s Know the probability that you have represented the population well Able to estimate confidence intervals for the statistic Make inferences to total target population Whole target population need to be subscribed in sampling frame Costly Time consuming

16 Sampling methods Sampling Methods Pro s Non- Probability Con s Easy to execute Quick Cost effective May or may not represent the population well Cannot make inferences to total target population

17 Factors influencing sampling methods

18 Culture Factors influencing quality of sampling procedures Lack of sampling frames Quality of sampling frames Difficulty to access available data (census) Inconsistency of sampling variables Lack of lower level estimates: aggregated data Lack of understanding how to execute sampling method

19 Culture Factors influencing quality of sampling procedures Lack of understanding demarcation of country Unavailability of enumeration area (EA) maps Kish grid bias Substitution Lack of understanding weighting procedures or lack of relevant information needed for weighting process Time constrains Budget constraints

20 Factors influencing sampling methods South Africa SA by Munic Province SA by SA by SA by Prov and Munic SA by Prov and MD DC or Metro DC or Metro DC or Metro Municipality Includes Metro Local Municipality DMAs Municipality Includes Metro Local Municipality DMAs MD MD Municipality Includes Metro Local Municipality DMAs Main Place DC Sub Place Main Place DC Sub Place Electoral ward Enumeration Area

21 Factors influencing sampling methods Colour and black/white EA maps

22 Factors influencing sampling methods This is in order to use when household size does not vary greatly. But in many countries there is huge variation in household size. The Kish Grid produces a bias towards smaller households and needs to be revisited.

23 Factors influencing sampling methods

24 Factors influencing sampling methods Substituting the household means that the selected household will not result in a completed interview but a neighboring household. When a household is being substituted it means that the same demographic profile and the same Kish Grid allocated to the primary household will be used for substituting the neighboring household. Reasons for substitution You get to the specified address and there is no household on that stand You received an outright refusal either from the person at the address or from the respondent as per Kish Grid You contact the address three times in order to set up an appointment and each time there is nobody at home You make three attempts to contact the select respondent, to no avail Visit House 1 st 2 nd 3 rd 4 th 5 th 6 th 7 th

25 Key considerations

26 Key considerations Define target population clearly Choose relevant data collection method Can get around unavailable/out dated population data Select relevant sample method based on objectives of research Ensure a well defined sample size, especially when weighting is necessary Educate/train field workers Ensure good quality execution of sampling plan

27