Method

Background

All fieldwork was conducted between February 1 and March 29, 2017. This is the fifth wave of the Media Use in the Middle East survey. The 2017 survey replicated many of the questions included in the 2015 and 2013 surveys; longitudinal comparisons are provided where applicable. The 2014 and 2016 surveys focused on entertainment media in the Middle East and are only referenced in a small number of questions. Data in all countries were collected via face-to-face interviews, except in Qatar, where telephone interviews were employed via random-digit dialing. Multistage random probability sampling was used in Jordan, Lebanon, Saudi Arabia, Tunisia, and the UAE.

Method

A total of 6,169 respondents participated in the survey, with approximately 1,000 respondents per country.  Fieldwork took place between February 1 and March 29, 2017. Respondents were offered to take the interview in Arabic or English in all countries, and also French in Lebanon and Tunisia. By respondent preference, interviews were completed in Arabic only in Lebanon and Jordan.

Below is a summary of completed interviews and response rates by country for 2013, 2015, and 2017.

Country[1]	Completed interviews (N)			Response rate (%)			Completed interviews  by nationality (N)
	2013	2015	2017	2013	2015	2017		2013	2015	2017
Jordan	1,250	-	1,009	75%	-	58%	Nationals	1,044	-	881
							Non-nationals	206	-	128
Lebanon	1,256	1,000	1,018	70%	54%	49%	Nationals	1,232	995	1,018
							Non-nationals	24	5	0
Qatar	1,253	1,000	1,140	54%	53%	37%	Nationals	352	280	508
							Non-nationals	901	720	632
Saudi Arabia	1,252	1,005	1,016	81%	77%	76%	Nationals	848	633	627
							Non-nationals	404	372	389
Tunisia	1,250	1,012	1,000	21%	85%	87%	Nationals	1,150	909	896
							Non-nationals	100	103	104
United Arab Emirates	1,264	1,005	1,013	70%	86%	73%	Nationals	313	250	481
							Non-nationals	951	755	532
Total	7,525	5,022	6,169

Survey Design

The average interview length was 30 minutes, with variations based on language and number of questions asked of each respondent. The total number of questions asked varied based on responses to previous questions. For example, some questions were asked only of internet users or those who use specific social media platforms.

Approval of the survey and method were required by the governing agency in Jordan, but not the other countries.  As a result, a sub-set of questions were omitted in Jordan or alternative wording was required. These exceptions are noted in the report where applicable.

Survey Procedures

 The following sections describe the detailed survey method and sampling plan in each of the six countries. While survey administration and sampling procedures varied somewhat by country, the method was designed to ensure representation of the national adult population in each country. The samples in all countries except Lebanon include citizens and expatriates. The sample in Lebanon includes just citizens due to the small number of expatriates in this country.

Groups that are not represented in the research: visitors with no residence permit, farmers, servants, the mentally disabled, and those in army barracks, hospitals, dormitories, prisons or labor camps. In Lebanon, residents in areas with heavy Hezbollah presence were also excluded.

Detailed Method: Jordan, Lebanon, Saudi Arabia, Tunisia and the UAE

Surveys in Jordan, Lebanon, Saudi Arabia, Tunisia and the United Arab Emirates were administered via face-to-face interviews, using multi-stage random probability sampling. This approach was designed to produce samples representative of the population of residents in legal households in the surveyed areas.

An overall master sample design was applied in these countries. Each country’s governorates or provinces were divided into cities, towns and villages, which were further divided into administrative units or sectors, and then into clusters, each comprised of several blocks. A block was defined as the PSU (Primary Sampling Unit). In each block a starting point was randomly selected and interviewers followed a pre-defined random path through the block.

A constant fraction sampling procedure was implemented. The constant fraction sampling procedure can be defined as “a constant small number of interviews per Primary Sampling Unit (consisting of the number of interviews per man-day from each block) which was assigned at the rate of eight (8) full interviews per day.”

The total number of Primary Sampling Units per country was 128 (Total Sample: 1000/ 8 interviews per PSU).

• In each block, a starting point was randomly selected and a random path defined.
• The interviews were conducted at regular intervals (of 4 dwellings) following the randomization as defined below:
  • The interviewer was directed to the starting point initially defined by the researcher.
  • The interviewer skipped a number of homes equal to the sample interval (4 houses) and conducted one interview in the next selected household.
  • The interviewer asked for a list of the household members who are 18 years and older, starting with the eldest to the youngest member, and used a random numbers table (or Kish grid) to select the prospective respondent for the interview.
  • The counting of households was continuous, uninterrupted starting from the top floor, clockwise in descending order, from one building to the next following the random path indicated in advance and carefully documented his passage through the block in a detailed Fieldwork Register.

Detailed Method: Qatar

Surveys in Qatar were conducted via telephone interviews. Potential respondents were sampled from mobile lines sourced by local official directories.

Prior to extracting a sample for the survey, the records were reshuffled to keep all numbers in random sequence within each stratum, and a special extraction program was then used to select phone numbers at regular intervals within the structured list.

A multi-stage random probability selection of telephone numbers from the tele-database was thus performed that yielded a representative sample for the interview:

• If available and accessible immediately, the interview would be conducted right away.
• If unavailable at the first call, five further attempts would be scheduled by the system to call again at different times when respondent was likely to be available. If this failed, the number would be skipped and another number would be selected at random from the tele-database.

The interviews proposed to be conducted in the language preferred by the respondent: Arabic for Arabs (as preferred by respondents), and English for non-Arabs.

Data Weighting

Weighting was applied—specifically, rim weighting—in all countries surveyed to bring the data in line with the population in each country. The weighting factors include gender by age, age by nationality, and gender by nationality.

Rim weighting uses a mathematical algorithm to help provide an even distribution of results across the entire dataset while balancing certain characteristics to pre-determined totals. It weights the specified characteristics simultaneously and disturbs each variable as little as possible. The weighting factors used in the countries included:

• KSA: Gender by age, age by nationality, geography
• Qatar: Gender by age, age by nationality, geography
• UAE: Gender by age, geography
• Jordan: Gender by age, geography
• Lebanon: Age, geography
• Tunisia: Age by nationality, geography

The sampling plans in these countries, developed based on census data, were designed to be representative of these populations by controlling for age, gender and region/nationality.

While the data within countries were weighted to be representative, the overall sample of 6,169 was not weighted across countries. That is, we do not claim that aggregated data is “representative,” per se, of the Mideast region as a whole. Weighting across countries was not applied due to the variable population sizes across the participating countries. 

Margin of Sampling Error

The descriptions below show the margin of sampling error based on all interviews conducted in that country, supporting a 95% confidence level. For results based on the full sample in a given country, one can say with 95% confidence that the error attributable to sampling and other random effects is plus or minus the margin of error.

TOTAL SAMPLE

Sample size:                                            6,169

Gender split:                                           52% male, 48% female

Mean age:                                               35 years old

JORDAN

Sample design:                                      Multi-stage random probability sampling

Mode:                                                     Face-to-face, adults 18 plus

Languages:                                            Arabic

Fieldwork dates:                                    March 14 to March 26, 2017

Sample size:                                           1,009

Gender split:                                           50% male, 50% female

Mean age:                                                35 years old

Margin of sampling error:                       +/- 3.1 percentage points

Representative:                                     Adult populations, less visitors with no residence permit,                                                   farmers, servants, the mentally disabled, and those in                                                                    labor camps.

LEBANON

Sample design:                                     Multi-stage random probability sampling

Mode:                                                     Face-to-face, adults 18 plus

Languages:                                             Arabic

Fieldwork dates:                                     February 22 to March 25, 2017

Sample size:                                            1,018

Gender split:                                            50% male, 50% female

Mean age:                                                37 years old

Margin of sampling error:                      +/- 3.1 percentage points

Representative:                                     Adult populations, less visitors with no residence                                                                          permit, farmers, servants, the mentally disabled, and                                                                    those in labor camps, and potential respondents in                                                                        areas with heavy Hezbollah presence.

QATAR

Sample design:                                     Randomized sample within the household using a                                                                          constant-fraction sampling procedure

Mode:                                                     Telephone, adults 18 plus

Languages:                                             Arabic, English

Fieldwork dates:                                     February 7 to March 23, 2017

Sample size:                                            1,140

Gender split:                                           56% male, 46% female

Mean age:                                               34 years old

Margin of sampling error:                      +/- 3.3 percentage points

Representative:                                     Adult population, less those in army barracks, hospitals,                                                                dormitories, and prisons

SAUDI ARABIA

Sample design:                                       Multi-stage random probability sampling

Mode:                                                      Face-to-face, adults 18 plus

Languages:                                             Arabic, English

Fieldwork dates:                                     February 6 to March 23, 2017

Sample size:                                             1,016

Gender split:                                           53% male, 47% female

Mean age:                                               33.5 years old

Margin of sampling error:                      +/- 3.2 percentage points

Representative:                                     Adult populations, less visitors with no residence                                                                          permit, farmers, servants, the mentally disabled, and                                                                    those in labor camps

TUNISIA

Sample design:                                       Multi-stage random probability sampling 

Mode:                                                      Face-to-face, adults 18 plus

Languages:                                             Arabic, French, English

Fieldwork dates:                                     March 8 to March 27, 2017

Sample size:                                             1,000

Gender split:                                            51% male, 49% female

Mean age:                                                 37 years old

Margin of sampling error:                       +/- 3.2 percentage points

Representative:                                      Adult populations, less visitors with no residence                                                                           permit, farmers, servants, the mentally disabled, and                                                                     those in labor camps.

UNITED ARAB EMIRATES

Sample design:                                       Multi-stage random probability sampling

Mode:                                                      Face-to-face, adults 18 plus

Languages:                                             Arabic, English

Fieldwork dates:                                     February 1 to March 29, 2017

Sample size:                                            1,013

Gender split:                                           53% male, 47% female

Mean age:                                               33.5 years old

Margin of sampling error:                      +/- 3.6 percentage points

Representative:                                     Adult populations, less visitors with no residence                                                                          permit, farmers,servants, the mentally disabled,and                                                                      those in labor camps

Margin of sampling error = 1.96* sqrt ((50%*(1-50%)/n)). Reported margins of sampling error take into account data weighting.

Detailed Country Sampling Plans

The sample distribution for each country is noted below.

Jordan Sampling Plan

Governorate	Weighted %	Unweighted Count
Amman	39.0	401
Zarqa'a	15.0	148
Balqa'a	7.0	72
Madba'a	2.0	20
Irbid	18.0	179
Mafraq	5.1	49
Jerash	2.9	31
Ajloun	2.0	20
Karak	3.8	38
Tafialh	.9	9
Ma'an	2.1	21
Aqaba	2.1	21
TOTAL	100	1,009

 

Age	Weighted %	Unweighted Count
18-24	30	242
25-34	28	280
35-44	19	219
45 and above	23	268
TOTAL	100	1,009

 

Gender	Weighted %	Unweighted Count
Male	50	503
Female	50	506
TOTAL	100	1,009

Lebanon Sampling Plan

A representative sample was followed based on the sampling frame source which “Living Condition Survey 2007” is done by Central Agency of Statistics (CAS).

Governorates	Weighted %	Unweighted Count
Beirut	14	157
Mount Lebanon	35.6	350
North	20.2	204
Beqaa	13	131
South	10.2	107
Nabatieh	7	69
TOTAL	100	1,018

 

Age	Weighted %	Unweighted Count
18-24	25	242
25-34	24	251
35-44	22	225
45 and above	29	300
TOTAL	100	1,018

 

Gender	Weighted %	Unweighted Count
Male	50	509
Female	50	509
TOTAL	100	1,018

Note: Sample exclusions included areas with heavy Hezbollah presence (Bent Jbeil, located in Nabatieh, and Baalbek, located in Beqaa).

Qatar Sampling Plan

City	Weighted %	Unweighted Count
Doha	52	556
Rayyan	27	278
Al Wakra	12	128
Umm Slal	2.9	54
Al Khor	2.9	51
Al Shamal	2.8	65
Al Daayen	0.4	8
TOTAL	100	1,140

 

Age	Weighted %	Unweighted Count
18-24	18	175
25-34	38.1	424
35-44	26.3	313
45 and above	11.6	147
Don’t know/Refused	6	81
TOTAL	100	1,140

 

Gender	Weighted %	Unweighted Count
Male	56.1	656
Female	43.9	484
TOTAL	100	1,140

 

	Weighted %	Unweighted Count
Qatari	24	508
Arab Expatriates	35.7	250
Asian Expatriates	27.9	231
European/American Expatriates	12.4	151
TOTAL	100	1,140

Saudi Arabia Sampling Plan

A representative sample was followed based on the statistical book 2007, ”Central Department of Statistic & Information.”

Governorate	Weighted %	Unweighted Count
Jeddah	27	284
Taif	9	84
Dammam\Khobar	8	72
Abha\Khamis Masheet	3	24
Riyadh	22.7	221
Madina	5.3	49
Hail	16	200
Mekka city	9	82
TOTAL	100	1,016

 

Age	Weighted %	Unweighted Count
19-18	7.3	62
24-20	16.4	139
29-25	18.7	162
34-30	16.5	158
39-35	13.5	153
44-40	9.9	121
49-45	6	81
54-50	5.1	62
55 and above	6.5	78
TOTAL	100	1,016

 

Gender	Weighted %	Unweighted Count
Male	53.2	527
Female	46.8	489
TOTAL	100	1,016

Tunisia Sampling Plan

A representative sample was followed based on the sampling frame source which is “General Census of the Population and Housing” (April 28, 2004).

Governorate	Weighted %	Unweighted Count
Tunis	22.8	194
Ariana	9.3	132
Ben Arous	10.5	104
Manouba	8.2	79
Bizerte	13.7	139
Sfax	22.6	217
Sousse	12.9	135
TOTAL	100	1,000

 

Age	Weighted %	Unweighted Count
18-24	21.2	257
25-34	26.4	273
35-44	22.8	200
45-54	19.2	139
55 and above	10.4	131
TOTAL	100	1,000

 

Gender	Weighted %	Unweighted Count
Male	51	535
Female	49	465
TOTAL	100	1,000

 

	Weighted %	Unweighted Count
Tunis	95	896
Expatriates	5	104
TOTAL	100	1,000

 United Arab Emirates Sampling Plan

A representative Sample was followed based on the UAE 2005 Census.

GOVERNATE	Weighted %	Unweighted Count
Abu dhabi/Al Ain	39.5	390
Dubai	29.5	340
Sharjah	17.3	157
Ajman	7.3	73
Ras Al Khaimah	2.8	27
Umm Al Quwain	2.9	21
Fujeirah	0.9	5
TOTAL	100	1,013

 

AGE	Weighted %	Unweighted Count
19-18	6.9	85
24-20	10.9	141
29-25	19.6	176
34-30	18	173
39-35	18.3	163
44-40	11.8	117
49-45	8.7	84
54-50	3.7	47
55 and above	2.1	27
TOTAL	100	1,013

 

GENDER	Weighted %	Unweighted Count
Male	53.2	500
Female	46.8	513
TOTAL	100	1,013

 

Survey in Egypt

Diplomatic hurdles delayed data collection in Egypt until well after the primary data collection described above. From June 18 to July 27, 2017, interviewers conducted 1,000 face-to-face interviews among the general population in Egypt under the direction of Harris Poll by Pan-Arab Research Center. Consistent with the methods described above, the same multi-stage sampling techniques were employed.

The average interview lasted 30 minutes, and the total number of questions asked varied based on responses to previous questions. Additionally, Egyptian authorities required  deleting several survey questions and modifying a few others.  

A summary of completed interviews and response rates for 2013, 2015, and 2017 is as follows:

Rim weighting brought the data in line with the population in each country. The mathematical algorithm used in rim weighting helps provide an even distribution of results across the entire dataset while balancing certain characteristics to pre-determined totals. It weights the specified characteristics simultaneously and disturbs each variable as little as possible. The weighting factors include gender and age.

A summary of Egypt data collection is included below.

Sample design:                       Multi-stage random probability sampling

Mode:                                      Face-to-face, adults 18 plus

Languages:                             Arabic

Fieldwork dates:                     June 18 - July 27, 2017

Sample size:                           1,000

Gender split:                           50% male, 50% female

Mean age:                               35 years old

Margin of sampling error:        +/- 3.0 percentage points

Repres

Representative:                      Adult population, less visitors with no residence permit, farmers,
                                                servants, the mentally disabled, and those in labor camps

United States Survey

Periodically, the report references the results of a survey conducted in the United States in April 2017.

This survey was conducted online by Harris Poll from April 6-10, 2017 among 2,250 U.S. adults[1] [2]  (age 18+years-old). Figures for age, sex, race/ethnicity, education, region and household income were weighted where necessary to bring them into line with their actual proportions in the population. Propensity score weighting was also used to adjust for respondents’ propensity to be online.

Respondents for this survey were selected from among those who have agreed to participate in Harris Poll surveys. The data have been weighted to reflect the composition of the adult population. Because the sample is based on those who agreed to participate in our panel, no estimates of theoretical sampling error can be calculated.

Data Collection Vendors

In the Arab countries surveyed, The Harris Poll partnered with the Pan Arab Research (PARC) to administer interviews.

Harris Poll

Over the last five decades, Harris Polls have become media staples. With comprehensive experience and precise technique in public opinion polling, along with a proven track record of uncovering consumers’ motivations and behaviors, The Harris Poll has gained strong brand recognition around the world. The Harris Poll offers a diverse portfolio of proprietary client solutions to transform relevant insights into actionable foresight for a wide range of industries including healthcare, technology, public affairs, energy, telecommunications, financial services, insurance, media, retail, restaurant and consumer packaged goods.

Pan Arab Research

The Pan Arab Research (PARC) is a full service research agency founded in 1976 in Kuwait by Mr. Sami Raffoul. Headquartered in Dubai, PARC operates in 13 markets with 11 offices across the GCC and Levant Region – including the UAE, Saudi Arabia, Lebanon and Egypt – and employs over 450 full-time staff.

Key staffers from PARC are members of ESOMAR, and adhere to the code of ICC/ESOMAR^[2] in all methods and practices adopted in the research work. In 1996, PARC joined EuroPanel association of consumer panels research suppliers, as it offers Consumer Panels research in the Arab World. PARC is also a part of the Worldwide Independent Network of Market Research (WIN). WIN includes some of the most prestigious independent firms of Market Research throughout the world (formerly identified as Gallup International).

PARC has been a part of Harris Poll’s group of preferred vendors for many years and has conducted data collection research on behalf of The Harris Polls’s offices in the US and Europe.