Safety
The extent to which a drug is free from harmful effects.

Sales Force Monitoring
The process of measuring the effectiveness of a team of sales representatives against a range of pre-determined targets. This may include performance against sales targets, call rates, coverage of target customers, etc.

Sales Promotion
Any activity which is not face to face, but which is concerned with increasing sales. Term is often taken to exclude above-the-line advertising.

Sampling
The method of selecting a specified portion, called a sample, from a population, from which information concerning the whole can be inferred. (see also Sample and Sampling Methods)

Sample - A part or subset of a population.

·Weighted Sample - Weighting data is simply a means of adjusting the sample to better reflect the population. For example, if the population were known to be made up of 50% men and 50% women, but our sample contained 52% men and 48% women, we could up-weight the women and down-weight the men to make our sample match the population. (see also Weighting)

• Sampling Error - An error (leading to research bias) made in sampling, due to a failure to accurately represent the entire population (universe). The error could be due to poor definition of the universe (e.g. selecting an age range of 18-65, and missing older patients); error in preparing the sample frame; or error in the sampling method.
• Non-Sampling Error - An error (leading to research bias) lying outside of sampling. This could be due to non-response (non-contactable, untraceable, refusal); or interviewer bias; or a deliberate incorrect answer (measurement error); or due to a mistake in handling data.
• Sample Frame - A list which can assist in sampling respondents. The sample frame should be relevant, accurate and up-to-date, consistent over time, and (preferably) easy to use.
• Pre Selected vs Field Sampling
- Pre Selected Sampling - Sample is selected prior to fieldwork.
- Field Sampling - Interviewer makes final selection of sample.

Sampling Methods
Market research sampling techniques fall into one of three main categories:

• Random (or Probability) Sampling
• Non Random (or Non Probability) Sampling
• Pseudo Random Sampling

These types and their sub-types are defined below

(1) Random Sampling (or Probability Sampling)
A random sample is one in which every population member has a known (and often equal) chance of being selected for inclusion in the sample. Random sampling is almost never possible in market research, due to the lack of information about the population of interest.

• Simple Random Sampling (SRS) - Taking a complete list of every member of the population, individuals are selected from the list at random to be a part of the sample. Notwithstanding the difficulty of obtaining a list of population members, this sampling method could result in unacceptable samples. For example, the sample could be geographically spread which would be uneconomical to interview. Alternatively, it could lead to a sample concentrated in a single area, which would not represent the whole population.
• Stratified Random Sampling - With this type of sampling, the population is divided up into sub-populations, known as strata. For example, the US could be divided into strata representing its States. A random sample is then drawn within each stratum (e.g state). Stratified Random Sampling can either be proportionate or disproportionate.
• Proportionate Stratified Random Sampling - With proportionate sampling, the number of people selected in each stratum is proportional to the size of the stratum. For example, 8.1 million people live in New Jersey but 33.1 million people live in California. We may therefore interview four times as many people in California as New Jersey.
• Disproportionate Stratified Random Sampling - With disproportionate sampling, the number of people selected in each stratum is dependent on the level of variation within each stratum. For example, although the populations of New Jersey and Pennsylvania are roughly equal, if people living in New Jersey are likely to have more varied responses we would interview a greater number.
• Cluster Sampling - With cluster sampling, the population is first split into clusters and then some of those clusters are selected at random. Within each selected cluster, all or some of its members are selected for inclusion in the sample. Although similar to stratified sampling, with cluster sampling we are aiming to recreate a microcosm of the population in each cluster. With stratified sampling, we want each stratum to represent a segment of the population.

(2) Non Random (or Non Probability) Sampling
Non-random sampling methods are most commonly used because it is impractical to carry out random sampling. Non-random sampling has no theoretical basis and its subjective nature means that biases can be introduced. In practice, of course, non-random samples (especially quota sampling) can yield results which are accurate and the simplicity of the approach means that the sample can be achieved more quickly and cheaply.

• Quota Sampling - The most common form of non-random sampling is quota sampling, in which interviewers (or recruiters) make the decision who to include in the sample. By setting quotas it is hoped that the resultant sample will be representative of the population in terms of certain key criteria. The quotas can range from very simple to extremely complex interlocking quotas. As an example, a simple quota sample may ask for 60% Primary Care Physicians (PCPs), 40% Specialists and 50% aged under 40, 50% aged 40 and over. An interlocking quota involves a more detailed specification, e.g. 30% specialists aged 40+, 10% specialists aged under 40, 20% PCPs aged 40+, 30% PCPs aged under 40.
• Convenience Sampling - A convenience sample is one that is drawn simply from those it is convenient to interview. Its main use is for piloting a questionnaire prior to main-stage fieldwork, where the results themselves are not important rather the ease of interviewing.
• Purposive (or Judgmental) Sampling - Purposive sampling is often used in business to business research or in recruitment for group discussions. Respondents are selected for the sample based on their appropriateness for the situation. For example, when recruiting for a creative group discussion it may be decided to select respondents who are likely to work well together - based on the recruiter's judgment.
• Snowball Sampling - A technique used mainly for difficult-to-access populations, where statistical validity is less important than finding respondents. Each respondent is asked if they could refer someone else who could be eligible for the survey.

(3) Pseudo Random Sampling
Pseudo-random sampling techniques tend to have some random elements, but also include some aspect that means that the probability of selection is not known for every population member. Most of the pseudo-random sampling techniques are used for consumer research, although they can be adapted for use in a healthcare setting.

• Random Route Sampling - The interviewer is given a starting address and proceeds to walk a 'random' route calling at houses according to a pre-determined system. For example, "For the starting address, walk around the block calling at every 7th house. When you come back to the road on which you started, continue in the opposite direction with the same rules." Random Route sampling can be very time-consuming and costly, and is often not feasible in rural areas. Interviewers must call-back at every house selected until they receive an answer - if this is not done then the method can be biased in the same way as quota samples. After selecting a household for inclusion in the sample, a Kish Grid is often used to select a specific member of the household.
• Kish Grid for Selecting Household Members - The Kish Grid is a framework for ensuring that respondents within the house are selected without interviewer bias. The number of people living in the house and the last digit of the questionnaire serial number are combined using the grid to indicate which person should be interviewed.

Saturation Point
Level at which any further expansion of distribution in a market is unlikely to be achieved and where further sales are restricted to the potential arising from replacement needs or population growth.

Scale in Market Research (see Research Scales)

Scamp
Rough layout of design for printed literature or advertisement.

Scenario
A set of assumptions fed into a market model to simulate a particular set of future market dynamics.

Screener (see Recruitment Screener)

Secondary Research (see Research Studies)

Segmentation
Segmentation is a view that not all customers the same. Markets consist of a number of 'segments', Each segment consists of 'homogeneous' items. Typically segmentation can be based on age, social class, income, sex, geographical location, condition, treatment, symptoms, etc. There are two main segmentation approaches.

• Needs Based Segmentation - Revolves around customer needs. A process of segmenting the market based on understanding needs of the end-user. This is fundamentally a strategic process so should come early on in product development. This is also known as Market Segmentation or Strategic Segmentation.
• Characteristics Based - Revolves around attributes of the customer and/or area. A process of segmenting customers based on their characteristics, attitudes or behavior. This drives development and execution of customer strategy and targeting (which ones and how) so comes later in product marketing. This is also known as Customer Segmentation or Demographic Segmentation, i.e. the breakdown of a market into discrete groups which have the same tangible characteristics identifiable elements, and which may have its own special requirements. Factors such as optimum price, quality, packaging, and distribution are likely to differ between one segment and another. (see also Cluster Analysis)

Semantic Differential Scale (see Rating Scale)

Smart Mapping
Technique for profiling competitive products on market-driven criteria, A step used in development of positioning and competitor strategies, key messages and to plan future studies.

SMO
Site Management Organization
A company contracted out to manage the sites conducting clinical trials (see also CRO)

Standard Deviation (see Statistical Terms)

Standard Error (see Statistical Terms)

Standard Gamble
A technique of generating patient preferences based on decision theory of choices under conditions of uncertainty.

Star
Highly successful, profitable product which has established a competitive track record and is still growing, i.e. high market share in a market with high growth. (see also Product Portfolio)

Statistical Significance
The probability with which the findings of a study cannot be attributed to chance.

Statistical Terms
Statistics used to describe the nature of an observed (usually a sample) distribution and to compare distributions

• Mean - A measure of central tendency, computed by dividing the sum of the values by the number of values. More precisely known as the arithmetic mean.
• Mode - A measure of central tendency, the value that occurs most frequently.
• Median - A measure of central tendency, the value for which one-half of the observations (when ranked) lie above that value and one-half lie below that value. When the number of values is even, the median is the mean of the two middle values.
• Variance - A measure of dispersion, it is a statistical measure of how similar the observed values are. It is calculated as the average squared distance of all measurements from the mean. It is a statistical measure of how similar the observed values are. The standard deviation is its square root.
• Standard Deviation - A measure of dispersion, it is a statistical measure of how similar the observed values are. The square root of the variance.
• Standard Error - Used in significance testing, it is the theoretical standard deviation of all possible samples. Estimated from the sample variance and sample size.
• Skewness - A measure of the symmetry (or asymmetry) of a distribution. A symmetrical distribution will have a skewness of zero. A significant negative skewness indicates that the distribution has a long left tail, a significant positive skewness indicates a long right tail. In a skewed distribution the mean, median and mode will be different from each other and the mean may be misleading due to the outlier values.
• Kurtosis - A measure of dispersion, the extent to which values are dispersed around the average values. A normal distribution has a kurtosis of zero. A distribution with negative kurtosis (known as 'platykurtic') has greater dispersion than the normal distribution - i.e. is flatter than the normal. A distribution with positive kurtosis (known as 'leptokurtic') has less dispersion than the normal distribution - i.e. is less flat than the normal.

Stochastic Analysis
A form of economic analysis in which values of parameters and occurrence of outcomes are determined and expressed in terms of probabilities, usually with 95% confidence intervals (Cl). Most variables in economic analytic models are determined and expressed stochastically e.g. length of hospitalization = 5 days; 95% Cl = 3.5-10 days, and total cost of care = $1000; 95% Cl = $500-$1500 for parameters and outcomes respectively.

Storyboard
Sequence of animations (cartoons) outlining the key scenes of a proposed video or film sequence.

Strategic Segmentation (see Segmentation)

Strategy
Plan for reaching certain objectives, usually quantified and more often on a relatively long time base.

Stratified Random Sampling (see Sampling Methods)

Structured Interview (see Research Interviews)

Structured Questionnaire (see Research Questions)

Study (in Market Research) (see Research Studies)

Survey Research (see Research Studies)

SWOT
Strengths, Weaknesses, Opportunities, Threats
Used in situation assessment used for marketing planning.

Syndicated Research (see Research Studies)

Synectics
The study of processes leading to innovation. The aim to solved practical problems using a Synectic group of people of varied imagination and ability and interests.